WIRELESS TELEGRAPHY -MlBookCoJne PUBLI SHERS OF BOO K.S F O R_, Coal Age ^ Electric Railway Journal Electrical World ^ Engineering News-Record American Machinist ^ The Contractor Engineering S Mining Journal ^ Power Metallurgical & Chemical Engineering Electrical Merchandising WIRELESS TELEGRAPHY BY DR. J. ^ENNECK, PROFESSOR OF PHYSICS AT THE TECHNICAL HIGH SCHOOL OF MUNICH. TRANSLATED FROM THE GERMAN BY A. E. SEELIG, E. E. FIRST EDITION FIFTH IMPRESSION McGRAW-HILL BOOK COMPANY, INC, 239 WEST 39TH STREET. NEW YORK LONDON: HILL PUBLISHING CO., LTD. 6 & 8 BOUVERIE ST., E. C. 1915 COPYRIGHT, 1915, BY THE McGRAw-HiLL BOOK COMPANY, INC. First printing, August, 1915 Second printing. May, 1916 Third printing, July, 1917 Fourth printing, January, 1918 Fifth printing, March, 1918 . ; '' 93 THE MAPLE PRESS YORK F JL Y/ EXTRACT FROM AUTHOR'S PREFACE TO THE FIRST EDITION This book was written at the suggestion of the publisher, Dr. Enke. It was originally intended to be an abridged form of my larger book " Elektrornagnetische Schwingungen und drahtlose Telegraphic" (Stutt- gart, 1905). It has, however, developed into something quite different; evidence of this lies in the fact that only 79 of the 332 illustrations of the larger book have been reproduced here. Since I began writing this book (winter 1905-1906) conditions in wire- less telegraphy have changed greatly. The mere fact that new devices and methods have appeared would be of relatively little importance; but the points of view determining the consideration of many of the problems of the art have changed entirely. This necessitated rewriting many portions of the book, which would otherwise have been out of date from its very publication. I need hardly dwell on what this has meant for both the publisher and myself. The mathematical premises are the same as in my larger book. In the text, knowledge of only elementary mathematics the use of differ- ential and integral calculus would have offered no advantage, in the Notes, knowledge of the electromagnetic theory is assumed. The phys- ical premises are somewhat higher than in the larger book ; knowledge of experimental electro-physics and of the phenomena of alternating cur- rents, in short the ground covered by the first four chapters of my larger book, is necessary for a thorough understanding of this volume. I have been somewhat more sparing with the bibliography, for since a year ago, Dr. G. Eichhorn, in the " Jahrbuch fur drahtlose Telegraphic, " gives detailed references to the literature on the subject. As regards the commercial form of the apparatus, most frequent refer- ence has been made to the German manufacturers (Ges. f. drahtl. Tel. and Amalgamated Radio-telegraph Co., i.e., C. Lorenz, A. G.). In so doing I had no desire to show these firms any preference. Description of all the different makes of apparatus would have been prohibitive, and I have simply chosen as examples the apparatus of those firms which placed exact data and photographs at my disposal. Moreover other makes of apparatus are fully described in other books; I might mention the excellent works of J. A. Fleming and particularly of J. Erskine- Murray in this connection. J. ZENNECK. BRAUNSCHWEIG, PHYSIKALISCHES INSTITUT DER TECHNISCHEN HOCHSCHULE, Dez., 1908. V 373881 AUTHOR'S PREFACE TO THE SECOND EDITION Only two and a half years after the appearance of the first edition, a second one has become necessary, even though a French edition had already appeared in the meantime. The book, therefore, has been ac- corded a much more favorable reception than I had dared hope. This served particularly to spur me on to do everything within my power to make the second edition representative of the present status of wireless telegraphy. Due to its rapid development, this meant an ex- tensive revision of the entire book. Unfortunately I found it impossible to carry out this revision without extending the scope. In view of this wider scope, the book has been renamed " Textbook" ("Lehrbuch") instead of " Elements" (" Leitfaden") "of Wireless Telegraphy." In choosing my subject matter, I was guided chiefly by the standpoint of the physicist. I have frequently discussed arrangements or devices involving a new physical idea, even though knowing that they had either not been used to date or are no longer used in practice. To confine ourselves to what is of practical importance will only be proper when once it has been fixed what really is of " practical importance." On this point, however, the views of experts have changed very rapidly during recent years; even to-day individual views diverge widely and seem to be influenced less by scientific reasons than by patent rights. Unquestionably, theoretical investigation, laboratory experiments and experiences in practice have cleared much in recent years. Nevertheless, there still remain a number of problems which find no answer in the re- sults obtained to date. If then my presentation of these problems falls short of the necessary clearness, the fault does not rest entirely with me. In this edition, as in the first, I have received friendly cooperation from many sources: from Dr. L. W. Austin (Washington, D. C.), H. Boas (Berlin), Dr. L. Cohen (Brant Rock), F. Ducretet and E. Roger (Paris), Dr. Erskine-Murray (London) , the Gesellschaf t f iir drahtlose Telegraphic (Telefunken Co., Berlin), Dr. E. Huth (Berlin), the C. Lorenz Co. (Berlin), the Marconi Wireless Telegraph Co. (London), Dr. E. Nesper (Berlin), Dr. E. H. Riegger and Dr. Rukop (Danzig), Dr. G. Seibt (Berlin), the Societe frangaise de radioelectrique (Paris), and Prof. C. Tissot (Brest). To all these I herewith express my thanks. vii viii AUTHOR'S PREFACE TO THE SECOND EDITION Particular thanks are due Dr. A. Meissner (Berlin), Prof. Vollmer (Jena), and Prof. M. Wien (Jena). These have gone to the great trouble of reading through the entire proof, and by their valuable advice have guarded me against many errors and defects. Lastly, I thank the publisher, Dr. A. Enke (Stuttgart) for the kind interest he has evidenced in the preparation of the book in its final form. J. ZENNECK. DANZIG-LANGFUHR, PHYSIKALISCHES INSTITUT DER TECHNISCHEN HOCHSCHULE, Nov., 1912. TRANSLATOR'S INTRODUCTORY NOTE Few students of wireless telegraphy need to be introduced to "Zen- neck." To many, however, this splendid work has remained a " closed book" due to lack of knowledge of the author's language. Hence this translation in the hope that it will fill a real need. Aside from the comprehensiveness of the work as a text-book, the author, without failing to give full credit to the best that has been done in America, naturally pays most attention to the work done in Europe, especially in Germany, and thus gives us an insight into the excellent results accomplished abroad by inventors and engineers in developing the art based on Marconi's fundamental invention. Rather than take the risk of distorting the author's precise meaning, the translator has at times retained a very literal translation in preference to adopting the customary English phraseology. Moreover, when tempted to add something to or modify the original (as for instance in connection with recent "high frequency" apparatus), the translator finally decided to let Zenneck be Zenneck. A word of thanks is due to Dr. L. W. Austin for occasional friendly assistance, as well as to the publishers for their cooperation in the preparation of the book. A. E. SEELIG. WELLSVILLE, N. Y., August, 1915. CONTENTS CHAPTER 1 THE NATURAL OSCILLATIONS OP CONDENSER CIRCUITS PAGE 1. Oscillations Produced by Charging the Condenser 1 1. The Frequency 2. Experimental Determination of the Frequency 3 3. Calculation of Frequency (Thomson's Equation) . . 6 4. Condensers in Series and in Parallel 7 5. The Practical Importance of Thomson's Equation 9 2. The Damping 6. The Transfer of Energy 9 7. The Various Causes of Damping 11 8. Condenser Circuit without Spark Gap. Damping Due to Heat Loss . 11 9. Condenser Circuit with Spark Gap. Damping Due to Spark .... 13 10. Methods for Determining the Spark Gap Damping 16 11. The Factors Which Determine the Amount of Gap Damping .... 16 12. Spark Gaps in Series (Multiple Gaps) 20 13. Energy Losses in the Dielectric of the Condensers 20 14. Energy Lost by Leakage Discharge 21 15. Energy Lost by Eddy Currents 22 16. Relative Importance of the Various Energy Losses 23 CHAPTER 11 OPEN OSCILLATORS 1. The Lineal Oscillator 17. The Fundamental and Upper Harmonic Oscillations 24 18. Current and Potential Distribution in the Fundamental Oscillation. . 24 19. Frequency of the Fundamental Oscillation 26 20. The Electromagnetic Field of the Fundamental Oscillation 27 21. Damping of the Fundamental Oscillation 31 22. Upper Harmonics of the Lineal Oscillator 32 23. Coils 33 2. General Properties of Open Oscillators 24. Current and Potential Distribution along a Wire 34 25. The Electromagnetic Field at Great Distances from the Oscillator. . 35 26. The Radiation of an Oscillator 39 27. Effective Capacity and Effective Self-inductance of an Oscillator . . 40 3. Various Forms of Complex Oscillators 28. Lineal Oscillator with Two Equal Capacities, One at Each End (Hertz Oscillator) 41 29. Lineal Oscillator with Capacity at One End 42 30. Lineal Oscillator Containing Series Condensers 43 31. Lineal Oscillator Containing Series Inductance 44 32. Lineal Oscillator with Both Inductance and Capacity 45 33. Grounded Oscillators : 46 xi xii CONTENTS CHAPTER 111 THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT PAGE 1. Resistance, Self-induction and Capacity 34. Current Distribution in Cross-section of Solid Wires 47 35. Coefficient of Self-induction 47 36. Resistance of Straight Wires 48 37. Resistance of Coils 50 38. Coils Having Variable Self-induction 51 39. Condensers of Constant Capacity 54 40. Variable Condensers 59 2. Current and Voltage 41. Relations between Current and Voltage Amplitudes 62 42. The Break-down Voltage and Gap Length . 64 43* Insulation of Conductors 66 3. Measurement of Current 44. The Indications of Hot-wire Instruments 67 45. Commercial Hot-wire Instruments 71 46. The Hot-wire Air Thermometer 71 47. Bolometer, Barretter 72 48. Thermoelement or Thermocouple 74 49. The Thermogalvanometer ! 75 50. Comparison of the Sensitiveness of Various Measuring Instruments . . 76 51. Measurement of Very Small Currents 76 CHAPTER IV COUPLED CIRCUITS 1. Coupling in General 52. Magnetic, Galvanic, Electric Coupling 79 53. Loose and Close Coupling . . . . 81 54. Methods of Coupling 82 2. Loose Coupling of Damped Oscillating Circuits 55. Coupling of Oscillator to Closed Circuit 84 56. Extremely Loose Coupling of Two Oscillators 85 57. Loose Coupling of Two Oscillators 87 3. Close Coupling of Tuned, Damped Oscillating Circuits 58. Form of the Oscillations 87 59. The Frequency of Coupling Waves 88 60. The Decrements of Coupling Waves 90 61. Amplitude and Phase of the Oscillations 91 4. Quenching Action in Coupled Circuits 62. Form of the Oscillations 93 63. Various Types of Quenched Gaps 95 64. Requirements for Good Quenching 95 65. Concerning the Nature of the Quenching Action 97 5. The Coupling of Undamped Oscillating Circuits 66. Coupling with a Closed Circuit 99 67. Loose Coupling with an Oscillator 100 68. Close Coupling with an Oscillator 101 69. Difference between Damped and Undamped Oscillations 103 CONTENTS xiii CHAPTER V RESONANCE CURVES PAGE 1. The Resonance Curve of the Current Effect 70. General Remarks 104 71. Measurement of the Frequency 106 72. Calibration of the Measuring Circuit 108 73. Determination of Capacities and Coefficients of Self and Mutual Induction by Resonance 112 74. Determination of the Sum of the Decrements of the Primary and Secondary Circuits (v. Bjerknes) 113 75. Abnormal Forms of the Resonance Curves 116 76. Determination of the Decrements of the Primary and Secondary Circuits 118 77. Measurement of Small Changes in the Decrement 120 78. Measurements with Resonance Circuits in General 120 79. Commercial " Wavemeters" 125 2. Resonance Curve of the Dynamometer Effect (L. Mandelstam and N. Papalexi) 80. General 132 81. Determination of the Frequency (Wave length) 134 82. Decrement Determination 135 83. The Dynamometer 135 3. Use of Resonance in the Study of Condensers 84. Determination of the Frequency Factor 137 85. Energy Absorbed by Dielectric Hysteresis 138 86. The Brush Discharge of Condensers 138 4. The Use of Resonance Curves for Investigating Coupled Circuits 87. Coupling of Tuned Circuits 142 88. Close Coupling of Tuned Circuits 145 89. Coupling Untuned Circuits 147 90. Investigation of the Quenching Action in Spark Gaps 148 CHAPTER VI THE ANTENNA 91. General 150 1. The Various Kinds of Antennae 92. Form of the Aerials 150 93. Comparison of the Different Forms of Aerials 155 2. Grounding 94. Ground and Counterpoise . 157 95. Energy Consumed by the Earth Currents 158 96. Ungrounded Antennae for Airships 163 3. The Oscillations of Antennce 97. Frequency, Capacity and Self-induction 164 98. Regarding the Effect of Coils 165 99. The Damping of Antennae and Its Causes 167 100. Determination of the Decrement 168 xiv CONTENTS CHAPTER Vll TRANSMITTERS OF DAMPED OSCILLATIONS PAGE 101. The Different Types of Transmitters , 173 1. The Simple (Marconi) Transmitter 102. General 173 103. The Damping 174 2. The Braun Transmitter 104. Nature of the Coupling 175 105. Coupled Transmitter for Antenna? Having High Damping. Very Loose Coupling 175 106. Coupled Transmitter for Antennae Having High Damping. Close Coupling 176 107. Coupled Transmitters for Slightly Damped Antennae 178 108. Commercial Form of the Braun Transmitter 179 3. Quenched Spark Gap Transmitter (Wien's Transmitter) 109. Impulse Excitation 182 110. The Connections 184 111. Practical Construction of Quenched Spark Gaps ......... 186 112. Commercial Construction of the Wien Transmitter 192 4. General Consideration of Transmitters of Damped Oscillations 113. Operation by Means of Interrupted Direct Current 194 114. Alternating-current Operation . 195 115. Direct-current Operation ...... i 198 116. Measurement of Energy Supplied ; Determination of the Efficiency . . 200 117. The Key 202 118. Spark Gaps with Rotating Electrodes 203 5. Comparison of the Different Types of Transmitters 119. Difference between the Coupled and the Simple (Marconi) Transmitter 208 120. Comparison of the Braun and Wien Transmitters 210 CHAPTER Vlll HIGH FREQUENCY MACHINES FOR UNDAMPED OSCILLATIONS 121. The Alexanderson-Fessenden Machines 213 122. Goldschmidt's High Frequency Generator .'." 216 CHAPTER IX UNDAMPED OSCILLATIONS BY THE ARC METHOD 1. The Various Arrangements 123. The Problem and the Solution by V. Poulsen 220 124. Commercial Construction of the Poulsen Generators ........ 222 125. Use of the Poulsen Arc for Measuring Purposes . . . 225 126. Circuit Connections of the Poulsen Transmitter 227 127. Devices for Producing Signals 228 128. The Multitone Transmitter of J. C. Lorenz 229 2. Study of the Action of the Arc 129. Characteristic of the Arc 231 130. Ty pel Oscillations:/! < 7 233 CONTENTS XV PAGE 131. Type 11 Oscillations: /I Q > 7 ; no Re-ignition 234 132. Type 111 Oscillations: II Q > 7 ; Re-ignition Present 236 133. Energy of the Oscillations 237 134. Frequency of the Oscillations 215 238 135. Practical Conclusions for Type 11 Oscillations 239 136. Regularity of Type 11 Oscillations 241 137. The Terms "Spark" and "Arc" 217 245 CHAPTER X PROPAGATION OF THE WAVES OVER THE EARTH'S SURFACE 1. Over Plane or Spherical Homogeneous Ground 138. Ground Having Plane Surface and High Conductivity 246 139. Over Flat Ground of Not Very High Conductivity 248 140. Effect of the Spherical Shape of the Earth 255 2. Wave Propagation over Uneven or Non-homogeneous Ground 141. Uneven Surfaces 258 142. Rain and Ground Water 260 143. Distribution over Land and Water 262 3. Effect of Atmospheric and Other Influences upon the Waves 144. Effect of the Condition of the Atmosphere 263 145. lonization of the Atmosphere 264 146. Actual Measurements of the Wave Propagation 269 147. Effect of Grounding the Transmitter upon the Wave Propagation . . 270 148. The Safety Factor 271 CHAPTER XI DETECTORS 247 1. Thermal Detectors 149. Bolometer and Thermogalvanometer . . 272 . 150. Thermocouples, Thermal Detectors . 273 151. Relative Importance of the Thermal Detectors 274 2. Magnetic Detectors 251 152. The Fundamental Physical Principles 274 153. Marconi's Magnetic Detector 274 154. Other Forms of Magnetic Detectors 275 3. Imperfect Contacts 155. Metallic Granular Coherer 255 276 156. Mercury Coherers 278 157. Carbon or Graphite Coherers. (Microphone Contact) 279 4. Electrolytic and Other Detectors 158. Anticoherers 279 159. The Electrolytic Detectors of FERRIE, FESSENDEN, NERNST, and SCHLOMILCH 280 160. Crystal Detectors 282 161. Incandescent Lamp Detectors. Gas Detectors 283 5. General Consideration of Detectors 162. The Nature of the Action in Various Detectors 285 163. What do the Different Types of Detectors React upon? 287 164. Testing the Sensitiveness of Detectors 289 xvi CONTENTS PAGE 6. Receiving Apparatus 165. Telephone Reception 290 166. Amplification of the Sound in Telephone Reception 291 167. Automatic Recording of Messages 294 168. Recording Apparatus for the Metallic Granular Coherer 298 169. Call Signals 299 170. Comparison of the Different Kinds of Detectors 300 CHAPTER Xll RECEIVERS 171. The Aerials of the Receiving Stations . 303 172. General Consideration of the Receiving System 304 1. The Original Marconi Receiver 173. The First Arrangement 307 174. The Marconi Transformer 308 2. Receivers for Tuned Telegraphy with Damped Oscillations 175. Receivers for Highly Damped Receiving Antennae 310 176. Receivers for Weakly Damped Antennae 313 177. Tuning the Receiver for a Double Wave Transmitter 314 178. Adjustment of the Energy Delivered to the Receiver 314 179. Receivers for Two Different Detectors 315 180. The Sharpness of Tuning 316 181. R. A. Fessenden's Method for Maintaining Secrecy of Telegrams. . . 323 182. Multiple Telegraphy 324 183. Methods for Overcoming Atmospheric Disturbances 326 184. Achievements of Tuned Telegraphy 328 185. Methods for Preserving Secrecy of Messages 330 3. Receivers for Undamped Oscillations 186. General 332 187. Methods Employing the Ordinary Detector 333 188. The Ticker '334 189. Construction of Interrupter for Ticker Method 335 190. Special Arrangements for Undamped Oscillations 335 191. Practical Achievements 336 CHAPTER Xlll DIRECTIVE TELEGRAPHY 192. Characteristic of the Distance Effect 338 1. The First Attempts 193. Use of Reflectors 340 194. Attempts at Screening 340 2. Methods Employing Several Antennce 195. The Field of Several Antennae General Consideration 341 196. The Field of Several Antennae Special Cases 342 197. Double Antennae, One-half Wave Length Apart 345 198. The Methods of E. Bellini and A. Tosi 347 199. The Methods of F. Braun . . 352 CONTENTS xvii PAGE 200. Production of Any Desired Phase Difference with Undamped Oscil- lations 352 201. Production of Any Desired Phase Difference with Damped Oscillations 353 3. Aerials Having Horizontal or Inclined Portions 202. Marconi's Bent Antenna 356 203. The Action of the Bent Marconi Antenna when Transmitting .... 357 204. The Bent Marconi Antenna Used for Receiving 361 205. Inclined Antennae 363 206. Horizontal Antennas. Ground Antennae 364 207. The Advantages of Directive Signalling 365 CHAPTER XIV WIRELESS TELEPHONY 1. The Transmitter 208. Source of Energy 371 209. Connections 371 210. Microphones 373 2. The Receiver 211. Connections 374 212. The Action in the Detector Circuit 376 The Development of Wireless Telegraphy During the Years 1909-1912 TABLES Table 1. The Natural Frequency of Condenser Circuits 384 " 11. The Natural Wave Length of Condenser Circuits 386 " 111. Frequency and Wave Length 388 " IV. Oscillation Curves for Various Decrements 389 V. The Spark (Arc) Constants 392 " VI. Equations for Calculations of the Coefficient of Self-induction . . 393 " VII. Effective Resistance of Copper Wires 396 " Vlll. Maximum Diameter of Resistance Wires 398 " IX. Gap Lengths and Corresponding Minimum Discharge Voltages . . . 399 X. Determination of Percentage Coupling 401 " XL Resonance Curve of the Current Effect 403 " Xll. Resonance Sharpness 405 " Xlll. The Radiation Resistance of Antennae 405 BIBLIOGRAPHY AND NOTES ON THEORY INDEX . 408 SYMBOLS AND ABBREVIATIONS The following explanations of symbols and abbreviations used in the text apply throughout unless distinctly stated otherwise : E = Electric field strength M = Magnetic field strength 8 = Electromotive force = e.m.f. IJL Permeability e = Dielectric constant o = Dielectric constant of air k = , usually referred to as the dielectric constant mf. = Microfarad c.g.s. = Units of the absolute electromagnetic (centimeter-gram-second) system S = Radiation W = Energy W e = Energy of the electrical field W m = Energy of the magnetic field V = Voltage V z = Ignition voltage / = Current (frequently i is used in the illustrations) r = Resistance L s = Coefficient of self-induction ! for stationary field C s = Capacity R = Resistance L = Coeff. of self-induction for oscillations C = Capacity i^ or Ls 2]L = Coeff. of mutual induction with quasi stationary current Z/i 2 or L 2l = Coeff. of mutual induction with non-quasi stationary current Ro = Gap resistance RZ = Radiation resistance K = Coeff. of coupling K' = Degree or percentage of coupling T = Period N = Frequency = number of complete periods per second co = 2-n-T = -^ = number of periods in 2ir seconds = circuit frequency X = Wave length VL = Velocity of propagation f = Discharge frequency = number of discharges per second d = (Logarithmic) decrement dj = Joulean decrement d h = Hysteresis decrement ds = Radiation decrement d g = Gap decrement a = Lineal decrement xix XX SYMBOLS AND ABBREVIATIONS a = Form factor of an antenna p = Sharpness of resonance a, b = Constants of the spark or arc e = Base of the natural (Naperian) logarithms a = Proportional to; varies as <^ = Much less than ^> = Much greater than EMS = Zenneck's "Elektromagnetische Schwingungen u. drahtlose Tele- graphic." Stuttgart, 1905. ETZ = Elektrotechnische Zeitschrift Jahrb. = Jahrbuch fur drahtlose Telegraphic. Leipzig. Joh. Amer. Barth. El. = The Electrician. London. C.R. = Comptes rendus de 1'Academie des Sciences. Paris. WIRELESS TELEGRAPHY CHAPTER I 1 THE NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 1. Oscillations Produced by Charging the Condenser. The simplest form of a condenser circuit is that shown in Fig. 1 : a condenser, C, and a conductor, AFB, joining the metallic coatings of the condenser. a. Let this circuit be broken at some point, F, and each side connected to one pole of an electric influence machine, an induction coil or an alternating-current transformer (Fig. 2). If then the influence machine or induction coil is put into operation, the condenser becomes charged, one of its coatings, say A, receiving a certain quantity of positive elec- To Induction Coil FIG. 2. tricity, the other, B, an equal amount of negative electricity. The resultant electrical field and difference of potential are obtained not only between the coatings A and B } but also between the poles FI and F 2 of the gap in the circuit. If the condenser charge and thereby the tension between FI and F 2 are gradually increased, a "spark" finally passes be- tween Fi and F 2 and the space FiF 2 , the "spark gap," becomes conductive. 6. The difference in potential between the coatings A and B produces an electric current in the direction of the arrow in Fig. 2, from the positive to the negatively charged coating. This holds good only at the start, however. For the current, assuming that the resistance of the conductors 1 WIRELESS TELEGRAPHY is not extremely high, is an oscillating or alternating current of the kind represented by the curve in Fig. 3, a photographic reproduction made by the aid of BRAUN'S Kathode Ray Tube, 2 which is specially adapted for such purposes. The absciss of the curve are proportional to the time, the ordinates to the current values at any instant. This alternating current is in one respect distinctly different from the alternating currents in ordinary commercial use as produced by A.C. generators, viz., it has a constantly decreasing amplitude. An alternating current of this kind is said to be "damped" to distinguish it from an "undamped" alternating current of constant amplitude. c. As every current produces a magnetic field whose strength, at least in the vicinity of the current-carrying conductor, is proportional to the current, it may be concluded that the magnetic field varies similarly to the current; it is a "damped alternating magnetic field." ''"... M. Time FIG. 3. During the period in which the current has the direction shown in Fig. 2, it must bring a positive charge from the coating A to B, and when its direction is reversed, its action upon the condenser coatings is also reversed. Hence the condenser charge also % oscillates and the electric field between its coatings is also a damped alternating field. d. The entire phenomenon, i.e., the alternating current with its accompanying alternating electric and magnetic fields is called an "electromagnetic oscillation." Oscillations, which, as in this case, may be produced in a condenser circuit without the influence of other oscillations, are said to be the "natural" or "free oscillations" of that circuit. e. With the arrangement of Fig. 2, the natural oscillations are caused by the spark. In general, however, the presence of a spark is not essential for the production of natural oscillations, which may also be obtained in a condenser circuit having no spark gap [Art. 109]. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 3 1. FREQUENCY 2. Experimental Determination of the Frequency. a. Even with condenser circuits, whose natural oscillations are too rapid to allow of photographic reproduction by the aid of a Braun tube (see Fig. 3) or an oscillograph, the "frequency" of the oscillation, i.e., the number of complete cycles per second, can be directly determined by means of a rotating mirror FEDDERSEN'S method if the condenser circuit contains a spark gap. The gap, placed in a horizontal position and viewed in a mirror which rotates about a horizontal axis e.g., fixed on the shaft of a small electric motor (Fig. 4) appears during a discharge in the form shown in Fig. 5. At those moments during which the current passing over the gap is a maximum, the most light is produced in its path, which is very dark when the current is at its minimum; so that the illumination FIG. 4. of the path of the discharge varies periodically with the current. Hence, in the rotating mirror, in which the successive images of the spark appear at different points, a row of alternately light and dark stripes is obtained. The distance between two adjacent light stripes corresponds to the time of a half period of the oscillation. If the image in the rotating mirror is photographed and if, from the number of revolutions of the mirror and the dimensions of the apparatus, the speed with which the image of the spark moves over the photographic plate is determined, then the time of one cycle and hence the frequency of the oscillations are easily obtained from the distance between two or more light stripes. This method is not only of great practical value, but is of special interest as having been used by W. FEDDEESEN S in the first experimental WIRELESS TELEGRAPHY m m FIG. 5. demonstration and study of the natural oscillations of condenser circuits, which constitute the foundation of the science of modern radio-telegraphy. b. Gehrke's incandescent oscillo- graph tube offers another method for the direct determination of the fre- quency of condenser circuits. 4 It consists of a glass tube of the form shown in Fig. 6a, with wire or sheet metal electrodes (Fig. 66) and filled with pure nitrogen under slight pressure. If current is sent through this tube, the length of the incandes- cent portion of the negative electrode is approximately proportional to the strength of the current. If the tube is connected through a sufficiently high series resistance (tube of water, R, in Fig. 7) to the condenser coatings, then the current passing through it, and hence also the length of the incandes- cence, are proportional at any instant to the voltage between the condenser coatings. By photographing the image of the tube in a mirror whose axis is parallel to the axis of the tube (Fig. 8)*, a picture of the form shown in Fig. 9 (H. DiESSELHORST) 4 is ob- tained. The distance be- tween the light stripes is a measure of the duration of a cycle (see a). As this tube absorbs considerable energy and as the length of the negative incandescence is not al- ways exactly proportional to the amount of current passing through it, it is adapted for demonstration purposes rather than for accurate measurements. * Fig. 8 shows oscillograph made by the firm H. Boas : the tube is in a box at the upper right and below this is the holder for trie photographic plate upon which the concave mirror, mounted on the shaft of the motor, reflects the image of the tube. FIG. 6a. FIG. /i (mm FIG. 7. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS c. More convenient but indirect methods for determining the frequency will be discussed later [Art. 71, 81]. FIG. 3. Calculation of Frequency (THOMSON'S Equation) . a. The follow- ing formula for the natural frequency, N, of condenser circuits has been deduced by LORD KELVIN (SiR WILLIAM THOMSON 5 ' 1 ) : ^ 777^ [Table I]* 1 CO = VLC t in which L is the coefficient of self- induction of the circuit and C its capacity, while o> is the number of oscillations in 2w seconds. Simi- larly the period, T, of the oscillation is given by FIG. 9. T= ~ = * This relation holds only if the damping is not extremely great (i.e., d < 2-n-) and, therefore, applies to all practical cases. The exact formula is: where d = the decrement [Art. 8]. t The wave length [Art. 19] is given b ' = 6*- VL CGS . . 10 10 cm. \/L CGS MP meters 59.61-v/L^rr , f- C 2 + Cz Ci + Cs + CG If all the condensers are of equal capacity Ci, then r^ r< so that the resultant capacity of the given combinations of four and nine condensers is the same as that of any one of the condensers placed in the circuit as shown in Fig. 1. d. Hence, so far as the resultant capacity is concerned, the combina- 8 WIRELESS TELEGRAPHY tions of Figs. 12, 13 and 2, assuming all the individual condensers to be of equal size, are exactly alike. The difference, this being one of the advantages of the combined series and parallel connections as compared to the single condenser, lies in the distribution of the charge among the individual condensers. In Fig. 2 the potential across the condenser is the same as that between the electrodes of the spark gap, while the potential across each condenser in Fig. 12 is only one-half the total gap potential and in Fig. 13 only one-third of the gap potential. 5- The Practical Importance of Thomson's Equation. THOMSON'S equation offers a very simple means for rough calculations in determining an approximate value of the frequency or, on the other hand, the value of the capacity required for a given frequency. Usually, however, it is not possible to determine these values with the accuracy required in practice. Not that the Thomson formula is inaccurate, but in con- denser circuits having no spark gap, for which alone the Thomson rela- tion holds good, the value of the capacity and coefficient of self-induc- tion are mostly not known exactly; in condenser circuits with a spark gap, the spark affects the frequency. a. The values of C and L substituted in Thomson's equation must, of course, be those which correspond to that frequency which is to be determined. For air condensers the capacity C, under the conditions prevalent in wireless telegraph work, is practically the same as the capacity C 8 of the same condenser holding a static charge, and can therefore be easily measured with sufficient accuracy. For condensers, however, having a solid or liquid dielectric, the capacity may vary widely with the frequency. The ratio between capacity C of a condenser in an oscillating circuit to its capacity C 8 for a static charge is termed the " frequency factor." When mica or micanite is used as the dielectric this factor may be as low as 0.7-0.8, and for certain kinds of glass it may differ considerably from 1.0, while for other varieties, as for example certain flint glasses, and particularly for certain oils, such as petroleum or well-dried paraffin oil, the frequency factor is practically unity. 6 b. That care must be taken in choosing the proper value of L, the coefficient of self-induction, for use in Thomson's equation follows from the fact that the value of L may be quite different for the same coil with an alternating than with a direct current [Art. 35]. A further complication arises from the fact that L is the coefficient of self-induction of the entire circuit. For example in the case of Fig. 10 this comprises not only the main conductor AFB, but also the con- denser coatings and their leads (ACi, BCi, AC 2 , BC Z ). However, if the circuit contains a coil of several turns this need not be considered, as the rest of the circuit adds very little to the relatively large coefficient of NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 9 self -induction of the coil and may be neglected without materially im- pairing the accuracy. But in some cases it is desirable to keep the self- induction of the circuit AFB (Fig. 10) as low as possible and it may happen, especially when using a larger number of condensers with their connections, that the resultant coefficient of self-induction is much greater than the value calculated from the general dimensions of the circuit AFB. c. The frequency of condenser circuits containing a spark gap may vary as much as 10 per cent, from that indicated by Thomson's equation (M. WIEN, H. RiEGGER 7 ). However, this variation is great only if the electrodes of the spark gap are made of copper or silver and if the gap itself is at the same time very short (say ^ 2 mm.)* the variation be- comes greater in proportion to the shortness of the gap and the small- ness of the condenser, other things being equal. If tin, zinc, cadmium and especially magnesium are used for the electrodes and if the gap length is greater than 4 or 5 mm., the frequency can be determined from THOMSON'S equation within a fraction of 1 per cent, for condenser circuits including a spark gap, assuming of course that the values of L and C are accurately known. 8 2. THE DAMPING 6. The Transfer of Energy. a. As long as the current, 7, has the direction of the arrow in Fig. 2, positive electricity is flowing away from the condenser coating A, so that the positive charge, + e, of this coating is decreasing. When the current is reversed this charge is increasing. The same applies to the potential difference, V, between the con- denser coatings, as this and the charge hold the well-known relation: e = CV If curves be plotted showing the variation of V and 7 respectively as ordinates with the time as abscissae, the results will be as in Fig. 14. Voltage and current have a phase displacement of practically 90. b. The energy W e , in the electric field of a condenser of capacity C, charged to a potential V, is known to be Similarly the energy W m in the magnetic field of a circuit whose coefficient of self-induction is L [9] , is known to be ^ W m = \ LI 2 * Translators' Note: Just these conditions prevail in the modern quenched spark gap. A. E. S. 10 WIRELESS TELEGRAPHY where / is the current flowing in the circuit. And the total energy of the field of the condenser circuit at any moment is equal to the sum of the energies of the electric and the magnetic fields, i.e., W = W e + W m c. Fig. 15 shows W e (broken line), W m (thin full line) and W, the sum of the other two (heavy full line). The current curve, /, from Fig. 14 has also been drawn in again for direct comparison. I I i I I I ! ' I ! I I i I I I I Time Timer At the start when the current is zero, we have W = W e i.e., the total energy of the circuit consists of the electric energy of the charged condenser. One-quarter of a period later the voltage is zero (Fig. 14) and the current is just at its maximum. We then have W = W m or the total energy of the condenser circuit is equal to that of its magnetic field. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 11 After another quarter period the current is zero again and W = W e , and so on. In short, the oscillations are really interchanges of the energy between the electric field of the condenser and the electromagnetic field produced by the current. 7. The Various Causes of Damping. a. If no energy were consumed by this transfer, the total energy W and also the current and voltage amplitudes, in view of the relations explained in Art. 66, would remain constant. Any consumption of energy, however, means a reduction in FIG. 15 the total energy, i.e., a decreasing amplitude, resulting in a " damping" of the oscillations. The question of the various causes of damping is therefore identical with the question of the various energy losses. b. The energy lost in the oscillations of condenser circuits can be divided as follows : that lost in 1. Heat developed in the metallic circuit. 2. The spark gap. 3. The insulation of the condensers.* 4. " Brush" leakage of the condensers. 5. Eddy currents induced by the alternating magnetic field of the current, f 8. Condenser Circuit without Spark Gap Damping Due to Heat Loss. a. The heat developed by a direct current / in a conductor of resistance r during the time t is rlH * And possibly also in the insulation of the coils [Art. 37c]. f The energy lost by radiation is extremely small [Art. 25e]. 12 WIRELESS TELEGRAPHY while for an alternating current during the time of one cycle T the heat developed is where R is the "effective" resistance and I\ ff is the mean value of 7 2 , I eff being the "effective" current. For undamped oscillations, the wave form being sinusoidal, /Vr -!/.. (where I max is the maximum amplitude of the current for that cycle) which relation, however, is also practically true of the damped oscillations to be considered in wireless telegraphy. Under these conditions there- fore the heat developed per cycle is 1 2 From Art. 66 and 6c it follows that the total energy transferred in one cycle (two alternations) is -2X2 LI 2 max = LI 2 max Hence the energy lost in heat is proportional to the total energy of the oscillations of a condenser circuit. 6. If the energy lost in heat is the only loss, then it can be demonstrated that the curve showing the decrease in the amplitude with time the amplitude curve" is an exponential curve. Its characteristic property is the fact that the ratio of the amplitude, A i, at the beginning of a cycle to that, A 2 , at the end of the same cycle remains constant during the entire oscillation, i.e., -j- = const. (1) The greater this ratio is, the greater is the percentage decrease in amplitude per cycle. Hence the value of this ratio is a measure of the damping. Instead of the ratio itself, however, it is customary to use the natural logarithm of its value: d = log nat. j- 1 (2) d is called the "logarithmic decrement" or simply "decrement" and where the heat loss is the only cause of damping, as in the preceding, it is distinguished as the "Joulean decrement," d,-. c. The value of the amplitude A at any time, t, is given by A = A* = A*' (3) NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 13 in which N is the frequency, e is the base of the natural logarithms and A Q is the "initial amplitude" when t = 0.* Fig. 16 shows the decrease in amplitude per cycle for different decre- ments, while in Table IV the oscillation curves have been drawn out for various decrements. d. It follows from a that the Joulean decrement can be determined from the ratio of the energy lost in one cycle to that transferred in the same cycle. Hence, substituting 7 for I max , we have a > LI __ B _R_ ~ 2L 1 2NL or replacing T by 123456789 10 Number of Cycles or Periods FIG. 16. (Art. 3a) we have or from the foot-note in Art. 3a RC X 9. Condenser Circuit with Spark Gap. Damping due to Spark. a. The curves AI and A 2 of Fig. 17 are the amplitude curves of condenser circuits containing a spark gap (J. ZENNECK IQ ) obtained with the Braun Tube, AI being for a circuit of very low, A 2 for one with higher ohmic resistance. Comparison with Fig. 16 shows a marked difference from the cases in which the damping is due to heat loss only. The amplitude * In Fig. 14, V in the upper, 7 in the lower curve. , , Af . n , Rohms . CMF Cftor . Rohm, . CMF Rohms t dj = 6007T 2 - = 5920 - or = ^7:7^ ' ^meters A meters lOU . . approximately. 14 WIRELESS TELEGRAPHY curve is now no longer an exponential curve but approaches a straight line more and more as the energy absorbed by the spark exceeds the energy lost as heat. 10 This condition is obtained when the spark-gap electrodes are of copper, brass, aluminium or silver, n while with magnesium electrodes the amplitude curve tends toward the exponential form (D. ROSCHANSKY 2 ). If the amplitude curve is a straight line the amplitude A at any time t can be obtained from A = A in which A is the initial amplitude and a is the "lineal decrement" which determines the decrease in amplitude just as d, the logarithmic decrement, does for the exponential curves. K 0.25 0.20- 0.15 31 i ^n L ^\ * / \ X X.. / f> \ "N, X / ^ 15 N X / -N / / ^ / / X / 0.10 0.05 w ^ ^^ \ V x X ^ \ N 2- ' ^' X X s _a- --&''' ' \ *7 X 3- _-0_- -u-- \" *x N Time- FIG. 17. 6. If the amplitude differs from the exponential form, this is evidence of the fact that the conditions for the absorption of energy in the spark are not the same as for absorption due to ohmic resistance, but are similar to those in an electric arc (A. HEYDWEiLLER 12 ). For this con- dition the energy A g , absorbed per second in terms of the current is A a = al + b (1) (a and 6 are constants for the particular spark gap [Table V]) which for larger currents becomes A, = al (2) But since A = IV V g being the tension across the gap, it follows from (2) that the gap NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 15 voltage remains practically constant during the entire series of os- cillations,* i.e., V g = a (3) c. The result is that the ratio of initial to final amplitude of the A same cycle or period, -j > and hence also [Art. 86] the decrement, do not A-2 remain constant for all cycles, f The increase of the decrement is shown by the curves BI and B 2 (Fig. 17) for the successive cycles as the ampli- tudes AI and A 2 die out. In short no one definite characteristic decre- ment exists for the entire series of oscillations. Nor can a definite resistance be ascribed to the spark gap for those cases in which the energy loss is not proportional to 7 2 . If then in practice, an equivalent gap resistance or simply a "gap resistance" R g is referred to, this is intended to mean that resistance which, if sub- stituted for the gap, would absorb the same amount of energy as is actually absorbed by the spark gap during the entire oscillation series following the same amplitude curve, t If the condition V a = a applies, and we have a straight-line amplitude curve, then R g = f- (/o being the initial amplitude). "trl o For the other extreme, when the energy loss in the circuit due to resist- ance is by far the greater and the curve is exponential we have (H. BARKHAUSEN 12 ) 8a A constant gap resistance R g would have [Art. 8d] a corresponding "gap decrement" d g = irR g ^Jj~ (4) so that the total decrement for a condenser circuit with spark gap would be d = dj + d g As this value of the decrement is constant for the entire series of oscilla- * This, however, does not hold during a single half period, but is approximately correct if V g be considered as the average value of the gap voltage for a half period. Even this average value does not remain absolutely constant for the entire train or series of oscillations, but gradually increases from cycle to cycle for copper and silver electrodes and gradually decreases with magnesium electrodes (D. RoscHANSKY 2 ). t For the extreme case, in which the energy lost in the gap is the determining factor and the amplitude curve is a straight line, we have the difference of A\ and Az, instead of their ratio, constant. %i.e., RgI 2 ef f [Art. 44] is the average energy actually absorbed by the spark gap during 1 second. 13 16 WIRELESS TELEGRAPHY To Induction Coil F, tions, it does not properly characterize the decrease in amplitude from cycle to cycle, but is the average value of the gradually increasing dec- rement, its use in practice being very convenient for the qualitative con- sideration of condenser circuits having a spark gap and corresponding approximately to the single and definite decrement which is a precise and sufficient characterization of the time-decrease of the amplitude for condenser circuits having no gap. d. Aside from the change in form of the amplitude curve caused by the spark, it has been observed that in a condenser circuit with gap the oscillations may abruptly cease as soon as the amplitude has fallen to a more or less small fraction of the initial amplitude. 10. Methods for Determining the Spark Gap Damping. Two methods have in general been used for measuring the gap damping and resistance. a. The first of these, the so-called reson- ance method, is based on a procedure for determining the total decrement, which will be considered in detail later [Art. 74, etc.]. \R The total decrement is first measured with and then without the spark gap [Art. 78c]; the difference of the two values obtained is then the gap decrement d g , from which the gap resistance, R g [see equation (4) Art. 9] can also be determined. b. The second 14 is the substitution method. In Fig. 18, F is the spark gap whose resistance is to be measured, A is a hot-wire ammeter and R is a very high ohmic resistance (or self-inductance) through which the condenser, C, can be charged in spite of the gap, but sufficiently high so as not to appreciably affect the oscillations passing through F. First the indication of A is noted with F in circuit. Then a variable non- inductive resistance is substituted for F and is adjusted until A has the same indication as before. The spark-gap resistance is then the same as the substituted resistance, if the coefficient of self-induction of the condenser circuit, the discharge frequency and the spark gap FI have been held constant in both cases [see Art. lie, 2]. 11. The Factors which Determine the Amount of Gap Damping. 15 a. Relation to the Current Amplitude. The gap resistance, other things being equal, and particularly for the same gap length, varies inversely with the current amplitude. Within the limits encountered in .wireless telegraph practice the relation - T. < between gap resistance and current amplitude is approximately accurate. FIG. 18. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 17 Assuming that the circuit contains only a single spark gap of constant length, the voltage amplitude must remain constant. If then the current amplitude is varied by changing the coefficient of self-induction of the circuit, then it follows from (1) that If, on the other hand, the self-induction is kept constant and the current varied by changing the capacity, we have I t!6] Thus an increase in the self-induction of the circuit as in the first case causes an increase in the gap resistance, while an increase in the capacity reduces the gap resistance. It follows that within the limits for which. the relation (1) holds, the spark-gap decrement is practically independent of the capacity and self-induction of the circuit, being de- termined only by the gap itself. b. The gap resistance and decrement are however not independent of the resistance of the circuit, both increasing for an increase of the circuit resistance. c. The effect of the gap itself upon the gap resis- tance depends upon: 1. The material of the electrodes. 2. The form of the gap, and if the electrodes are spheres, upon the radius of these. 3. The gas or medium through which the spark passes, and 4. The length of the spark. As to the material of which the electrodes are made, it has been found that copper and silver cause a very high, magnesium, tin and zinc, a very low resistance, while aluminium stands between these groups. The radius of spherical electrodes, particularly for long sparks, greatly affects the gap decrement; the latter is much smaller with balls of large radius than for small spheres, the gap length being the same throughout. For disc-shaped electrodes the decrement is practically the same as for spheres of very large radius. If the gap medium is hydrogen, a very high decrement is obtained, it being less with illuminating gas, carbon dioxide, air, oxygen and par- ticularly low for sulphur dioxide. For the same electrodes, i.e., of a given material and radius and for a given gas in the gap, the decrement becomes larger as the discharge voltage becomes smaller for a gap of constant length; or, again, with constant voltage the gap decrement becomes smaller as the gap is shortened. 18 WIRELESS TELEGRAPHY d. For the relation between the gap decrement and the gap length the substitution method [Art. 106] gives curves similar to B* in Fig. 19, if the length F in Fig. 18 is varied without changing the rest of the circuit; the increase in gap resistance with increasing length is at first very gradual, then quite rapid. If the gap decrement is determined directly by the resonance method [Art. 10a] for different gap lengths, a curve of the form of A* in Fig. 19 is obtained. Here we have the gap decrement first rapidly and then more slowly, falling off as the gap length is increased. The curve determined by M. WiEN 17 with spherical zinc electrodes is shown in Fig. 20. f The following explanation accounts for this difference in the results 0.06 0.05 0.04 0.03 0.01 \ 0.5 \ 1.5 10\ 11 lt\ 13 \U x 10* Volts 15 FIG. 20. Gap Length in Cm. obtained. In the resonance method giving curve A, the gap under in- vestigation is the only gap in the circuit and its length determines the voltage and current amplitudes. Hence as the gap length is varied the current is correspondingly changed and the gap resistance is determined for a different current amplitude with each observation. An increase in the gap length alone would cause an increase in gap resistance, but an * Abscissae <* gap length, ordinates <* gap resistance. f The values given are for the following condenser circuits: Radius of Electrodes C L Circles O 220mm. 4.25 X 10~ 4 MF. 40,900 C.G.S. Units Crosses X 50 mm. 4.25 X 10~ 4 MF. 40,900 C.G.S. Units Dots 50mm. 6.3 X 10~ 4 MF. 40,500 C.G.S. Units Circles with dots G 50mm. 5.8 X 10~ 3 MF. 40,500 C.G.S. Units Squares D 50mm. 5.8 X 10~WF. 7,300 C.G.S. Units NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 19 increased gap length also means increased current amplitude, which latter decreases the gap resistance. We therefore have two factors with opposite tendencies; with short gaps the action of the current amplitude is the more effective of the two, but as the gap becomes longer it is partly compen- sated by the effect of the greatly increased gap length. In the substitution method (Fig. 18), current and voltage are deter- mined by the length of the gap FI, and therefore practically independent of the gap Fj whose resistance is to be measured. Hence as the length of F is varied the current amplitude remains practically constant. In addition there is an essential difference in the nature of the two methods. In the substitution method we find that resistance which, when put in place of the spark gap, produces the same c'urrent effect [Art. 43a]. The resonance method gives that value of the resistance 1 2 3^ 5 6 78 9 10 11 12 13 U 15 Length of Gap F in mm FIG. 21. which, when replacing the spark gap produces the same degree or rather sharpness of resonance in a loosely coupled secondary circuit [Art. 64c.] These two are not necessarily identical. e. From a and b the following conclusions in regard to the substitution method may be drawn: 1. The gap resistance of F (Fig. 18) is dependent not only upon its dimensions, the capacity and the self-induction of the circuit, but also upon the length of the gap FI (Fig. 18) upon which the current ampli- tude depends. How great this effect is may be seen from Fig. 21, in which the gap resistance of F is shown for several different lengths of FI. IB Any statement of the resistance of F without an accompanying statement of the dimensions of FI is therefore just as useless as stating that the resistance of a metallic filament incandescent lamp is so and so without mentioning the voltage or current at which the measurement was made. 20 WIRELESS TELEGRAPHY 2. Results obtained by the substitution method must not be con- sidered as conclusive in case there is only a single spark gap in the con- denser circuit. For when FI, having a larger resistance, is in the circuit, thi s has a considerable effect on the resistance of F. Furthermore the form of the amplitude curve and the conditions in the gap FI must be some- what influenced by placing an ohmic resistance as a substitute for the gap F, so that it does not follow from the fact that the current effect is the same in both cases, that the energy absorbed at F is also the same. 19 12. Spark Gaps in Series (Multiple Gaps). If a number of spark gaps are connected in series in the same condenser circuit the interesting question arises: Is the decrement for the several gaps in series greater or less than that obtained for a single spark gap with the same initial voltage? Investigations 20 intended to answer this question have shown that up to potentials of about 80,000 volts and down to capacities as low as 0.6 X 10~ 3 M.F. the series gap has a higher decrement than the simple gap. 13. Energy Losses in the Dielectric of the Condensers. 21 The alternating field produced in the insulating material (dielectric) between the coatings of the condensers by the oscillations, involves an energy loss for practically all insulators. It is due to the so-called " dielectric hysteresis" which is the electrical analogue of magnetic hysteresis. a. Such investigations as have been made so far with various materials indicate that, independently of the frequency of the oscillations, the energy absorbed per cycle in the condenser is proportional to the total energy in the condenser during that period. Hence the "hysteresis decrement," d h , i.e., that portion of the total decrement due to the di- electric hysteresis losses, is independent of the frequency of the oscilla- tions or the dimensions and capacity of the condenser, and is deter- mined solely by the dielectric material, that is by: 1. Its chemical composition, 2. Its temperature. As to the kind of material, it has been found that the hysteresis decrement is not appreciable for air, very small for well dried paraffin oil and transformer oil (dh = 0.001 0.002), also for good flint glass* (dh = 0.006 0.01) and somewhat greater for certain grades of hard rubber. It may run very high for certain kinds of glass, e.g., ordinary window glass, other grades of hard rubber, mica and the otherwise very convenient insulating material, micanite. Increasing the temperature causes an increase in the hysteresis decrement, at times in fact a very considerable increase. * Can be obtained from Molineaux, Webb & Co., of Manchester (Ancoats, Kirby Str.) and the glassworks at Ehrenfeld near Cologne. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 21 b. With certain materials the hysteresis decrement depends upon the energy load, We, the relation being of the form d h = a + fiW e* for some materials, and for others d h = aWf where a and (3 are constants for the particular material. In some cases this effect of the energy load is only an indirect one; due to the increased amplitude of the oscillations and the consequent increase in the energy absorbed, a higher temperature is produced, which in turn in- creases the hysteresis decrement. 14. Energy Lost by Leakage Discharge. a. For our purposes it is necessary to distinguish between leakage of two kinds. The first of FIG. 22. these 22 occurs whether the conductor is charged by oscillations or has a static charge. It consists of the well-known phenomenon of fine brush discharges emanating from conductors charged to a very high potential, particularly from edges or points. This is due to the fact that under the influence of the strong electric field the air becomes a conductor (ionized) and hence a part of the charge is led off. This phenomenon is observed very frequently on influence machines and occasionally in the air con- densers used in wireless telegraphy, wherever the plates have uneven surfaces or along their edges; especially also on antennae or coils charged to a high potential. * The " energy load," We, is the maximum energy contained per cc. of the dielectric material. It is where k = ratio of the dielectric constant of the material to that of air and E = strength of the electric field. 22 WIRELESS TELEGRAPHY The second kind of leakage discharge occurs only with oscillations; it appears at the moment when the oscillation commences, continuing even if the conducting elements in question do not have their potential increased during the remainder of the oscillation. A leakage discharge of this nature is observed in Ley den jars (see photographic reproduction in Fig. 22) or other condensers having a solid dielectric. It is char- acterized by long, fine, branching rays which spread out over the surface of the dielectric from the edge of the condenser coatings. 6. The essential requirement for a noticeable leakage discharge in both cases is a sufficient ionization of the air, which in turn means a sufficiently strong electric field. If sharp edges and points, at which the field strength assumes especially great intensities, are avoided as much as possible, these discharges need not be feared, as long as the potential is kept within a few hundred volts. c. When, however, a discharge as just described, occurs, it means an actual loss of energy in any case. How great this loss may become is not known. But what has been definitely determined is that if Leyden jars are properly constructed the increase in the decrement due to this energy loss can be kept below 0.002 for a voltage amplitude of 30,000 volts, and below 0.007 for 40,000 volts [Art. 86]. In addition it has been found that a discharge of the second kind has a tendency to produce a fluctuation in the frequency [Art. 79]. 15. Energy Lost by Eddy Currents. The alternating magnetic field produced by the oscillations in a condenser circuit induces so-called "eddy currents" in all conductors through which the lines of magnetic flux pass. The energy which these currents dissipate in the form of heat, other things being equal, is much greater for the high frequencies used in wireless telegraphy than for the lower frequencies customary in com- mercial power and lighting circuits. Being a direct loss to the total energy of the condenser circuit, it causes a corresponding increase in the damping. All conductors in the immediate vicinity of the condenser circuit, particularly those conductors (such as terminals) which are inside of coils where the magnetic field is concentrated, are subject to eddy currents. Very dangerous in this respect are the coatings of condensers which, in view of their extensive surface and thinness,* may cause con- siderable eddy current losses. Leyden jars are very troublesome on this account; with plate condensers it is much easier to place the coatings in such a position as to minimize the magnetic flux cut by them. Care should be taken in using such artificial insulating materials as are frequently substituted for hard rubber or marble, if placed in a strong high frequency magnetic field. The conductivity of such substances may be great enough to cause considerable energy losses. 23 * Very thick masses of metal are less dangerous. NATURAL OSCILLATIONS OF CONDENSER CIRCUITS 23 16. Relative Importance of the Various Energy Losses. a. Con- denser Circuits with Spark Gap. The important question here is: Which energy losses come into consideration as compared to the energy dissi- pated in the spark? If the conductors of the circuit are copper wires or tubes of sufficient diameter, it is safe to conclude that the Joulean (heat) decrement will be entirely negligible as compared to the spark gap decrement. Eddy current losses may become very considerable if provoked by clumsy connections or arrangements, particularly with the condensers. With proper care, however, these losses may also be so reduced as to have no material effect on the total decrement. In view of the high potentials for which condenser circuits with spark gaps are usually designed, a dielectric of high insulating quality is essential. If, then, compressed air condensers [Art. 396] or condensers filled with a good oil are used, the hysteresis decrement becomes negligible as compared to the gap decrement. The hysteresis losses in good flint glass are also much smaller than in the spark gap. As soon as other dielectrics are tried, however, losses comparable to or even greater than those occurring in the spark gap must be anticipated, especially if the condensers are to be highly charged. The energy lost by leakage discharge in condensers can, by careful design, be reduced to a quantity negligible in comparison to the gap losses. In any case this loss is, in general, far less important than that caused by the frequency fluctuations [Art. 86]. 6. Condenser circuits without a spark gap are usually designed for comparatively low voltages. For that case, losses due to leakage dis- charges usually disappear. Furthermore, as air condensers even at atmospheric pressure can generally be used, losses due to dielectric hysteresis can be entirely avoided. If, however, such potentials as may be encountered do produce leakage discharges, the losses may be far greater than those due to Joulean heat. The latter may be greatly reduced by the use of sufficiently thick and massive copper wires or, better yet, copper bands or strips, and especially by properly wound braided wire consisting of individually insulated conductors [Art. 36d]. This, however, tends to increase the eddy current losses, and if these losses are not minimized by the greatest care, it is not possible to bring the decrement below 0.01. In fact, decrements of about 0.003 are the very lowest attained in practice. CHAPTER II 24 OPEN OSCILLATORS In a condenser circuit, the condenser itself offers the only break in the continuity of the circuit. It is therefore usually referred to as a "closed oscillator" or "closed oscillating circuit," as distinct from systems in which the metallic conductor is not even approximately continuous and which are therefore "open oscillators." 1. THE LINEAL OSCILLATOR 17. The Fundamental and Upper Harmonic Oscillations. The simplest form of open oscillator is the straight lineal oscillator, i.e.j a straight metal wire or rod. If its two halves are given a positive and a negative charge respectively until a sufficiently high potential is reached, so that a spark discharge passes between the two halves (at To Induction Coil FlG. 23. Fj Fig. 23), an electromagnetic oscillation takes place here just as in a condenser circuit.* In general this oscillation is not a simple one, but is made up of a number of component oscillations of different frequencies, different current and voltage distributions and different electric and magnetic fields. It will therefore be necessary to consider separately the so-called "fundamental oscillation" that having the lowest frequency, as in acoustics and its "upper harmonics" or "upper partial oscillations." This subdivision in the treatment is further justified by the fact that each of these oscillations can be produced independently of the others. 18. Current and Potential Distribution in the Fundamental Oscilla- tion. a. In a condenser circuit, as used in practice, the quantity of * The natural oscillation can be induced in open oscillators as well as in condenser circuits without the existence of a spark gap in the oscillator [Art. 109]. 24 OPEN OSCILLATORS 25 electricity passing through a cross-section of the circuit in a certain length of time, just as in ordinary direct-current circuits, is practically the same at all points,* i.e., the currentf has the same phase and ampli- tude throughout the entire circuit. We can therefore speak of a definite phase and amplitude of the current just as for the alternating currents used in power and lighting circuits. In a lineal oscillator the current f may also be considered as having the same phase at all parts of the oscillator. But the current amplitude is entirely different at different parts of the oscillator. If a curve be plotted FIG. 24. giving the current amplitude at each point of the oscillator as ordinates, this "curve of current distribution " J is found to be an approximate sine curve (dotted line in Fig. 23). The current amplitude is greatest at the middle and zero at the ends of the oscillator. In other words, there are "current nodes" at each end and a "current anti-node" at the middle. 6. Correspondingly, if we plot the electric charge or rather the potential values along the length of the oscillator, we obtain the FIG. 25. "curve of potential distribution" as shown in Fig. 23 by the full line (sinusoidal) V. It should be noted that "potential" or "voltage anti- nodes " occur at each end of the oscillator, the " potential node " being at the middle. * Hence called a "quasi-stationary current." Compare footnote in Art. 24c. f The current = quantity of electricity passing through a cross-section of the circuit in 1 second. t This should not be confused with the " current curve" of Art. 16, which gives the variation with time. 26 WIRELESS TELEGRAPHY c. Just as in a condenser circuit, and for the same reasons, the current and voltage in an open oscillator have a 90 phase displacement. The distribution curves of the current and the voltage are shown in Figs. 24 and 25 respectively for successive eighths of a cycle, curves bearing the same number in the two figures being for the same instant. 19. Frequency of the Fundamental Oscillation. The simplest way to arrive at the fundamental frequency is by the following consideration : a. The current and potential distribution curves of Fig. 23 are of the same type as the so-called "stationary waves" encountered in other physical phenomena (as in acoustics). Such stationary waves result when two advancing waves of the same amplitude and frequency but of opposite direction occur simultaneously. The wave-length of the stationary wave is then the same as that of the advancing waves, if by wave-length of the stationary wave we understand twice the distance between two consecutive nodes or anti-nodes. As is well known, the "wave-length," X, of an advancing wave is equal to the distance traveled by the wave in one complete cycle. The propagation velocity, V L , is the distance traveled in 1 second. If the duration of a cycle or period is T seconds, then I/ T or N complete cycles occur per second. Hence we have the relation V L = N\=* (1) b. As we are j ustified in considering the oscillations of a lineal oscillator, as shown in Fig. 23, as stationary waves, we can apply equation (1), writ- ing it in the form N = ^ (2) N being the frequency. From a and as shown in Fig. 23, one-half the wave-length is equal to the total length, I, of the oscillator, i.e., c. The velocity of propagation of the electromagnetic waves occurring in air along a conductor 25 is practically equal to the velocity of light in air, hence: V L = 3 X 10 10 cm./sec. whence: Ar _ 3 X 10 10 cm./sec. . , ~2lU.) ~ This simple relation if not quite accurate is approximately correct.* * This and what follows is based on the assumption that the oscillator is in free. space, i.e., for practical consideration, its distance from conductors or high insulation must be large in comparison to its own dimensions. OPEN OSCILLATORS 27 20. The Electromagnetic Field of the Fundamental Oscillation. a. Direction of the Electromagnetic Field. The magnetic field is com- paratively simple, the lines of induction being circles whose axes coincide FIG. 26. with the axis of the oscillator. Fig. 26* shows the lines of induction in the equatorial plane | at a given instant. The lines of force of the electric field are shown in Figs. 27 to 30, for undamped oscillations!)! at each eighth period during one-half a cycle as calculated by M. ABRAHAM 26 and drawn by F. HACK 26 . Fig. 27 repre- sents the moment at which the charge of the oscillator is zero, while * This and the following figures do not indicate the falling in amplitude with distance [r, d]. f That is, the plane perpendicular to the oscillator at its middle. | For damped oscillations the nature of the phenomenon would not be notice- ably different. 27 28 WIRELESS TELEGRAPHY the current is a maximum; the following figures show conditions at each successive eighth period until after Fig. 30, Fig. 27 would again apply but with opposite signs, and so on. To better comprehend these figures consider first that at the moment of zero charge, as in Fig. 27, no lines of force emanate from the oscillator. Immediately thereafter, however, the oscillator becomes charged, for example as in Fig. 28, the upper half positively, the lower part negatively, and lines of force emanating from the upper half reenter in the lower half. This process continues cumulatively until the maximum charge OPEN OSCILLATORS 29 is reached at the end of the quarter period (Fig. 29) . Then the lines of force in the oscillator gradually decrease again until zero is reached after half a period. A part of the lines of force which have emanated from the oscillator (Fig. 30) during the first quarter period go through a peculiar contraction during the second quarter, assuming a kidney-like shape, and at the same time continue to move farther away from the oscillator. What happens to them as they pass off into distance is shown in Figs. 295 and 296, the first representing conditions at the moment of maximum charge, the second at zero charge. The advanc- ing lines of force gradually become arcs of circles. b. Phase of the Electromagnetic Field. Advancing Waves. Neither the magnetic nor the electric field has the same phase at any moment throughout the entire space affected. Both assume the form of a wave advancing out from the oscillator with the velocity of light. The following will explain what is understood by an advancing electromagnetic wave, in the simplest case, when the amplitude remains constant. If over each point of the line of direction (OX in Fig. 31) C\ FIG. 31. of the advancing wave we were to plot as ordinates the field intensity at any given moment, a sine curve, such as the full line curve in Fig. 31, would result. It represents the distribution of the field strength along OX at this moment. A moment later a similar sine curve is ob- tained, but slightly displaced from the first one in the direction of the advancing wave front (as shown by the arrow). This is indicated in Fig. 31 by the dotted line curve. Hence a conception of the process may be formed by considering the sine curve to move in the direction of and with the velocity of the advancing wave, its position at any instant indicating the distribution of the field intensity at that moment. From the preceding, it follows that at any one point there exists a simple alternating field whose frequency is VL \ N = in which X is the wave-length of the advancing wave and V L its velocity, in this case the velocity of light. 30 WIRELESS TELEGRAPHY It is evident that the phase varies from point to point.* It is the same, however, for two points lying in the direction of the advancing wave and separated by the distance of one wave-length, or a multiple thereof. If the two points are just a half wave-length apart, the phase difference will be 180. Or, in general, we have the phase difference is 2Trx 360.z , . ,, ,. , _ where x is the distance between the two points. A A Similarly, if x is the difference in the respective distances of two points in the equatorial plane from the oscillator, the phase difference between the fields at these points is also -r . A c. The Amplitude of the Field. Neither the amplitude of the magnetic nor that of the electric wave remains constant for different distances, r, from the oscillator; the amplitudes decrease as r increases. The magnetic wave amplitude in the immediate proximity of the oscillator roughly Distance below Surface of Wire in mm FIG. 49. decrease may be so rapid that practically the entire current is restricted to a very thin outer sheath of the wire (the so-called "skin effect"). Fig. 49* shows the drop in current density in copper wire as the depth from the surface is increased, for various frequencies. 35. Coefficient of Self -induction. 44 If the skin effect is very decided, there is practically no magnetic field within the wire. While for direct currents the coefficient of self-induction of the circuit is made up of two parts, one originating from the field inside of the wire, the other from the * The minimum wire radius for which the curves of Fig. 49 still hold good, at N = 0.5 X 10 6 /sec., i s about 3 mm.; at N = 2.5 X 10 5 /sec., about 1.6 mm.; and at N = 5.0 X 10 5 /sec., about 1.1 mm. With thinner wires, the drop in current density is not so rapid. 47 48 WIRELESS TELEGRAPHY field without the wire, for high frequency alternating currents the first part (which for non-ferromagnetic straight wires of length I cm. amounts to ~ C.G.S. units) practically disappears. No great error will be made if Z for straight or nearly straight solid wires the first part is neglected and the "effective coefficient of self-induction," L, for high frequencies is calculated by deducting ~ C.G.S. units from the value applying to direct currents (see Table VI) . For wires which are much bent, however, the relations are not so simple* (see Art. 37). If the development of a skin effect is prevented by the use of properly woven and twisted braid, consisting of individually insulated wires [Art. 36d], the effective coefficient of self-induction, L, for oscillating currents will not differ materially from that, L s , for direct current, this being true not only of straight wire circuits, but also of coils wound in a single layer. 45 36. Resistance of Straight Wires. A further result of the uneven distribution of the current is that the cross-section of the thin outer sheath, in which the flow of current is concentrated, rather than the section of the entire wire, determines its resistance to high fre- quency currents. In fact the so-called "effective" resistance, R, of a wire also called the alternating-current resistance for high frequency oscillations [Art. 8a], is something quite different from the resistance for direct current. This difference increases as the frequency, the radius of the wire, its conductivity and its permeability become greater. 46 a. Table VII at the end of the book gives the resistance of copper wires of various sizes and for different frequencies encountered in radio practice. The resistance of iron wire is much higher on account of high permeability so that for this reason alone its use in practice is forbidden. b. For very thin wires, particularly when made of metal having low conductivity, the effective resistance at radio frequencies is but little different from that for direct current, the difference decreasing as the size of wire decreases. In Table VIII are given those sizes of wire of different material and at different frequencies for which this variation from the direct-current resistance is just 1 per cent. Resistances 47 which are practically non-inductive and practically independent of the frequency can be made up of thin wires of constantan, manganin and nickelin for small currents, while braids of these wires individually insulated, arc lamp carbons, graphite rods and also glass tubes containing an electrolyte, such as CuSO 4 solution, serve for larger currents. * The coefficient of self-induction of coils made of heavy wire may be about 20 per cent, less for high frequency oscillations than for direct current. 45 Formulae for the coefficient of self-induction, L, of coils are given in Table VI. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 49 c. The following conditions are closely associated with the skin effect : 1. A copper tube with walls not extremely thin has, to all intents and purposes, the same resistance as a solid wire of the same diameter and material (i.e., of course, for high frequency currents). 2. Tinned copper wire is not desirable, as the current is carried mostly by the poorly conducting tin, making the resistance higher than for the untinned wire. 3. Copper-clad steel wires have a resistance only very little higher than copper wires, and combine the high conductivity of the copper with the greater tensile strength of the steel, which is very advantageous for antennae submitted to high wind stresses.* d. An important difference between the resistance for direct currents and that for high frequency currents lies in the relation to the wire radius, r, for in the first case the resistance c 2J while in the latter case (for wires not too thin) it oc -. . r In other words the direct-current resistance simply depends on the total cross-section of the conductor, whether this is a single wire or made up of a number of wires in parallel giving the same total cross-section. For oscillating currents it is preferable to replace heavy solid wires or tubes by braids of very thin individually insulated wires or flat bands made up of such braids woven together, f But care must be taken that the current does not distribute itself much the same as it would in a solid wire, i.e., mainly in those of the smaller wires lying near the outer sur- face. This is provided for by so twisting and interweaving the component wires that each of them lies at the outside just as many times as on the inside of the circuit, resulting in a uniform current amplitude in all the wires. Furthermore, while for direct currents the resistance, aside from the specific conductivity of the material, depends only on the area of the cross-section, the form of the section also plays a part in determining the effective resistance of a conductor carrying high frequency oscillations, e.g., thin copper bands 49 in general have a lower resistance than cylindrical copper wire of the same cross-sectional area, though the resistance of the bands also increases rapidly with increasing frequency unless they are exceedingly thin. * For example, the antenna of the Eiffel Tower has galvanized steel wires. f The first suggestion to use woven ropes of thin insulated wires for high frequency circuits probably originated with N. TESLA. SO F. DOLEZALEK was the first to intro- duce them into actual practice. Braided wire of this kind is furnished by many manufacturers, but by no means always of equal value. Braids of enameled wire (i.e., wire having very thin enamel insulation) of 0.07 mm. diam. are very satisfactory. 4 50 WIRELESS TELEGRAPHY 37. The Resistance of Coils. 45 The only conductors having appre- ciable self-induction encountered in radio circuits are usually in the form of either "cylindrical coils" (Figs. 50 and 51) or "flat spirals" (Figs. 52, 53 and 54; see also the much used form in Fig. 236 marked "28"). a. If these coils are made of solid wire the current distribution over the cross-section is subjected to a fur- ther complication as compared to the simple straight solid wire. The current amplitude is no longer dis- tributed symmetrically to the wire's axis but is consid- erably greater on the inner side of the coil than on the m m outer side. This results in a further increase of the re- sistance, so that the effec- tive resistance of coils as used in radio work is apt to run as high as one and one-half to two times that of the same wire when unbent. The dissipation of energy and hence the effective resistance of coils is considerably increased if they are so constructed that a large proper- FIG. 50. FIG. 51. FIG. 52. FIG. 53. tion of the magnetic force cuts the turns of the wire (as for example at the ends of the coil). In this case, however, the effective resistance may often be reduced by the use of a wide copper strip or band in place of wire having a THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 51 circular section, or better yet, conductors made up of small, individually insulated twisted wires (or braids or bands woven out of such conductors), the thinness of the individual wires, the method of twisting and inter- weaving them and the form of the coil determining the resultant effective resistance. 6. With coils wound in several layers a further loss due to dielectric FIG. 54. hysteresis may be added. With alternating current a relatively high difference of potential exists between adjacent layers, causing a cor- respondingly intense alternating electric field which may result in energy losses in the insulating material affected. For this reason and also be- cause they otherwise tend to increase the energy losses, coils wound in several layers are not generally desirable. 38. Coils having Variable Self-induc- tion. 52 a. Changes of the self-induction in large steps are most easily attained by varying the number of turns connected in circuit, say through the use of plug or clip contacts; thus in Fig. 55 the current enters through A and leaves the coil either at B or CorZ). If the plug contact is at B, then the portion BD together with parts of the cur- rent circuit may constitute an oscillator which is directly coupled [Art. 526] with the current circuit; the oscillations of this system may at times produce undesirable disturbances. Furthermore, losses may re- sult from eddy currents induced in the free portion (BD) through which flows the magnetic flux generated in the connected portion (AB). It is 52 WIRELESS TELEGRAPHY therefore advisable to so choose coils that the free end, BD, does not become too long. Under no circumstances should the variation of the self-induction be obtained by short-circuiting a number of the turns (e.g., BC, Fig. 56). Heavy currents would be induced in the short-circuited por- tion, causing a large energy loss. 6. Self-induction variations in small steps may be obtained by the use of sliding con- /** FIG. 57. FIG. 58. tacts. Fig. 57 shows this method as applied to a cylindrical coil and Fig. 58 to a modification of this, the "ring coil." The latter has the advan- tage of enclosing practically all its lines of magnetic force, thereby minimizing eddy current losses in neighboring conductors and disturb- ances in near-by circuits. The ring coil, however, involves greater construction difficulties. Care must be taken with coils of the form of Figs. 57 and 58 that the sliding contact provides good conductivity and that it does not touch more than one wire at the same time, thereby short-circuiting the turn included between them. c. A uniformly gradual change of the self-induction is attainable in a particularly simple manner with flat coils (Fig. 59) which are pro- vided with a rotating arm, K, and a movable sliding contact, SC, for this purpose. Cylindrical coils may also be so arranged, in that the coil is rotated about its axis, the turning causing a sliding contact to move up and down along its length as in the Kohl- rausch bridge or by having the wire of the coil, which is bare and flexible, wound and unwound to any desired extent on a bare metallic cylinder or roll, as in the Wheatstone resistances. FIG. 59. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 53 The arrangement used most widely for gradual variation of the self-induction (called " variometer" in radio practice) consists of two coils connected either in series or in parallel and whose relative position to FIG. 60. each other may be varied. Fig. 60 shows a variometer designed by G. SEIBT and C. LORENZ, in which one of the coils is turned around inside of the other. The self-induction of these is a maximum when the two coils stand parallel to each other and the current flows through both in the same FIG. 62. direction, and is a minimum when the coils are still parallel but carry the current in opposite directions. Another similar method is sketched in Fig. 61; the two cylinders shown are intended to be placed one inside of the other, one of them being turned on its axis. A widely used arrange- 54 WIRELESS TELEGRAPHY ment is shown in Fig. 236 (TELEFUNKEN*) in which the middle one of the three flat coils (marked "28" in Fig. 236) can be swung from side to side.f A particularly elegant construction is found in the variometer devised by R. RENDAHL 63 (Telefunken) . Two flat coils wound as shown in Fig. 62 are mounted face to face on a common axis (in Fig. 62 they are shown next to each other instead of face to face) . FIG. 63. FIG. 64. One of them turns on its axis. If it is turned so that those halves of the two coils carrying the current in the same direction are superimposed, the coefficient of self-induction will be at its maximum. If turned 180 from this position, the coefficient of self- induction becomes a minimum. The advantage of this variometer lies in its compactness (Fig. 63 shows the manu- factured instrument for heavy currents and rather high potentials) and in the low stray magnetic field outside of the coils; by alternate series and parallel connection of the coils a very wide range in the self-induction can be obtained. 39. Condensers of Constant Capac- ity. 52 a. Plate Condensers. Plate con- densers for large capacities with paper as the dielectric are adaptable only for low voltages, unless a sufficient number are joined in series. Otherwise mica (Fig. 64) or glass plate condensers with coatings of tin-foil or thin sheet metal are used. Mica as the insulating mate- " Telefunken" is the trade-name of the German Company of Wireless Teleg- raphy " Gesellschaft fuer drahtlose Telegrafie, m.b.H.," Berlin. t Translator's Note: This is sometimes referred to as the "butterfly" type of variometer coil. FIG. 65. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 55 rial, in view of its very high resistance and its comparatively high di- electric constant, permits of very small dimensions* but causes quite heavy losses through dielectric hysteresis if the load is not kept very low. If the condenser losses are to be minimized, air or oil must be used as FIG. 66. the insulating material. Two constructions of air condensers are shown in Fig. 65 (GIEBE) and Fig. 66 (E. HUTH) respectively. A somewhat different arrangement is shown diagrammatic ally in Fig. 67 for an oil condenser as designed by J. A. FLEMING. With air condensers great care must be taken that the advantage of practically no energy dissipation is not lost by leakage discharge (of the first kind described in [Art. 14o]) or poor in- sulation of the non-conducting parts which serve to hold the plates in position. It is advisable to enclose these condensers in containers of glass or the like and to dry the air within thoroughly by means of metallic sodium. An air condenser for high pressures (compressed air condenser) as built by the National Electric Sig- nalling Co., at the suggestion of R. A. FESSENDEN is represented in Figs. 68 and 69. (See b for the ad- vantages of compressed air.) b. Cylindrical condensers. The best-known form of cylindrical condenser, the Leyden (also the * The dimensions of a mica condenser for a breakdown potential of 1000-1500 volts for example are 26 X 54 X 8 mm. for about 0.01 MF., 26 X 54 X 14 mm. for about 0.2 MF. (C. LORENZ). FIG. 67. 56 WIRELESS TELEGRAPHY Kleit) jar, has glass for its dielectric. Glass is chosen in view of its low (dielectric) hysteresis losses and its low conductivity, while the form of the jar is chosen on account of its low leakage discharge [Art. 86]. In this connection a long narrow form of jar is always the most ad- vantageous (note the battery of jars, as built by TELEFUNKEN, in Fig. 70). The Leyden jars of J. MOSCICKI* (Fig. 71) in which the upper ends are made narrower and heavier than the main body (Fig. 72) are particularly FIG. 68. FIG. effective in minimizing the leakage discharge. The thickening of the glass at the top also increases the breakdown voltage as experience has shown that Leyden jars mostly break down at the top edge of the coatings. The construction of these jars is evident from Fig. 72, in which PI and P 2 are the terminals of the two coatings, L is a metal tube, b is a rubber stopper and / is a porcelain insulator. The coating consists of a thin layer of silver chemically deposited and covered by a thicker * Manufactured by Messrs. WOHLLEBEN & WEBER, in Saarbriicken, from whose descriptive pamphlets Figs. 71 and 72a and b are taken. 66 THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 57 layer of copper. The jars are filled with a mixture of distilled water and glycerine having a low freezing point and serving to secure a good contact between the terminal PI and the inner coating. This also pre- vents a too rapid heating of the jar. Another method for raising the breakdown voltage and minimizing r\ V FIG. 72a. FIG. 727>. FIG. 70. FIG. 73. the effect of leakage discharge is shown in Fig. 73 (Allgemeine Elektri- zitatsgesellschaft). The dielectric is split into two parts at the top, the outer part being bent out like an umbrella. For purposes requiring particularly low energy loss the very handy compressed gas condensers as designed by M. WIEN IT are very conven- ient. Their design is shown in Fig. 74. The use of carbonic acid gas, 58 S, 3 Section A-B FIG. 74. WIRELESS TELEGRAPHY Volt 4 6 8 10 12 U 16 18 20 22 Pressure in Atmospheres FIG. 75. FIG. 76. FIG. 77. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 59 under a pressure of twenty atmospheres, greatly raises the breakdown voltage, bringing this even higher than for oil filling, so that these condensers may be used without difficulty up to about 35,000 volts, in spite of the very small space (3 mm.) between the cylinders. (See the curve in Fig. 75 for the relation of break- down voltage to gas pressure in con- densers of this type.) The high pres- FIG. 78. FIG. 79. sure also reduces the leakage discharge to such an extent that it has not been possible to measure it up to potentials of about 35,000 volts. 57 40. Variable Condensers. 52 Conden- sers whose capacity is changed in steps, as that shown in Fig. 76, are seldom used. Instead of this, it is customary to use bat- y FIG. 80. teries of Leyden jars, and vary their number according to the required capacity. a. Continuous variation in the capacity of condensers is usually ac- 60 WIRELESS TELEGRAPHY complished by varying the relative position of the two coatings or con- ducting plates. This form of condenser was probably first introduced into radio practice by A. KopSEL 57a in the form represented by Fig. 77. The conducting elements are made up of sets of semicircular plates or discs of which one is stationary and the other rotated into the spaces between the plates of the former. A pointer moving over a circular scale (see Fig. 77) indicates the po- sition of the movable element. The first form of this type of condenser built by TELEFTJNKEN is shown in Fig. 78. They are now made by many firms. For instance, Fig. 79 shows a construction developed by C. LORENZ, Fig. 80 represents a precision condenser of G. SEIBT, and Figs. 81 and 82 show an arrangement with vertical plates made by C. LORENZ. The latter is said to allow of a better circulation of the FIG. 81. oil and to prevent air bubbles from collecting on the plates. The con- denser plates as made by H. BOAS (Fig. 83) are also vertical, but cylin- drical in shape. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 61 The desire to combine maximum capacity with minimum space underlies the design of Fig. 84, C. LORENZ. It consists of a combination of two (or three) condensers of the form of Fig. 77 in such manner that the two movable sections g and h occupy a common space in one position. FIG. 83. The problem of minimizing space is solved with particular nicety in the condensers of the Marconi Co. These also have the movable section made up of semicircular plates, similar to those of Fig. 77, but differ FIG. 85. in having two stationary (AiA 2 , Fig. 85) and two rotating (B iB 2 , Fig. 85) sections arranged as shown. One stationary and one rotating set, say AI and J5i,are connected to one terminal, while the others, A 2 and B 2 , are joined to the other pole. Then the capacity is greatest 62 WIRELESS TELEGRAPHY when BI entirely covers A 2) B% covering A\. This capacity, for the same total volume occupied and the same distance between plates, is double that of a similar condenser having only one stationary and one movable section of plates. 2. CURRENT AND VOLTAGE 41. Relation between Current and Voltage Amplitudes undamped sinusoidal oscillations the relation is given by 7o = For (1) in which R and L are the resistance and co- efficient of self-induction respectively of the circuit whose end-points have a difference of potential F . For damped oscillations within the limits encountered in practice,* this relation also holds approximately. It assumes an even simpler form for all wire circuits, unless these consist of particularly thin wires of low con- ductivity, as in these the inductance, in view of the high frequencies customary in radio practice, increases much more rapidly than the resistance. We may therefore write ap- proximately : /o = (2) b. If a current / (Fig. 86) divides itself into two paths one having a resistance ^i and a coefficient of self-induction LI, the constants of the other being R 2 and L 2 , then we have for the ratio of the currents I\ and 1 2 in each of the parallel paths (3) + If both branches are made of fairly heavy wire, then the lower the resistance is in comparison to the inductance, the more nearly accurate will be the approximate relation so that the splitting of the current depends not upon the resistance but upon the coefficients of self-induction of the branches. * i.e., d is much less than 2ir. t It is assumed that the two branches do not affect each other inductively. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 63 The two branches may be intentionally so adjusted that the resistance of one of them, say Ri, is much greater than its self-induction while the reverse is true of the other branch. We then have: /2 I-leff Rl This gives a simple measure of w and the frequency N* from the ratio of the branch currents. It has been frequently suggested to make use of this relation for measuring the frequency by noting the current indicated by ammeters, A i and A 2 , in each branch circuit. This scheme is very neatly carried out in FIG. 87. FERRIE'S "frequency-meter," which gives direct readings of the frequency. 62 The two ammeters are so arranged that their pointers AI Zi and A 2 Z 2 (Fig: 87) cross each other. At a given frequency, N , for any deflection, i, of the pointer of AI only one definite deflection, 2 , of the other instrument, A 2 , will correspond, so that the pointers will intersect at a definite point. For another current of the same frequency passing through the system, other deflections, /?i and (3 2 (dotted lines in Fig. 87), and another definite point of intersection correspond. By thus varying the current at a constant frequency, the successive points of intersection develop a curve (/ in Fig. 87) which geometrically locates the frequency N on the face of the instrument. By repeating this process with other frequencies, individual curves (II, III, etc., Fig. 87) are obtained for each frequency. These curves once determined to measure an unknown * The same in fact is true of the more general equation (3). 64 WIRELESS TELEGRAPHY frequency it is only necessary to observe on which curve the instrument pointers intersect, which will indicate the desired frequency. If one of the paths contains a coil of very high self-induction, while the other path contains neither high self-induction nor high resistance, then the oscillations will flow through the second path almost entirely. The first path is said to be " choked" (high frequency "choke coil"). c. In applying equation (2) to an entire condenser circuit (AFB, Fig. 1) the difference of potential between the condenser coatings* must be taken for V and the coefficient of self-induction of the entire circuit substituted for L. If capacity is introduced in place of self -inductance, we have: Jo = coC.Fo 63 (4) To illustrate the application of this formula, consider the condenser circuit formerly installed at the German station in Nauen (TELEFUNKEN). Its effective capacity was 0.44 MF., the frequency about 1.5 X 10 5 /sec. With 60,000 terminal volts, we have 7o = 27T X 1.5 X 10 5 X 0.44 X 10~ 15 X 60,000 C.G.S. units = approx. 2500 C.G.S. units = 25,000 amp. It should be noted that the current amplitude is very great, even from the standpoint of commercial light and power circuits. d. Equation (4) holds in general for any condenser in the circuit, if C is its capacity, V the potential difference of its coatingsvand 7 is the current in the circuit. If the capacity C is very great, F becomes very small ; in this case the condenser acts as a short circuit for the oscillations, while it would offer an infinitely great resistance to a direct current. It may therefore be used to "block" or protect the circuit against a direct current without appreciably affecting the oscillations. 42. The Breakdown Voltage and Gap Length. 64 A given voltage, V, say that existing across the plates of a condenser, may be measured by its "breakdown gap" i.e., the length of a gap in air or gas over which the voltage V would just discharge itself. f The relation between the length of the gap and the breakdown potential depends on the form of the electrodes (on their radius in case of spheres), on the particular kind of gas in the gap as well as its condition, and the method of charg- ing the electrodes, i.e., whether a static charge has been supplied by a friction or influence machine, or whether the charge is produced by os- cillations or by an induction coil. * For several condensers in series this would be the sum of their potential differences. t This is also known under various other names such as "discharge voltage," "rup- ture voltage," and is also identical with the "ignition voltage," V zt mentioned in Art. 129. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 65 a. The relation of gap length and breakdown voltage for air and static charges is given in Table IX. From these curves it will be noted that for short gaps the size of the electrode (radius of the sphere) has but little effect. Its importance in- creases, however, with each increase in gap length, so that with very small spheres the breakdown voltage increases only very slowly for increasing gap length, while with large spheres it remains in approxi- mate proportion to the gap length up to much greater distances. With plate or disc electrodes (Fig. 88) the relations are similar to those for spheres of very large diameter. b. If the charge on the electrodes is produced by oscillations, the relation between breakdown potential and gap length is also affected by the fre- quency. The higher the frequency, the higher is the voltage necessary to jump a gap of given FIG. 88. length. 65 This is due to the fact that when the voltage is reached at which a discharge would finally occur if this voltage were maintained, i.e., the normal breakdown voltage (Table IX), the discharge does not take place immediately and the voltage will have risen above the normal dis- charge value at the instant at which the discharge actually takes place.* This phenomenon is called "retardation" or "lag of the discharge" (E. WARBURG). 66 It plays an important part in wireless telegraphy, as in radio practice the high potential usually exists only for very brief periods (e.g., in induction coil interrupters, alternating-current transformers and even more so with high frequency oscillations). The cause of this phenomenon lies in the low number of ions contained by the gas in the gap. Its occurrence can be prevented by providing a sufficient quantity of ions in the gas. This is most easily attained by subjecting the negative electrode (both electrodes in the case of alternating-current operation) to ultraviolet light, thereby inducing the emission of negative electrons. This method is advisable wherever it is important that the spark discharge occur always at the same potential. In fact, if properly applied even for radio frequencies the breakdown voltages and gaps will be practically the same as for static charging, and the values of Table IX may be applied without appreciable error. 65 c. The discharge voltage is reduced under the conditions encountered in radio practice by heating the electrodes, in fact by any strong ioni- zation of the gas. In practice ionization is usually produced by im- mediately preceding discharges. If a number of spark discharges are passed over a gap in rapid succession, the voltage may be reduced very considerably from that required for the initial discharge. The breakdown potential may be increased by raising the pressure * Apparently the phenomenon described in c is also due to this condition. 66 WIRELESS TELEGRAPHY above atmospheric. Up to about 10 atmospheres the discharge voltage is approximately proportional to the pressure [see Art. 396]. The breakdown potential is not much different for various gases such as air, nitrogen, oxygen, carbon dioxide, etc. However, it is only about one-half as great for hydrogen as for those mentioned, and much lower still for helium and argon. d. So-called "micrometer gaps" as illustrated in Fig. 89 serve for measuring the breakdown gap. KiK% are the spherical electrodes, GiG$ good insulators of glass or, better yet, porcelain, Si a micrometer screw, Sz the lever head of a set screw, not otherwise visible in the illustration. If S 2 is loosened the electrode on the left can be moved away from or nearer to the other electrode, while the micrometer screw, Si, serves for the fine adjustments. FIG. The radius of the spheres Ki and K% should be chosen at least as great as the gap length being measured. Moreover, the field between the electrodes must not be disturbed by any conductors in its vicinity if results for general comparison are desired and the values of Table IX are to be used. 43. Insulation of Conductors. a. In view of the high voltages which occur when working with damped oscillations, there is often great danger of a spark discharge between two points of the circuit. Hence the con- ducting circuit must be carefully insulated against spark discharges. For example, if a spark jumps across from A to B in the condenser circuit shown in Fig. 90, practically the entire current will flow via AFiB, as this path offers a much lower impedance than the path ADB, and thus the entire oscillation will be changed. 6, On the other hand, insulation against current losses in circuits THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 67 To Induction Coil charged by damped oscillations is not so essential 67 as it is for high tension direct current or commercial alternating current or even for undamped high frequency current. For instance, if A and B in Fig. 90 were joined by a poor insulator, say a wooden strip, this would not perceptibly impair the oscillations, in spite of the high voltage developed between A and B, for the length of time during which the potential between A and B is at all high is so short for damped oscillations of such decrements as come into question in practice that the loss across the strip of wood becomes very small unless the number of discharges per second is extremely great. Nevertheless, to insure against unnecessary energy losses the best insulating materials (porcelain and, second in rank, oil and hard rubber) should always be used. c. All parts subjected to high voltages from the induction coil or transformer must be insulated with the greatest care, otherwise very heavy losses may result. 67 In circuits having several condensers in series, only the portions FCi and FC 2 (Fig. 11) require heavy insulation; but if there is only one condenser or there are several in parallel in the circuit, then the entire circuit requires careful insulation. In this respect the connection of condensers in series may at times offer a considerable advantage. 3. MEASUREMENT OF CURRENT 44. The Indications of Hot-wire Instruments. a. Under hot-wire instruments, in the broadest sense, should be understood those instru- ments whose deflection is caused by the development of heat due to the current passing through a wire. The deflection of such an instrument is a measure of the average quantity of heat, Q, * developed per second. In general, the heat developed per second in a wire of effective resistance R is in which P eff is the mean value of 7 2 , the current effect. 68 For undamped sinusoidal oscillations smm D FIG. 90. i\tt = (2) so that Q-i * The deflection need not be proportional to Q, but is approximately so in most instruments. 68 WIRELESS TELEGRAPHY For damped oscillations whose amplitude curve is of the exponential form, the heat developed during one discharge 7?^ K 4Nd If then there are discharges per second, the total quantity of heat developed in 1 second o 2 (3) Comparing this with equation (1) we obtain /.// -ifew (4) For damped oscillations whose amplitude curve is a straight line 10 in which a is the lineal decrement [Art. 9a]. 6. As the effective resistance, R, of a wire depends on the frequency, the same is true of the indications of hot-wire instruments. These, however, can be made independent of the frequency (also usually making calibration with direct current possible at the same time) by the use of very thin wires [Art. 366] whose diameter is less than that given in Table VIII.* c. A hot-wire instrument calibrated with direct current gives direct readings for the current amplitudes of undamped oscillations, if the latter are approximately sinusoidal [equation (2)]. This is not the case with damped oscillations, as here not only the current amplitude but also the decrement, d, the frequency, N, and the number of discharges per second, f , enter as factors [equation (3)]. Only when these are known is it possible to calculate the current amplitude from the indication of a hot-wire instrument. d. A method for determining the frequency and the decrement for the case of exponential decrease of the amplitude will be given later [Art. 74, et seq.]. The number of discharges per second, when using induction coils or some form of motor-driven interrupter, is easily de- termined from the speed, on condition that each interruption corresponds to only one discharge, so that the number of interruptions and the number of discharges are identical. The same relation holds between the number of alternations and the number of discharges when operating with alternating current. In general, the number of discharges and the number of interruptions (or of alternations) are not identical. If the * The instrument's independence of the frequency is again destroyed as soon as a shunt is connected to the instrument for adjusting its sensibility. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 69 primary current is sufficiently strong, each interruption (or each half period of alternating current) will be accompanied by several "partial discharges" or "partial sparks." Whether or not this is occurring is easily determined by observing the spark image in a rotating mirror. If this appears as shown in the photograph reproduced in Fig. 91, there FIG. 91. are no partial discharges, while an image as shown in Fig. 92 indicates the presence of partial discharges.* If the image of the spark gap in a rotating mirror is photographed, FIG. 92. then the number of discharges per second can be calculated from the distance between the successive images on the photograph, the speed of the mirror and the dimensions of the outfit. If the spark itself is in- visible, an oscillograph (with incandescent lamp) or a Braun tube can be used in conjunction with a rotating mirror to count the discharge frequency. A more conven- ient indicator for this purpose is the discharge analyzer 'f of J. A. FLEMING, which consists of a GEISSLER (helium or neon) tube attached to the armature of a small motor. Fig. 93 shows the construction, Fig. 94 a finished instrument, as made by C. LORENZ. If the two terminals P\ and P% are respectively connected to two points of a condenser circuit or other oscillator, a high frequency current will pass through the helium tube (d, Fig. 93 1) which lights at each discharge. The speed of the motor is regulated to a point at which the image of the tube appears stationary to the eye. If it appears as shown in Fig. 95, it follows that there are four discharges during every complete revo- * With a little practice this can also be determined from the sound of the spark, which for partial discharges tends to become hissing rather than crackling. t Also frequently called "oscillation analyzer." t The metal rings m and n form the electrodes of a condenser, the rings k and i forming another. The tube is connected between these two. FIG. 93. 70 WIRELESS TELEGRAPHY lution of the motor, 510 while if it has the appearance of Fig. 96, there are four groups of three partial discharges each per revolution. FIG. 94. FIG. 95. FIG. 96. To Oscillator FIG. 97. Another simple and convenient method is that sketched in Fig. 97. 71 On the shaft of a small motor a photographic plate or film, P, is at- tached. Very near to this is a metallic point, S, which is conductively THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 71 connected to a point of the oscillator and is quickly moved across the plate by means of the handle H. Each oscillation is accompanied by a discharge between the point S and the plate, which, when developed, shows a series of dark points arranged on a spiral, each point represent- ing a discharge. From these points, knowing the speed of the motor, the number of discharges per second is easily obtained, independently of the velocity at which S is moved over P. 45. Commercial Hot-wire Instruments. Some hot-wire ammeters may be used for high frequency oscillations, without any shunt. It is preferable, however, to use instruments especially made for high fre- quency currents, as those, for instance, of HARTMANN AND BRAUN 72 (FRANKFORT A. M., Bockenheim, Germany). The type shown in Fig. 98 is intended for heavy currents, that FIG. 98. shown in Fig. 99 being designed for a minimum energy consumption. The scale of the former gives the value of I e /f in amperes, while the latter indicates the energy used in the instrument in watts,* which is propor- tional to Peff. In the latest and most sensitive instruments of this type, the energy consumed amounts to only about 0.015 watt. 46. The Hot-wire Air Thermometer. The air thermometer or hot- wire air thermometer devised by RIESS (Figs. 100 and 101) and brought into radio practice by F. BRAUN is a particularly simple laboratory instrument. It consists of a glass cylinder provided with an alcohol manometer and a glass stopcock, by means of which the difference be- tween the pressure within and the outside atmospheric pressure can be * This is not a sufficient excuse for the common misnomer of "hot-wire wattmeter" so frequently applied to this instrument. 72 WIRELESS TELEGRAPHY equalized. The hot wire, H, is at the bottom of the glass cylinder, 73 be- tween two heavy entrance wires which are led in through a stopper, the glass cylinder usually being surrounded by a vacuum chamber and sometimes in addition by a silver coating. Current passing through H heats this and also the air in the glass cylinder, causing an increase in the pressure, which is indicated by the manometer. These instru- ments are best calibrated with direct current. 47. Bolometer, Barretter. 74 The hot wire, w, in Fig. 102 is con- nected as one arm of a Wheatstone Bridge, which is adjusted until no current flows through the galvanometer, g. If now we send an alternating current, i, through w (AB), this wire becomes hot and its resistance in- creases, and the galvanometer deflection caused thereby is pretty nearly in exact proportion to the current effect of i. A somewhat different arrangement of this device, which is called a bolometer, is shown in Fig. 103. The branches pqrs and piqtfiSi which replace w and c in Fig. 102, respectively, are made of thin iron or platinum wire and as nearly alike as possible, and the arms pqr and s are so equalized that if direct current is ap- plied at E and F, the galvanometer g shows no deflection, so that the points C and D have the same potential with direct current. This arrangement has the following advan- tages: (1) The bolometer is less affected by variations in the room temperature, as pqrs and piqiTiSi are subjected to the same influ- ence; (2) at most, only a very small portion of the alternating current led in through A and B flows into the other circuits of the bridge or into the galvanometer,* as the points C and D remain at practically equal potential even with a variable current. On the other hand, the simpler arrangement shown in Fig. 102 has the advantage that the hot wire can easily be put into a vacuum in a glass tube. This greatly reduces the heat lost by convection, consider- ably increasing the sensitiveness (FESSENDEN, TISSOT). Similarly, the use of extremely thin wires in this arrangement is advantageous as com- pared to the method of Fig. 103. This also tends toward high sensi- tiveness. The calibration curve shown in Fig. 104 is that of a bolometer * Choke coils must be connected at each end of the hot wire for this purpose in the arrangement of Fig. 102. * FIG. 100. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 73 of BELA GATi, 74 having a gold wire* of 0.002 to 0.003 mm. diam., while a bolometer with a 0.0005 platinum wire gave a deflection of ten scale divisions for 0.034 milliampere, with the same galvanometer. B. S. FIG. 101. CoHEN 74 was able to measure currents as low as 5 X 10 3 milliamperes by means of a carbon filament in a vacuum. FIG. 103. * Galvanometer = movable coil galvanometer, direct reading; one scale division = 1 X 10~ 6 amp. BELA GATI makes use of a special compensation method of connection instead of the complete bridge arrangement. Using a single pivot galvanometer made by PAUL (London) (1 = 1 X 10~ 7 amp.) he obtained a deflection of 5 at 0.001 milliampere with the 0.0005 mm. platinum bolometer. 74 WIRELESS TELEGRAPHY 48. Thermoelement 75 or Thermocouple. a. KLEMENCIC adopted the form illustrated in Fig. 105* for the thermoelements used in the measurement of electric oscillations. A and B are thick wires through which the oscillations are led in, while the wires c and d connect to a galvanometer. ai 2 and 6i& 2 are very thin wires of different mate- rial (e.g., constantan and iron or 1.2 0.8 0.0 0.4 0.2 10 20 30 40 50 ~*"~ Galvanometer Scale Divisions FIG. 104. FIG. 105. constantan and platinum). If oscillations pass through the wires AB, the wires 6ia 2 become heated as do also the points of contact of the wires ai 2 and &i& 2 , the heat developed at these points of contact being FIG. 106. FIG. 107. greater than at the soldered points a^c and b z d. This uneven heating produces a thermoelectric EMF and a deflection of the galvanometer. * Greatly enlarged. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 75 b. The sensitiveness of these thermoelements is greatly increased by enclosing them in a high vacuum, as shown by P. LEBEDEW. H. BRANDES 75 has designed a very simple construction for this, shown in Fig. 107, while Fig. 106 shows a diagrammatic cross-section through two of the four wires.* A particularly good thermoelement is obtained by the combination of tellurium with constantan or tellurium with platinum (say a thin platinum wire sweated on to a small ball of tellurium) (L. W. AUSTIN). 75 c. An advantage of the thermoelement as compared to the bolometer is that no auxiliary cell (e, Fig. 103) and no equal- izing of the bridge are necessary. Both bolometer and thermoelement require only a very small amount of heat and hence only a very small amount of energy to produce a considerable deflection, particularly if a highly sensitive galvan- ometer is used, wherein lies their great advantage over hot-wire air thermometers or the commercial hot-wire instruments. For many purposes the very convenient direct-reading movable coil galvanom- eters are sufficient; however, if measurements neces- sitating the lowest possible energy consumption are to be made, a good mirror galvanometer, not too extremely damped, is more suitable. Calibration of these is best obtained with alter- nating current and an electrodynamic precision volt- meter without a multiplier. 49. The Thermogal variometer. There is one in- strument even more sensitive than either the ther- moelement or the bolometer of usual design, viz., the thermogalvanometer, constructed by H. DUD- DELL 76 following an arrangement of C. V. BOYS for measurements with high frequency oscillations. The principle is as follows: Between the poles N and S (Fig. 108) of a horseshoe magnet a movable wire frame L is suspended similarly to a movable coil galvanometer. A thermocouple (antimony-bismuth) is attached at the lower end of the suspended frame, giving a very high EMF. At one junction point a hot wire or thin strip of gold-leaf or strip of a platinum mirror on glass is attached, through which the oscillations are passed. This heats the strip and thereby also the junction point, producing an EMF and a current in the frame. The latter is thereby * Thermoelements and bolometers are disadvantageous when a vacuum is used, in that they cannot be repaired when they burn out, which occurs frequently as it is difficult to provide reliable fuses. When no vacuum is used, it is a simple matter to replace the burned wire. 76 WIRELESS TELEGRAPHY deflected just as in a movable coil galvanometer, a mirror and scale serving to measure the deflection. Fig. 109 illustrates a construction of such an instrument* said to be characterized not only by its sensitiveness, but also by its convenience. W. GEKLACH 77 has devised an arrangement similar to the thermogal- vanometer, having the highly sensitive thermocouple, which is acted upon by the hot-wire strip, connected to a separate sensitive galvanometer. 50. Comparison of the Sensitiveness of Various Measuring Instru- ments. 78 The following table gives the energy consumption at a deflec- FIG. 109. tion of 100 mm. or 100 scale divisions for various instruments, this serving as a measure of their sensitiveness. This, however, is not by any means a measure of their practical usefulness, which depends on quite other properties. 51. Measurement of Very Small Currents. 79 For the measurement of very small currents the various detectors discussed later (e.g., galena- * As made by the Cambridge Scientific Instrument Co. Figs. 108 and 109 are taken from a pamphlet issued by this company. THE HIGH FREQUENCY ALTERNATING-CURRENT CIRCUIT 77 i o r J2 a. G732 ^ 1 o o o o o O o o o o o o o pq | ^Og'l x X X X X X X X x X X X X X X X Is. ||j|i CM o CO CO CO <** CO o 8 8 co o d o o i o d d 2 d 10 o d il' 'O a is "S s a f 3 "g It It g It a .2 ft g a a Sa- il .2 |ft It a o 1ft a ^flection ,mp. a .2ft || sflection 6 amp. eflection " 6 amp. It Ja - If eu i Section 6 amp. :flection 6 amp. "i !; sl 73 ^ 73 * 03 03 0) 03 73^ 73^ ^3 co 73 "^ 73 73 73 ^0 "So So 73 73^ 73 03 l! O O 1 od p a o o !1 o ad Jd O iM O CO i* OOO lx sS ft O(N K ooo !^ X O>0 O(N 1J . iH aa O a " * W MO o .d a si ?s o co O 00 N co ^H T-H O 10 o O -2 G a O co Oi 3 -' 1 P4 S P^ 1 "" 1 PH ^ P4 " J3 o 03 3 a a a G ^^3 1 O a 0) o> H 3 cu 1 ?: a a '-3 1 a o -2 * 1 ^ a : E "o *p G 'o o a -2 I W3 cu cu 03 6 G oo' (^ ^ 'S " PQ h IM -2 03 s co' . 00* o ^J 'I 73 'S O le G G 00* 1 % 1 1 S 15 c3 .s c 03 a . o 8 ^ 03 o3 'S ^ ^ *bb 3 2 ft ft 1 3 G 73 G 73 r! 73 "3 "M a a a a PQ^M a "8 Tiometer. Cc cm. long. 1 S 60 G __0 l^ ^^ S G cu r l a B S 01 vacuum. Iro a 3 foege. Iron a acuum. 6 ment. Iron ai no vacuum. 7 a 3 ^anometer. G r . Platinum o a 3 G 03 s r. Platinum r. Platinum /. Platinum a a "ol S 1ft 11= .0 la. 73 c tn i* gco 1 S ~ls 2 I <+H ^* 0) . 1 13 03 13 13 a 13 PM g- W II |a r Ij PQ os 73 > XI ^ G la a^ > ^3 a | be a o 1 g bO o a |l ii Hot-wire air 0.02 mm. di || 'g8 11 w~ Bolometer oi 0.0025 mm. Bolometer w: mm. diam. 4 1 H g73 |a feq Is 2 The same wi Duddell-ther Duddell-ther Duddell-ther Duddell-ther Duddell-ther cu 73 s Duddell-ther '^73. i H ..' "1 78 WIRELESS TELEGRAPHY graphite [Art. 150], or red zinc oxide-copper pyrites [Art. 160], or the audion detector [Art. 161 c]) can be used to advantage. They are used in connection with a galvanometer (arranged for instance as shown in Fig. 375 in the circuit $ 2 C", by substituting the galvanometer for the telephone) or with a telephone (as in Fig. 375). In the first case the galvanometer deflection gives a direct measure of the alternating current* passed through the circuit, while in the second case there are two methods for arriving at the current value.* Either an adjustable resistance is connected in parallel to the telephone and varied until the sound heard in the telephone becomes just audible the smaller the resistance necessary, the greater is the alternating current measured (the so-called "parallel-resistance" method), or the unknown alternating current is caused to induce current through an adjustable coupling [Art. 54] in a circuit containing the detector and telephone the looser the coupling for a just disappearing sound, the greater is the measured current. Such devices must always be calibrated before being used, and even then are adapted for accurate measurements only if the detector can be relied upon for entirely constant action. Their great advantage, however, lies in that their sensitiveness is of quite another order 79 than that of the apparatus described in Arts. 47-49. * That is, the effective current value, when thermodetectors are used. CHAPTER IV COUPLED CIRCUITS 1. COUPLING IN GENERAL 62. Magnetic, Galvanic, Electric Coupling. Two electromagnetic systems (oscillators or closed current circuits) are said to be "coupled" if they are so arranged that oscillations in one of the systems always cause oscillations in the other. That system or circuit in which the energy is first supplied, say from an induction coil or similar source, is called the "primary circuit," the other being called the "secondary circuit." a. Magnetic or Inductive Coupling. In this case the mutual action of the two systems is procured only through their magnetic field : mutual induction* of the two circuits. Fig. 110 illustrates a case of this kind FIG. 110. for two condenser circuits; the bracket between the two coils SS is in- tended to indicate, here and in following diagrams, that the coils are mutually inductive. b. Galvanic or Conductive Coupling. In Fig. Ill, which shows a case of this kind, the parts drawn in heavy lines may be considered as con- stituting the primary circuit, while the fainter lines together with the coil S form the secondary system; the coil S is therefore common to both circuits. The arrangement of Fig. Ill may be conceived as having been developed from that of Fig. 110 by first winding the two coils S of Fig. 110 next to each other on a common core, as illustrated by the coils Si and $ 2 of Fig. 112, and finally superimposing them until they become a single winding. It is evident that in this case there is a magnetic coupling of the two circuits just as there is in Fig. 110 the * For the electromotive forces En and En induced in / by the secondary circuit and in // by the primary circuit, we have, as is well-known : En o = wL lo X /2 ; Ei- 2<> = o>L 2i X 7i 80 79 80 WIRELESS TELEGRAPHY current /i in the primary circuit produces a magnetic flux in the coil S, which flux in turn induces an EMF in the secondary circuit which also contains S. However to the magnetic coupling there is here added another kind of coupling. Even if S in Fig. Ill were a non-inductive (e.g., electrolytic) resistance and both circuits were so arranged that absolutely none of the magnetic lines of force of one could pass through the other (that is their mutual inductance were zero), there would nevertheless exist a coupling of the two circuits. The current in the primary circuit would cause a difference of potential between the ends of S which would in turn cause a current to flow in the secondary circuit. This kind of coupling is called "galvanic" or "conductive." Fig. Ill therefore illustrates a combination of magnetic and galvanic coupling, which is frequently referred to as "direct coupling." 81 FIG. 111. FIG. 112. In this case to the electromotive forces EH and E& produced in the primary and secondary circuits by their magnetic coupling there are added the electromotive forces E g i and E g2 caused by their galvanic coupling, the various values being: 81 E glo = Rh o ; E ilo = a slight reaction becomes noticeable. Consequently there is a change in the damping of the two oscillations. The decrement of the more weakly damped oscillation is increased and that of the more highly damped oscillation is reduced, that is the two decrements come nearer to each other in value. The following are the equations for the new decrements d 1 and d IT with primary circuits having no spark gap: -2 W f 1 z > \-dz ty V*X i *"* or, as long as K is small as compared to ZTT d 1 = di + ~ T- 2 1 d" - d, - /^- 2 1 3. CLOSE COUPLING OF TUNED, DAMPED OSCILLATING CIRCUITS 58. Form of the Oscillations. Assume two tuned oscillators, say two condenser circuits, which before coupling had the frequency N and wave- length X and the respective decrements di and dz, to be coupled. The coupling is not made loose, so that in any case -d 2 \ 2 t * For currents not quasi-stationary L S2]l should be replaced by L 2l in the equations for 7 and I max, which latter are to be understood as values at the current anti-nodes. t In the very unfavorable case of di = 0.08, d 2 =0.2, K must > 0.02 to meet this condition; for K > 0.1, K' becomes practically identical with K in all cases encountered in actual practice. 88 whence WIRELESS TELEGRAPHY Whenever the coupling is fairly close, K 2 is considerably greater than n -) ' so that the quantity K f is not much different from the coup- ling coefficient K. (See note f; P- 87.) Under these conditions there are in general 87 two distinct oscillations the so-called "coupling oscillations" ("coupling waves") produced in both the primary and the secondary circuit, having two distinct frequencies, N 1 and N 11 , and two distinct decrements, d 1 and d 11 . If, as heretofore, we use 7i and Vi to indicate current and voltage in the primary circuit, 7 2 and V 2 the same in the secondary then 7i (and Vi) as well as 7 2 (and V 2 ) are the results of two oscillations. Hence we may write : 7 = I T + I u \ y 1 = y T y jj for the primary circuit. I 2 Vz 4- I * ~r V 2 for the secondary circuit. The various oscillations have the following frequencies, wave-lengths and decrements. (and F 2 J ) 59. The Frequency of Coupling Waves. a. Primary Circuit without Spark Gap. Let the index 7 refer to the oscillation having the higher frequency and shorter wave-length. Then we have: n N " _ N" _ jl + K' '-" Vl^K'' and , n (2) Hence, the greater K f , i.e., the closer the coupling, the more will the frequencies (wave-lengths) of the coupling waves differ from each other and from the original common frequency (wave-length) . b. Primary Circuit with Spark Gap. In this case also the relations between the frequencies before and after coupling are of the form of equa- tions (1). It is not definitely known, however, though this is of no prac- COUPLED CIRCUITS 89 tical importance, whether the factor K' has the relation to the coefficient of coupling, K, and the decrements given by equation (1) of Art. 58. The quantity which actually determines the extent of the coupling and which may be directly measured by test [Art. 87] is the factor K' for cir- cuits with spark gap also. K' is called the "degree of coupling." Its value is frequently ex- pressed in percentage, thus: "3 per cent, coupling" means K r = 0.03. The relation between N 1 , N 11 and N, as well as X J , X IJ and X is given in Table X for different values of K'. c. The resultant oscillation produced from the two coupling oscilla- tions of different frequency is of the form shown diagrammatically in Fig. 130, and shown in Fig. 124 as ob- tained with an oscillograph (H. DIESSEL- HORST 88 ) and in Fig. 125 as photo- graphed from the spark discharges (H. RAU 88 ). The resultant oscillation may be conceived as having the frequency N and an amplitude which periodically increases and decreases, similarly to the beats or pulsations of a tone which are observed in acoustics. 1! fi ! II I'l FIG. 124. FIG. 125. The greater the difference between the frequencies of the two os- cillations, i.e., the closer the coupling, the greater is the number of pul- sations obtained per second. This number, S, which is the number of times per second that the amplitude passes through zero, is given by S = N 1 - N 11 = approx. NK' Hence the duration of one beat or pulsation is approximately =vr^; seconds = -^ periods. d. The energy relations, as is evident from Fig. 130, are as follows: Originally the entire energy resides in the primary circuit. After half of one pulsation the amplitude of the oscillation in the primary circuit is 90 WIRELESS TELEGRAPHY zero, while that in the secondary is a maximum and the entire energy has been transferred to the secondary circuit. After another half pulsation all the energy is again back in the primary and the secondary is at zero, etc., etc. In short, the energy continues to swing back and forth be- tween the primary and secondary circuits. 60. The Decrements of the Coupling Waves. a. Primary Circuits without Spark Gap (P. DRUDE 89 ). The relations of the decrements be- fore and after coupling are expressed by: + d 2 N^_ _ dd-f 2 ' N : 2 d 1 = ill d" = d, N J^ X' _V X" d 1 d 11 N 1 N n X' So that while for low degrees of coupling the decrements of the two oscillations are approximately equal to the average value of the decre- 0.25 0.20 0.15 0.10 0.05 =$=ZT Theor. d lj - Thcorr t 0.1 K' 0.2 0.3 FIG. 126. 0.4 0.5 ment before coupling, as the coupling becomes closer, the decrement of the oscillation having the shorter wave-length increases and that of the oscillation having the longer wave becomes less than the average value mentioned above. Theoretically the closest possible coupling exists when K' = 1; in practice about the highest value obtainable is approximately K' = 06. For this latter value, we have: N 1 = 1.6 N d 1 = 0.8 (di + d 2 ) = 0.8 N d 11 = 0.4 (dd + d 2 ) Hence in practice the frequency and decrement of the oscillation of shorter wave-length will at most be twice what they are for that of the longer wave. COUPLED CIRCUITS 91 b. Primary Circuit with Spark Gap. (C. FiscHER 90 ). In this case the relations of a do not hold. 1. The decrements of both oscillations, particularly if the coupling is loose, are greater than would follow from a. 2. It is by no means always the oscillation of shorter wave-length which is the most highly damped. On the contrary, this usually is more slightly damped than the oscillation of greater wave-length. The conditions obtained by coupling a condenser circuit containing a spark gap with another having no gap, were observed by C. FISCHER, whose results are shown in Figs. 126 and 127. Fig. 126 refers to the case of primary and secondary capacities being practically equal* while 0.25 0.20 0.15 0.10 0.05 in Fig. 127 f the capacity in the primary circuit is much greater than that in the secondary. 61. Amplitude and Phase of the Oscillations. 91 a. Amplitude.^ The current amplitudes of the individual oscillati'ons have the same re- lation, approximately, as their frequencies, i.e., III r 1n / N II _2 o _ II ~ II ~ II ~ I /, /2, A X * Ci = C 2 = 0.85 X 10- 3 MF. Li = L 2 = approx. 22,000 C.i, c. The relations given in a and 6 also hold if the frequency is varied in the primary circuit and the current effect is measured in the secondary. In that case C, N and X in the equations of a and b should be under- stood as the variable quantities of the primary circuit, while P eff repre- sents the current effect in the secondary. d. If the oscillations in the primary circuit are undamped the equations of a and b remain correct if we put di = 0; the decrement of the secondary circuit is then given directly. 75. Abnormal Forms of the Resonance Curves. In applying the expressions given in Art. 74 to different points of an experimentally determined resonance curve it may happen that the value of d\ + d^ will be different for different points, i.e., the form of the resonance curve is not that assumed in Art. 74. a. If the value determined at the various points fluctuates up and down irregularly, this is due to inaccurate observations in obtaining the resonance curve (irregular operation of the interrupter or spark gap). All that can be done in such a case is to take the average of the different * Theoretically xi should = x 2 . In practice, however, because of unavoidable inaccuracies in the measurements, x\ and Xz will always be slightly different. Hence it is' best to take x the mean value of xi and x 2 . t A zero method for determining the decrement has been developed on this principle by L. KANN. 109 RESONANCE CURVES 117 values found for di + d%; it is preferable, however, to redetermine the resonance curve. b. If the resonance curve is decidedly* unsymmetrical, as, for instance, curve b in Fig. 173, this indicates condenser leakage discharge; the sum of the decrements can then not be found at all from the resonance curve [Art. 86]. c. With condenser circuits having spark gaps, the following condition is often found : the resonance curve is symmetrical, but the values found for di + ^2 become systematically lower as we go down from the top of the curve (Fig. 148). An explanation! for this may be that the ampli- tude curve of the primary circuit is not, even approximately, an exponential curve. 110 In that case the decrement does not remain the same during an oscillation [Art. 9c]. If, in this case, as is usually done, the average of the values obtained at different points is taken, t then this f mean value is that value of di + d 2 which ^ would be obtained with a condenser circuit having an exponential decrease of the amplitude and giving the same resonance sharpness. This peculiarity is particularly noticeable in FlG 148 condenser circuits having very short spark gaps. With these, the value obtained for d\ and d% at nine-tenths of the height of the peak may be 50 per cent, greater than that obtained at one- third the total height. 111 This is probably due, at least in part, to the fact that the oscillations are abruptly cut off [Art. 9d]. At any rate, theory has shown (B. MAddi 112 ) that an abrupt cutting off of the oscillations may result in a deformation of this kind in the resonance curve. d. It may happen that the current effect retains relatively Targe values for a considerable distance to either side of the resonance point (Fig. 149). This may be due to the fact that the primary either directly or through some other circuit has an inductive effect upon the indicating circuit. Or the measuring circuit may have been coupled too closely with the primary circuit. The resonance curve will then be of the form shown in Fig. 150; the decrement values from such a curve will vary from point to point and be too high throughout. 113 e. If the resonance curve has two peaks, this may be taken as an * A slight dissymmetry appears when the factor A [Art. 556] is not constant, depending upon the frequency. f Furthermore the instrument deflections used as ordinates may not be exactly /.//. t Usually the value obtained at one-half the total height of the resonance curve is identical with this mean value. 118 WIRELESS TELEGRAPHY indication that there are two distinct oscillations in the primary circuit [Fig. 175]. Such curve forms, however [e.g., the heavy, full line in Fig. 151] may result in quite another way, namely, if spark or brush discharges pass between the coatings or plates of the variable condensers (or at FIG. 149. FIG. 150. other points of the measuring circuit). Thus in Fig. 151 the real reso- nance curve is shown by the broken line: the actually observed full-line curve having been caused by a reduction of the current effect at the middle portion of the curve due to spark discharges. 76. Determination of the Decrements of the Primary and Secondary Circuits. a. With a damped primary circuit, the resonance curve gives only the sum of the decrements of the primary and secondary circuits. To obtain the individual decrements, we may, for example, proceed as follows (method of V. BjERKNES 84 ). A known re- sistance R', is connected into the secondary cir- cuit,* which has been brought into resonance with the primary circuit (deflection in the indicating cir- cuit = otr). This will cause an increase in the secondary decrement dz by an amount A FIG. 151. (1) But the instrument deflection, d r , will be reduced to a value a. Then 1 -t - 1 (2) in which d is the value of di -f- d 2 obtained from the resonance curve. * In the measuring circuit of Fig. 141, the terminals AB are provided for this purpose. t Assuming that the deflections S 2 . The distance between S'i and S'z is chosen the same as that between Si and 2 . A circuit of the same dimensions and resistance as the indicating circuit is then constructed and joined to ' 2 . If this causes no change in the deflection of the instrument in the indicating circuit at resonance, * For the development of this method as applied to damped oscillations (see S. LowE 114 ). f If the measuring instrument used gives the effective value of the current, /,// K' and I' e ff respectively, then R = -j-^ le/f .. ' " 120 WIRELESS TELEGRAPHY then the indicating circuit has no effect upon the decrement of the measur- ing circuit. This condition can not be obtained when using hot-wire thermometers or commercial hot-wire meters, with which the decrement is largely dependent upon the indicating circuit and its degree of coupling to the measuring circuit. The values of d 2 obtained in calibration therefore hold only for those coils Si and S- 2 used in calibrating and only for the par- ticular position of those coils used during calibration. 77. Measurement of Small Changes in the Decrement. The method change in the current effect at resonance given in Art. 76, may also be used to advantage where small changes in the decrement (e.g., as caused by eddy currents) are in question. a. Equation (2), Art. 76, when d z is known, i.e., with calibrated meas- uring circuit, may be used to determine any change d f in the decrement, whether in the primary or in the secondary circuit. This gives the decrement change with much greater accuracy than by obtaining it from the resonance curve.* b. Sometimes, particularly for comparisons, it is very convenient to use the "equivalent resistance," instead of the change in the decrement, d' ', produced by some such cause as e.g., eddy currents. The equivalent resistance is that resistance, R', which would have to be connected into the circuit to increase the decrement by d'. This equivalent resistance,. R', can be calculated from equation (1), Art. 76, i.e., from the measured increase of the decrement, d', and the dimensions of the circuit. 114 If only R f is desired, it is much simpler to apply the method given above as a compensation method. For instance, assume a conductor brought near the secondary circuit has caused a reduction of the deflec- tion in the indicating circuit from a to a. To determine its equivalent resistance, the conductor is removed and resistance inserted in the secondary circuit until the deflection at resonance again falls from a r to a'.f The resistance having this effect is R'. 78. Measurements with Resonance Circuits in General. a. For the excitation or production of the oscillations in the primary circuit, four methods are available in the laboratory, viz. : 1. Undamped oscillations by the arc method. 2. Excitation by a quenched spark-gap circuit (Fig. 152). 3. Impulse excitation [d]. * Other things being equal, the accuracy of the determination increases as the sum of the decrements of the primary and secondary decreases; hence it is advisable to use a quenched spark-gap circuit for producing the oscillations [Art. 78c]. f The frequency may be influenced by the eddy currents in which case it must be brought back to its original value by suitable regulation. RESONANCE CURVES 121 4. Charging the condenser in the primary circuit and discharging through a spark gap. The requisite condition underlying the equations of Arts. 70, 74 and 76 are: 1. The primary oscillation must be of constant frequency. 2. The amplitude curve must be either a straight line parallel to the axis of abscissae (undamped oscillations) or an exponential curve. None of the methods of excitation given above strictly fulfills both re- quirements. The form of the resonance curve gives some indication of how nearly they are fulfilled [Art. 75]. Hence even when the measure- ments do not absolutely necessitate it, it is advisable to plot the resonance curve. To these requirements should be added another, of great practical importance, viz., constant amplitude and frequency of discharge. If these are not constant, the fluctuations of the instruments will prevent accurate measurements [also see g]. Quenched Gap Circuit Measuri FIG. 152. b. The presence of upper harmonic oscillations of higher frequency in addition to the fundamental, when using the arc method for undamped oscillations (Chap. IX 115 ) does not generally interfere if the secondary is a condenser circuit.* But if the frequency of the fundamental oscilla- tion is not entirely constant, its fluctuations will cause a widening of the resonance curve and application of the equations of Art. 74 will give too high a value for the decrement. The errors, however, which occur in this way with arcs especially intended for measurements [Art. 125], once skill is attained in their manipulation, are not large. But, what is most unsatisfactory, is not so much the average amount of the error as the un- certainty of its extent at any particular instant. Another disadvantage of the arc method is the difficulty of keeping the amplitude sufficiently constant. Otherwise, measurements, particularly of the decrement, are espe- cially convenient and simple with undamped oscillations. * With an open secondary circuit (antenna) this may cause disturbance if an upper harmonic of the secondary should happen to be in resonance with an upper harmonic of the primary circuit. 122 WIRELESS TELEGRAPHY c. If the primary oscillations are induced by means of a quenched spark-gap circuit, then from the moment the oscillations in the gap cir- cuit stop, those in the primary circuit fulfill the requirements given above, viz., the frequency is absolutely constant, the amplitude curve is an expo- nential curve.* Hence the resonance curves obtained by this method have the normal form as shown in Art. 74. Excitation by means of a quenched gap circuit^ moreover, has the great advantage that the oscillations in the primary circuit are much less damped than they would be if the primary circuit included a spark gap. In addition, a very high number of discharges per second may be used. This increases the current effect to such ah extent that the coupling between the primary and the measuring circuit and also between the latter and the indicating circuit, if one is used, may be made very loose. The regularity of the oscillations depends upon the kind of gap used. The mercury arc lamp and especially series gaps in hydrogen are well adapted for the purpose. Also the series spark gap of TELEFUNKEN with disc-shaped silver electrodes in air [Art. lllc], as well as PEUCKERT'S generator [Art. llle] can be used to advantage. 116 They are suitably operated by an alternating-current transformer (also an induction coil fed by alternating current) , or a resonance coil. d. The third method impulse excitation of the primary circuit [Art. 109d] is experimentally of extreme simplicity: a storage battery of a few cells and a suitable interrupter giving a high number of interruptions per second are all that is needed to produce the oscillations. To determine resonance in the secondary circuit a detector [Art. 51] with telephone or galvanometer is used, the simplest kind of detector, such as, e.g., is quickly made from a piece of galena and a point of graphite pressed lightly against the galena by a spring, suffices entirely. If the primary circuit is properly designed, its oscillations will be only very slightly damped. The high sensitiveness of the detector and the fact that the current effect (even though the oscillations have a small amplitude) is relatively large on account of the high number of discharges per second, permit extremely loose coupling between the primary and secondary circuits. Hence the various measurements based on the resonance principle [Art. 73] can be made with this method just as accurately as with any method using the current effect. For determinations of the decrement from the resonance curve, this method is suitable only if the detector works with great regularity and is calibrated. e. The fourth method, namely, charging the condenser in the primary circuit and allowing it to discharge through a spark gap [Art. 1], has the * The presence of two oscillations up to this moment will interfere less as the coupling to the quenched gap circuit is made closer, i.e., the faster the oscillations in the latter die out. 191 RESONANCE CURVES 123 advantage that the amplitude can be very easily varied over wide limits ; its disadvantage lies in the fact that the amplitude curve of the resultant oscillations does not approximate an exponential curve. Moreover the discontinuing of the oscillations in the primary circuit at a certain instant and the fact that the spark gap affects even the frequency [Art. 9] must be taken into consideration. Entirely aside from the questionable value of determinations of the decrement of the primary circuit by this method, it is very doubtful to what extent we may draw conclusions from the resonance curve under these circumstances as to the decrement of the secondary circuit. On the other hand, a great many determinations have shown that this method is entirely accurate for frequency measurements and gives at least approximately correct results for secondary decrements, on condition that such spark gaps as have been found to cause wide variations from the conditions existing in condenser circuits having no spark gap are avoided, i.e., spark gaps less than 5 mm. long and having copper or silver electrodes. Furthermore the following should be considered in regard to the primary circuit. The average value of d\, obtained from the resonance curve [Art. 75c], does not correctly characterize the decrease with time of the amplitude in the primary, but defines with sufficient accuracy the shape of the resonance curve and hence the sharpness of resonance and the maximum current effect attainable with the particular primary circuit in a loosely coupled secondary. But these are just the quantities on account of which the decrement is of practical interest. The decrease in the amplitude itself is only of minor importance in practice. To obtain the greatest possible regularity in the discharges only a low number of sparks per second should be used and partial spark discharges be avoided, assuming the use of metallic spark gaps in air. * Magnesium is the best electrode material in air; tin, zinc and aluminium are less suitable; copper and silver are especially bad for the purpose. The regularity of the sparks is affected by retardation of the discharge [Art/426], so that all means for reducing the retardation tend to increase the regularity. Subjecting the gap to ultra-violet light has already been mentioned in Art. 426 as one such means. Another method is to attach a fine point (say a pointed wire) to one electrode (Fig. 153). The position of the point can be so adjusted until the point discharge causes a very regular main discharge without materially changing the potential ampli- tude (W. EicKHOFF 118 ). If magnesium electrodes are used, however, this method usually need not be applied. An induction coil with D.C. supply * If a mercury arc lamp 117 or a metallic gap in hydrogen is used, many partial spark discharges may be used without resulting in irregularities; in this way the current effect can be greatly increased. However, the much higher damping of such spark gaps is the bad part of the bargain. 124 WIRELESS TELEGRAPHY and a mercury-turbine interrupter* or a large influence machine, but, best of all by far, a resonance coil with A.C. supply [Art. 114a] serve as suitable current sources. /. For the measuring circuit minimum damping offers the advantage that the frequency, due to the increased sharpness of resonance, as well as the decrement of any primary circuit can both be more accurately determined, other things being equal. If it is to be expected that the primary oscillations will be of relatively short duration (short spark gaps) then it would seem advisable to make the decrement of the measuring circuit about equal to that of the pri- mary circuit. 112 g. To the requirements of a should be added the very important one : the coupling between primary and secondary circuits must be extremely loose, i.e., so loose that there is no appreciable reaction. To Source of Current FIG. 153. FIG. 154. Whether or not that is the case, can be determined as follows. Con- struct a condenser circuit (///, Fig. 154) of about the same dimensions as the secondary circuit and coupled to the primary circuit about as closely as the secondary is. If this causes no change in the current effect in the secondary at resonance, it follows that no appreciable reaction exists between the secondary and primary circuits. 119 Commercial wave meters have sacrificed the fulfillment of these various requirements in order to make use of the convenient but rela- tively non-sensitive commercial measuring instruments. Hence the values for di + d% [Art. 74] and for d% [Art. 76] obtained by them are, in general, too large. The error may amount to even 30 per cent. 120 It can be decreased by increasing the current effect in the primary, as this allows a looser coupling to obtain a sufficient deflection in the indicating circuit. h. The wires leading to the current source (e.g., transformer) must be connected directly at the terminals of the spark gap (thus in Fig. 2, at * It is advisable to use a motor of somewhat larger capacity than that generally supplied by the manufacturers and to mount a fly-wheel on its axis. RESONANCE CURVES 125 the points FI and F 2 , and not at A and B) , as otherwise the damping of the primary circuit may be considerably increased. i. If a revolving coil, mirror galvanometer is used in the indicating circuit, it is usually advisable to ground the coil. Otherwise the coil may become charged and react similarly to the needle of a quadrant electrometer. 79. Commercial "Wave Meters". 121 a. Wave meters, which are simply commercial constructions of the measuring circuits described in Art. 71, are arranged for one or more of the following duties: 1. Determination of the natural frequency of any oscillator and through this, of the capacity, the self-inductance, the mutual inductance [Art. 73], and the degree of coupling [Art. 87]. 2. Determination of the decrement of any circuit. 3. Production of oscillations of any desired frequency (the wave meter used as primary circuit). These are all based on the resonance principle described in Arts. 70 and 74. Hence the essential part is a condenser circuit whose fre- quency can be continuously varied over a known range. For this purpose wave meters have either 1. A condenser of continuously variable capacity and one or more coils of fixed self-inductance (e.g., TELEFUNKEN Co., 122 MARCONI Co. 123 ), or 2. One or more condensers of fixed capacity and a coil of variable self-inductance (" variometer") (e.g., G. SEIBT [C. LoRENZ 124 ], IVES, DE FOREST), or finally 3. Both capacity and self-inductance are variable, in some cases the movable parts of the condenser and the inductive coil being linked so as to move in unison (e.g., J. A. FLEMING'S Kymometer, 125 PERi 126 ). The movable parts are usually provided with a pointer moving over a scale, which mostly permits a direct reading of the wave-length (or frequency) at each position of the pointer. 6. To measure the frequency (or wave-length) of an oscillator, e.g., a condenser circuit, the wave meter is set up near it and the wave meter's frequency varied until it is in resonance with the oscillator. Resonance is indicated either by 1. The lighting of a Geissler tube, or 2. The maximum deflection of a measuring instrument (e.g., hot- wire meter) connected directly into the measuring circuit or coupled to it, or 3. The maximum sound intensity in a telephone which is connected, together with a detector, in parallel to a portion of the measuring circuit or which is in a separate indicating circuit. Some wave meters have several of these arrangements provided at the same time. 126 WIRELESS TELEGRAPHY c. The decrement of an oscillator is rarely determined in practice by the resonance curve method of Art. 74a, it being customary to employ the simplified method given in Art. 746, using the measuring instrument mentioned in 62 [Art. 79]. Both methods give the sum of the decrements of the oscillator and the measuring circuit; the latter either being known or found as per Art. 76, the oscillator decrement follows. Another method, 123 which gives approximate values of the decrement without the use of an actual measuring instrument, only employing a detector and telephone, is that used in the MARCONI "Decremeter." It is based on the equation of Art. 74a2, N r - N Nr I Vfef The arrangement is shown diagrammatically in Fig. 155. The measuring circuit contains a variable condenser, C, a self-inductance, L, a small coil, L 3 , which can be either connected into the circuit or else short- circuited, and a coil, L 2 , having thirty-two turns of heavy wire. The coefficient of self-inductance of L 3 is so chosen that by the introduction of this coil the frequency of the measuring circuit is changed by 4 per cent., this change, as shown in Art. 3, being indepen- dent of the capacity of the cir- cuit. The detector, D, and the telephone, T, with the cell, E, are shunted across the coil / 2 from its end A to the sliding contact S. In order to determine the decrement of an oscillator the measuring circuit, without the coil La, is first brought into resonance with the oscillator (maximum sound intensity in the tele- phone). Then: 1. The measuring circuit is put out of resonance to the extent of 4 per cent, by inserting L 3 , and the tone in the telephone noted with the sliding contact at B, i.e., with thirty-two turns of L 2 in circuit. Let / 2 e // be the current effect in the measuring circuit under these conditions. 2. The coil L 3 is again cut out, bringing the measuring circuit into resonance again. The sliding contact is then displaced until the tone in the telephone is just as loud as it was in 1. Let n be the number of turns now between A and S t and let 7 r 2 e // be the current effect in the measuring FIG. 155. RESONANCE CURVES 127 circuit. Provision is made by suitable arrangements for quickly obtain- ing the conditions of 1 and 2. In the arrangement of Fig. 155, the current amplitude in the detec- tor during each period is proportional to the current amplitude in the measuring circuit and to the number of turns in parallel with the detec- tor. Hence the current effect in the detector must be proportional to the current effect in the measuring circuit and to the square of the number of parallel turns. Therefore, if the detector action depends upon the current effect, and if the tone in the telephone and therefore the current effect in the detector is the same in both cases 1 and 2, then we have FIG. 156. / J^lL\ _ / \ . Applying this to Art. 74a2, we obtain the sum of the \ 1 e ff I \n / decrements of the oscillator and the measuring circuit : X 0.04J (f > - ' d. In order to use a wave meter as a primary circuit, either 1. Small spark gaps are inserted in it so that oscillations may be pro- duced by means of an induction coil, or 2. Means are provided for producing oscillations by impulse excita- tion [Art. 109]. e. Only a few of the many types of commercial wave meters can be described here. Probably that of J. ZENNECK 127 was the first used in 128 WIRELESS TELEGRAPHY FIG. 157. FIG. 158. RESONANCE CURVES 129 wireless telegraphy. It consisted of a condenser of fixed capacity and continuously variable self-inductance. A Geissler tube or a spark gap served for frequency measurements by means of resonance, while a bolometer was used for decrement determinations. The next step in this direction is represented by the FRANKE-DONITZ (Telefunken) wave meter shown in Fig. 156; it consisted of a variable condenser, interchangeable coils for the different ranges and a hot-wire- air thermometer. Very similar in design and equally simple is the port- FIG. 159. able wave meter of the MARCONI Co., of which Fig. 157 shows the con- nections diagrammatically, Fig. 158 the finished construction; it consists of a variable condenser, a fixed self-inductance of rectangular shape mounted into the cover of the case and a carborundum detector [Ait. 160] with telephone. The later wave meter of the TELEFUNKEN Co. (Fig. 159) whose adjust- able condenser, in addition to its graduated scale, also has three scales of wave-lengths corresponding to the different coils, while more compli- cated, has a much wider range of usefulness. The same applies to the portable decremeter of the MARCONI Co. (Fig. 160). 130 WIRELESS TELEGRAPHY f. The direct-reading wave meter of R. HiRSCH*' 128 is based upon a very neat application of the resonance principle. FIG. 160. a C 1 1 / FIG. 161. FIG. 162. * Manufactured by DR. E. HUTH, G. M. B. H., Berlin, to whom the author is indebted for the cuts. RESONANCE CURVES 131 FIG. 163. FIG. 164. 132 WIRELESS TELEGRAPHY The measuring circuit (Figs. 161 and 162) consists of a fixed self-induc- tance, L, and a variable condenser having one fixed, C, and one movable, C 1 , set of plates, the latter being rotated by a motor. This also rotates a small helium tube, A, over a scale, B, the tube being connected in parallel with the condenser. The rotation of the movable element of the con- denser causes a continuous variation in the frequency of the measuring circuit. At that position at which the measuring circuit is in resonance with the oscillator under observation, the tube becomes illuminated and a bright line is seen on the scale at the point where the helium tube is at the instant of resonance. By indicating along the scale the wave-lengths of the measuring circuit corresponding to each position of the rotating element, the instrument becomes direct-reading Two forms of this wave meter are shown in Figs. 163 and 164. 2. RESONANCE CURVE OF THE DYNAMOMETER EFFECT (L. MANDELSTAM AND N. PAPALEXi 129 ) 80. General. a. Assume a movable coil, $ 2 , in a vertical plane, e.g., suspended on a vertical wire, placed within a fixed coil, Si, also in a ver- tical plane. If a current /i is passed through Si and /2 through $2, the turning moment to which the /|\ movable coil is subjected ^Iil^. If /i and 7 2 vary rapidly with time, as e.g., in high frequency alternating currents, the coil will in general not respond to the rapid variations and its motion will be determined by the average FIG. 165. turning moment, i.e., the average value of /i/g. This average value is called the "dynamometer effect," IiI 2 130 , from the use of this arrange- ment of a movable coil in the field of a fixed coil in the well-known dynamometer type of wattmeters. This arrangement also always makes it possible to measure IiI 2 . b. Assume now, as in Art. 70, that a primary circuit of constant fre- quency (and wave-length) acts inductively upon a secondary circuit of variable wave-length, e.g., an adjustable condenser circuit. Let I\ and 7 2 represent the currents in the primary and secondary circuits re- spectively. The dynamometer effect of the two currents is measured and a curve plotted in which the abscissae are the wave-lengths (or capacities) of the variable secondary circuit, the ordinates being the corresponding dynamometer effects. The resulting curve will be of the form shown in Fig. 165; as may be shown theoretically 129 this curve passes through the axis of abscissae when RESONANCE CURVES 133 the wave-length of the secondary circuit is equal to that of the primary, i.e., when the two circuits are in resonance. c. The form of this resonance curve, similarly to the current effect curve [Art. 70c] depends on: 1. The sum of the decrements of the primary and secondary circuits. 2. The degree of coupling between the two circuits. FIG. 166. FIG. 167. The effect of the size of the decrement is shown by curves / and // of Fig. 166* and the effect of the percentage of coupling is shown by curves I and II of Fig. 167. | In these curves the abscissae are the dissonance values, \ \ r r A.2 I\T -t / ^2 ~"~~ ^ T /2 ~cT x = X, I X' X" FIG. 168. d. If the coupling between primary and secondary circuits is extremely loose, we have the following relations (Fig. 168) : 1. Let xi and x 2 be the dissonance values at which the dynamometer effect has its maximum positive and negative values respectively. Then = 27T * / :di = 0.05, d 2 = 0.01; II :di = d 2 = 0.01. t di = 0.05, d z = 0.01; 7 : coupling extremely loose; II :K' = 0.3 per cent. 134 WIRELESS TELEGRAPHY 2. If a line is drawn parallel to the axis of abscissae and intersecting the resonance curve at the points whose abscissae are x' and x", then 81. Determination of the Frequency (Wave-length). a. A method for determining the frequency (wave-length) of a primary circuit follows directly from Art. 806. The primary is caused to act inductively upon a measuring circuit through an extremely loose coupling and the dynamome- ter effect 7 1/2 of the primary current l\ and the measuring current 7 2 is measured. The frequency of the measuring circuit is varied until the dynamometer effect becomes zero. That frequency (wave-length) of the measuring circuit at which this occurs is the desired frequency of the primary circuit. b. Instead of leading the primary and measuring currents directly through the dynamometer, it is more convenient to have both circuits act Dynamometer FIG. 169. inductively, through as loose a coupling as possible, upon two coils, Si and $ 2 (Fig. 169) which are connected to the dynamometer. It can be shown 129 that the dynamometer effect /'i/' 2 of the currents induced in these coils follows practically the same changes as /i/ 2 - c. Wave-length (or frequency) determination by means of the dyna- mometer effect has the following advantages over the determination by means of the current effect [Art. 71]. 1. It is much more accurate. In the current effect method we work around the peak of the resonance curve and slight variations in the frequency cause but very slight (percentage) changes in the deflection. Hence, in order to obtain the exact point of resonance it becomes prac- tically essential to plot the resonance curve or at least its upper part. The dynamometer determination, on the other hand, is a zero method. The slightest deviation from resonance produces a noticeable deflection in RESONANCE CURVES 135 the measuring instrument. The dynamometer method is therefore to be used wherever small changes in the frequency (or in the capacity, dielec- tric constant or coefficient of self-induction [Art. 73]) are to be measured ; in accuracy it surpasses by far all other methods. 2. The accuracy of frequency determinations by means of the current effect method, depends upon the accuracy with which the resonance curve can be obtained, i.e., upon the regularity of the discharges per second and the amplitude. The dynamometer method is independent of both these factors. 82. Decrement Determination. This is based upon the relations described in Art. 8Qd. As in the case of the current effect method, the sum of the primary and secondary decrements is obtained. The connections are those shown in Fig. 169, with the stipulation that the coupling between the primary and measuring circuits must be extremely loose. a. To find the sum of the decrements by Art. SOdl, the wave-length, X (capacity, C) of the measuring circuit is varied until the dynamometer effect is a maximum either on the positive (X = Xi, C = Ci) or the nega- tive (X = X 2 , C = C 2 ) side. Then, if X r and C r are the respective values of X and C at resonance, i.e., when the dynamometer effect is zero, we have: , 7 Xi X r _. X r X2 Xi X2 d\ + d 2 = 2w r - = 2ir r -- = approx. IT Xi X2 Xr Ci -Cr C, C, 7T Cl - C 2 With this method it is not necessary to determine the entire resonance curve. However, for most purposes the method is sufficiently accurate, as it is relatively easy to sharply locate a maximum and as the absolute value of the deflection at the maximum has no bearing upon the results. 6. If great accuracy is of importance, the method based on Art. 80d2 should be employed. The resonance curve is plotted with either the wave-lengths or capacities of the measuring circuit as abscissae. Then a line is drawn parallel to the axis of abscissae and intersecting the curve at two points whose abscissae are X' and X" or C' and C" respectively. Then if \ r and C r are the resonance values, we have : C" -C r C" -C r 83. The Dynamometer. Modified forms of the ordinary dynamome- ters may be used for measuring the dynamometer effect, a fixed and a movable coil, the latter suspended on a bronze strip and provided with a 136 WIRELESS TELEGRAPHY mirror similarly to the coils of a Deprez-d'Arsonval mirror type instru- ment. Both coils must have only a small number of turns. 131 A "short-circuit loop dyna- mometer" (MANDELSTAM and PAPA- LExi 129 ) has given very good results. a. Its construction, for labora- tory purposes is shown in Fig. 170, its diagrammatic connections in Fig. 171. There are two flat coils, Si and 82, perpendicular to each other and between the two, but coaxial with $ 2 an aluminium loop or ring with a small mirror is sus- pended on a fine thread. The two currents I'i and 7' 2 of Fig. 169 are sent through the coils Si and $2 in order to measure their dynamome- ter effect. The resulting action is as follows: The current 7' 2 , sent through $ 2 induces a current 7 3 , in the loop, which is in phase with 7' 2 and proportional in amplitude to I f 2, on condition that the inductance of the loop is greatly in excess of its resistance. 132 The current in /S 2 causes no turning force to act on the loop, as $2 and the loop are coaxial. But the current I'\ pass- ing through Si induces no current in the loop as their planes are perpen- dicular to each other; it does, however, produce a torque upon the loop proportional to the dynamometer effect I\I 3 and hence also ex JVV This is true accurately only at the zero position of the loop. Careful analysis of the conditions, however, 129 has shown that even when the loop has been turned from its position of rest through a small angle, its de- flection oc I'J'i. With the arrangement of Fig. 169, the deflection is FIG. 170. a . FIG. 171. in which b is the torsion moment of the suspension system, f is the number of discharges_per second and a and c are the constants of the apparatus. Hence # oc F RESONANCE CURVES 137 b. If the torsion moment of the suspension system is made very small (quartz thread), the factor b in the equation of paragraph a becomes very small in comparison to f . c7' 2 i e //. Then we have i.e., 5 becomes independent of the discharge frequency, f, and thereby independent of the more or less irregular operation of the interrupter, if an induction coil or interrupted direct current is used. 3. USE OF RESONANCE IN THE STUDY OF CONDENSERS 84. Determination of the Frequency Factor. The following will serve as a simple arrangement. Construct a primary circuit (condenser circuit /, Fig. 172), having that frequency at which the frequency factor of the FIG. 172. condenser is to be determined. Connect the test condenser, (7, into a circuit containing a variable self-inductance, S, by means of which this condenser circuit, 77, is brought into resonance with the primary circuit, 7. Then replace C by a calibrated adjustable air condenser and vary its capacity until circuit 77 is again in resonance with the primary. Then the capacity, C, of the test condenser is equal to that of the air condenser at resonance. Now find the capacity, C 8 , of the test condenser for static charges (see foot-note, Art. 72o). Then [Art. 5a], the frequency factor for the frequency in question is f .c_ 3 " c, This method is easily modified in numerous ways to adapt itself to any given case. The only essential feature is the replacing of the unknown capacity by that of an air condenser (which is independent of the fre- quency [Art. 5a]) and maintaining resonance. 138 WIRELESS TELEGRAPHY In applying this method it is important to choose the coefficient of self- induction of circuit II so that the frequency will not be appreciably affected by the currents in the condenser coatings or by such slight changes in the leads to the condenser as may be necessary in view of the different construction of the air condenser and the various test condensers. If it is desired to compare the frequency factors of a number of con- densers having different dielectrics, the same electric field strength in the dielectric should be used in each case, as this may affect the frequency factor. 85. Energy Absorbed by Dielectric Hysteresis. 21 a. The same arrangement as that of Fig. 172 applies. Assume the secondary circuit, //, which includes the test condenser, to be in resonance with the primary circuit and that the deflection of the measuring instrument in the indi- cating circuit is a'. The condenser, C, is then replaced by a variable air condenser which is adjusted until resonance is again obtained. Let a be the instrument deflection which now results. Then from these values and equation (2) Art. 76, we obtain the increase, d', in the decrement of the secondary circuit caused by the energy absorption in the condenser C and which characterizes the energy absorption of the particular dielectric material [Art. 13]. b. For comparing various condensers it may be simpler to determine their equivalent resistance [Art. 776] by substitution. For this purpose, after having replaced C (Fig. 172) by an air condenser and readjusted for resonance, sufficient resistance is connected into the secondary until the instrument deflection is again a'. This resistance, R', is the equivalent resistance of the condenser. 133 c. In applying this method, which may be modified in various ways, it is especially important to avoid eddy currents in the condenser coatings. Their effect can entirely destroy the accuracy of the results. In con- densers in the form of Ley.den jars it is quite difficult to avoid eddy currents, or even to determine whether the eddy currents have been eliminated. A convenient safeguard, applicable only to plate condensers, however, is to place the condensers in various positions or to use first zinc sheet electrodes and then copper electrodes. If this causes no change in the instrument deflection it may generally be concluded that the existing eddy currents are negligible. d. In comparing the energy absorption of different materials it is also important to use the same electric field intensity throughout in the dielec- tric, as this may affect the result. Similarly only values obtained at the same frequency should be compared. 86. The Brush Discharge of Condensers (W. EiCKHOFF 134 ). a. Curve a in Fig. 173 is the resonance curve of a condenser circuit whose condensers have no brush discharge; curve b was obtained with the same RESONANCE CURVES 139 circuit but with a heavy brush discharge from the condensers [Art. 14a]. The difference between the two curves is twofold, viz., 1. b is not symmetrical, falling off much more rapidly on the side of the higher frequencies, while curve a is symmetrical. 2. The resonance point (maximum current effect) in b occurs at a lower frequency (greater wave-length) than in a. Both these points are characteristic of condensers with brush discharge. b. The explanation of this phenomenon is to be found in the following: The brush discharge, by charging the uncoated portion of the con- denser, causes an increase in its capacity and a decrease in the natural fre- quency of the circuit. The effect, however, is not the same as when another second condenser is joined in parallel to the coatings of the first through a metallic connection, for the conducting path be- tween the coated and un- coated portions really consists of the brush discharge itself, which jumps from point to point very irregularly. Hence the amount of the charge held on the uncoated portion of the condenser is also con- tinuously fluctuating. The irregularity of this parasitic FIG. 173. capacity and its connection to the coated condenser will result in a varying frequency whose maxi- mum value is determined by the capacity of the coated portion.* Hence, if such a condenser circuit is caused to act upon a resonance circuit and if the frequency of the latter is gradually decreased, the cur- rent effect will rise with relative rapidity as soon as the maximum fre- quency just mentioned is approached. It will, however, retain compara- tively great values as long as the frequency of the resonance circuit remains in the range of the frequency fluctuations caused by the brush discharge in the primary circuit. The result therefore is a widening of the resonance curve in the direction of the lower frequencies. c. The widening of the resonance curve indicates a considerable reduc- tion in the resonance sharpness;^ it is caused mainly by the fluctuations in the frequency, as was shown in the preceding paragraphs. Hence the amount of the energy loss due to the discharge can not be * In all probability, these fluctuations in the frequency are accompanied by varia- tions in the initial amplitude and irregularities in the fall of the amplitude during each oscillation. f The resonance sharpness is about 24 for curve a and about 10.5 for curve 6 in Fig. 173. 140 WIRELESS TELEGRAPHY determined [Art. 78a] from the resonance curves which give only an upper limit for the loss. If the value di + d% is obtained from the resonance curves by applying the relations of Art. 74, there being no brush discharge on the condensers, and the value d'i + d 2 is obtained with a brush discharge on the con- densers, other conditions being the same, then the increase in the decre- ment due to the brush discharge can not be more than (d\ + d 2 ) (di + &) = d\ - d,. However, the resonance curves may also be used for obtaining a quantitative value of the effect of the brush dis- charge on the sharpness of resonance. The equations of Art. 74 are applied to the curve and di + d% is deter- mined. Subtracting from this the decrement of the measur- ing circuit, d 2 , we have d\, which may be considered as being the decrement of a condenser circuit having no brush discharge but having the same resonance sharp- ness. * The result of measurements made in this way (W. EiCKHOFF 134 ) is shown in Fig. 174 for Ley den jars of German flint glass, f The three full-line curves show the relation of di to the potential amplitude, first with the capacity consisting of a single Ley den jar,{ then with 2X2 Leyden jarsj connected as in Fig. 12, and lastly with 3X3 jars,J con- nected as in Fig. 13, the outer circuit remaining the same for each case. In the first case the entire potential difference exists between the con- * The actual reduction in amplitude caused by the brush discharge does not come into question. f M. WiEN 17 has found the following apparent increase in the decrement due to brush discharge: I 08 - $_ 1 , ( 1 f,^ _i fl |: 02 g / / ^ , / Wf / f ^ '' sVr' 0.1 - "1"~ *** *= = = r? '** 5 "" o . ,_ Single Jar under Oi "H 10 s Volt-* 10 20 SO 40 50 60 70 Spark Length 35 8 12 15 20 SO 34 in mm FIG. 174. Potential amplitude 0.9 X 10 4 volts 1.55 X 10 4 volts 2.2 X 10 4 volts Leyden jars of English flint glass 0.008 0.028 0.064 Jars made by H. Boas. 0.002 0.002 0.007 J All the jars had practically the same capacity, so that the resultant capacity in the various combinations remained just about the same. RESONANCE CURVES 141 denser coatings; in the second case only one-half; in the last case only one-third. The brush discharge is accordingly greatest in the first and least in the third case. The uppermost curve shows how very detrimental the effect of brush discharge may be to the resonance sharpness. A comparison of the three curves shows that this harmful effect may be combatted by series-parallel combinations of the condensers [Art. 4d], As a matter of fact, such series-parallel combinations of equal con- densers can produce the desired result only if the apparent increase in the decrement due to brush discharge with increasing potential varies more rapidly than TV. If it c Fo 2 , a simple consideration will make it evident that nothing is gained (L. W. AUSTIN 21 ) by series-parallel combinations. Above what potential the apparent decrement increase rises more rapidly than F 2 depends upon the form and material of the condensers. d. As these series-parallel connections involve considerable compli- cation, it is desirable to overcome brush discharge in some simpler way. This may be accomplished by placing the condensers, or at least the edges of their coatings, in a heavy oil. The extent to which this can reduce the detrimental effect of brush discharge is shown by the dotted curve in Fig. 174. However, this is a dangerous method. For, if the voltages are not comparatively low,* the condensers are almost certain to break down. Bad as a brush discharge may be it has one good feature about it, namely, a certain protection against breaking down of the condensers. e. From the preceding, we may draw certain conclusions of practical importance, viz., 1. Displacement of the point of resonance [a2], other things being equal, ^increases together with the potential amplitude, in fact is pro- portional to it under the conditions applying to such investigations as have been made. Hence, as brush discharge can not be entirely over- come in primary circuits, the tuning between primary and secondary must be done at the same potential at which the circuits will be used later. 2. The influence of brush discharge in cylindrical condensers becomes less according as the diameter is made smaller in comparison to the length, other things remaining equal, as this makes the parasitic capacity smaller in proportion to the normal capacity. Hence, from this point of view it is better to use long, narrow than short, wide jars. 3. Thickening the uncoated end of the condensers (Leyden jars) also reduces the parasitic capacity and the effect of the brush discharge (see Art. 396). * With the best flint glass, 5 mm. in thickness, 30,000 volts (corresponding to 1 cm. gap) is the extreme limit. 142 WIRELESS TELEGRAPHY 4. THE USE OF RESONANCE CURVES FOR INVESTIGATING COUPLED CIRCUITS (J. ZENNECK/ C. FISCHER, 90 M. WiEN 90 ) 87. Coupling of Tuned Circuits. Determination of Frequency, Dec- rement and Degree of Coupling. If the oscillations of coupled, tuned circuits [Art. 55, et seq.] are caused to act upon a measuring circuit, reso- nance curves of the form shown in Fig. 175 will be obtained, if the cir- cuits are quite closely coupled. The relations of Arts. 71 and 74 may be applied to both of the parts of these curves. The location of the two peaks gives the values of the frequencies N 1 and N 11 (and the wave-lengths X 1 and X 77 ) of the two oscillations, the form of the curve around the two peaks gives the decrements d 1 and d 11 and the degree of coupling [Art. 95] is 2.2 2.4 2.6 2.8 3 3.2 3.4Xl0 6 /Sec Frequency of the Measuring Circuit FIG. 175. 1 - K f = 1 + in which N and X are the frequency and wave-length respectively of both circuits before coupling.* If the coupling is not very close this may be simplified into X"-X' ~~ Table X gives the values of K' for different ratios of the frequencies. With loose coupling, however, the resonance curves assume the shape * If C 1 , C 77 and C represent the capacities of the measuring circuit corresponding to the wave-lengths X 7 X 77 and X, then K' = ^ ~ RESONANCE CURVES 143 of the full-line curve in Fig. 176. The two peaks do not, in general, occur at the points corresponding to the two frequencies. Hence N 1 and N 11 (as well asX 1 andX 11 ) cannot be determined from the location of the peaks, nor can the decrements be found by applying the methods of Art. 74. In this case we must proceed as follows. 136 a. The method in this case is based on the fact stated in Art. 616 that of the oscillations of the same frequency 7 1 1 j are approximately in Ii 11 1 are displaced approxi- 7 2 7 j phase. 7/ J j mately 180. This relation makes it possible to practically eliminate the effect of one of the pairs upon the measuring circuit, subjecting the latter only to the other pair. Tl ww uuuuu L F FIG. 177. 2.2 2.2b N 2.32 Frequency of the Measuring Circuit FIG. 176. b. The arrangement is shown diagrammatically in Fig. 177. Small wire loops, KI and K 2 , are connected into the primary and secondary circuits respectively, and similar loops, MI and M 2 , are joined to the measuring circuit (777) . KI acts inductively only upon MI, K 2 only upon M 2 . The phase relations of the electromotive forces, E, are the same as for the corresponding currents, and we have approximately Ei 1 ] E H ' ^// [ in phase, ^ u \ displaced 180. We will assume that these rela- tions instead of being only approximate are exact. The amplitude of these electromotive forces, aside from depending on the currents 7/, 7 2 7 , etc., also depend on the distances between KiMi and KzM z . If these distances are adjusted until the amplitudes of Ei 1 and E 2 H are equal, then Ei 1 and E 2 n will neutralize each other. 144 WIRELESS TELEGRAPHY The result is that oscillation II (frequency N 77 ) has absolutely no effect upon the measuring circuit, which acts as if only the oscillation of fre- quency N 1 and decrement d 1 existed. Hence if the resonance curve is plotted in this way, it will represent only this one oscillation, and N 1 or X 7 and d 1 can be obtained from it in the usual way. To obtain the opposite effect, that is, obviate oscillation / so that only oscillation II will be measured, all that is necessary is to revolve the loop MI (or else MZ) through 180 and then proceed just as before; EI and EZ now neutralize each other while Ei 1 and E% n are added to each other. c. The method of procedure therefore is as follows. First a resonance curve is plotted, having, in general, two maxima. Then the distance between MI and KI (or M% and Kz) is varied until only one maximum remains in the resonance curve, all indications of the second peak having disappeared; the curve is then the resonance curve of one oscillation. Then M i is turned through 180. If a trace of the former maximum remains, it should be eliminated by a final adjustment of the distance between MI and KI (or MZ and Kz). The curve then is the resonance curve of the second oscillation. The dash and dot-and-dash curves of Fig. 176 were obtained- in this way. They are the resonance curves of the two oscillations; from them may be obtained the frequencies N 1 and N 11 ) the decrements d 1 and d 11 and the degree of coupling K' '. d. In the practical application, the loops KI and KZ may be entirely omitted, the primary and secondary circuits being used in their normal form to act upon M i and MZ. The latter, however, i.e., loops M i and MZ, are best retained, as they simplify the manipulation. The following points should also be noted: 1. Moving MI (or MZ) must not change the self-induction of the measuring circuit. This is provided for by placing the leads connecting these loops to the rest of the circuit of wires very close together. 2. For precise measurements it is important to entirely eliminate the effect of the oscillation other than the one whose resonance curve is being determined. This can be done as follows: Assume oscillation // is to be eliminated, after an approximate value of N 11 has been obtained as described above. The measuring circuit is adjusted to have this fre- quency N 11 . Then the loops MI and KI are adjusted with respect to their relative position, until the electromotive forces E-f 1 and Ez 11 are added and the current effect in the measuring circuit becomes as great as possible. Only then is MI turned through 180; this procedure will result in a much more complete elimination of oscillation II than before. * *The necessity for this precaution is due largely to the fact that the deflections of the measuring instruments which can be used for this purpose depend upon the mean square of the current value. Even when the effect of oscillation II is not RESONANCE CURVES 145 e. The accuracy of this method is very high so far as determinations of frequency and degree of coupling are concerned. But in view of the assumption made in b not being strictly correct, the values of the decre- ments d 1 and d 11 found in this way may involve considerable errors, the extent of which can hardly be fixed in each case (B. MACKU 138 ). 88. Close Coupling of Tuned Circuits. Current Effect in a Third Circuit. 90 Consider a secondary circuit tuned to and closely coupled to its primary and at the same time very loosely coupled to a third (measuring) circuit. Two questions present themselves, viz., 1. How does the total current effect in the third circuit depend upon the latter's frequency? 2. If the third circuit is synto- nized with one' of the oscillations of II \a 0.5 K 0.4 FIG. 178. FIG. 179. the secondary circuit, how does the current effect in the third circuit de- pend on the coupling between the primary and secondary circuits? a. The answer to the first question follows directly from Art. 87. The heavy full-line curve of Fig. 176 shows how the current effect in the third circuit depends on the natural frequency of this circuit. Comparison with the dash and dot-and-dash lines (the resonance curves of the individual primary and secondary oscillations) shows that the maximum current effect in the third circuit does not occur when it has the same frequency as one of the oscillations in the secondary circuit. The maximum for the slower oscillation occurs at a somewhat lower frequency, that of the more rapid oscillation at a higher frequency. The curves of Fig. 176 represent quite a loose coupling (K r = 0.028). sufficient to show signs of a second maximum in the lower portion of the resonance curve of oscillation 7, it may nevertheless greatly influence the shape of the upper part of the curve. 10 146 WIRELESS TELEGRAPHY The closer the coupling becomes the more nearly do the maxima of the current effect in the third circuit coincide with the frequencies N 1 and N 11 of the oscillations in the secondary circuit. 6. The curves in Figs. 178, 179 and 180* show the relation of the cur- rent effect in the third circuit to the percentage of coupling, first with the third circuit tuned to the more rapid oscillation (7), then to the slower oscillation (77). These curves show: 1. In all cases there is a very decided maximum current effect for both oscillations, always occurring at a relatively very low frequency. The less the damping of the secondary and tertiary circuits, the more decided is the maximum. 2. Up to this maximum there is no notice- able difference between the more rapid (7) and b FIG. 180. 0.1 0.2 0.3 >- FIG. 181. the slower (77) oscillation. But as the coupling is increased beyond the maximum, the oscillation of the higher frequency (7) may produce a considerably greater current effect than the slower oscillation. The curves of Figs. 178, 179 and 180 were all obtained with primary circuits having a spark gap. The conclusion (2) just drawn from the curves would in fact not apply otherwise. If the primary circuit has no * Fig. 178: Ci = C 2 = 0.85 X W~ 3 MF. L, = L 2 = 22,000 cm.; di = 0.11 a: d 2 = 0.14 6: d 2 = 0.20 d 3 = 0.10 Figs. 179 and 180: Ci = 5.29 X 10~ 3 M^. C 2 = 0.45 X W~*MF. Fig. 179a: d* =0.034 d 3 = 0.031 Fig. 180c: d 2 = 0.21 d 3 = 0.20 Fig. 181: The letters correspond to the same conditions as in Figs. 179 and 180. Length of primary spark gap about 6 mm. d 3 = 0.20 = 6,230 cm. \ , _ n IK = 73,000 cm. / * Fig. 1796: d 2 =0.10 d 3 = 0.10 Fig. 180d: d 2 = 0.37 d 3 = 0.31 RESONANCE CURVES 147 spark gap, theory 90 shows that the current effect must be the same for both oscillations. c. It may at times be interesting to compare the current effect in a third circuit tuned to one of the coupling oscillations with the total current effect in the secondary circuit. The latter's variation with the coupling is shown in Fig. 181 for the same circuits referred to by Figs. 178, 179 and 180. The use of very short spark gaps (less than 1 mm.) alters these condi- tions materially. In this case, as the percentage of coupling is gradually increased, the current effect in the secondary circuit may pass through a succession of maxima and minima. Careful investigation has shown that the maxima are due to particularly thorough quenching, the minima to particularly poor quenching in the gap of the primary circuit (H. RlEGGER). 7 With thorough quenching in the primary gap, the current effect in the secondary may, under certain conditions, be greatest when the primary quenched gap circuit is slightly out of resonance with the secondary circuit. 139 This, however, is by no means universally the case; in many arrangements bringing the circuits out of resonance does not in the least increase the current effect obtained at precise resonance in the secondary. 89. Coupling Untuned Circuits. Current Effect in a Third Circuit (M. WiEN 90 ' 92 ). Conditions are somewhat altered if the primary and secondary circuits have slightly different frequencies before being coupled, that is, are slightly out of resonance. a. Theoretical investigation has led to the following conclusions for circuits without spark gaps. If the decrements di and d 2 of the primary and secondary circuits are different before coupling, it is possible, by bringing the two circuits out of resonance, to obtain a current effect in one of the two oscillations which is greater than when the primary and secondary are exactly in tune. 1. If did 2 , what has just been stated for the more rapid oscillation, holds for the slower and vice versa. The increase in the current effect is greatest for a certain dissonance. This amount of dissonance, other things being equal, increases as the difference between di and d 2 increases and as the coupling is made closer. In general only about 20 per cent, increase in the current effect is the most that can be obtained. 6. With a spark gap in the primary circuit the relations, so far as can be concluded from such investigations as have been made to date, are qualitatively the same. But in general the strengthening of the current 148 WIRELESS TELEGRAPHY effect is not quite so great as for primary circuits without a spark gap and under certain conditions the more rapid oscillation (7) seems to be the most favored. 90. Investigation of the Quenching Action in Spark Gaps. The reso- nance curves of the current effect given in 1 are especially well adapted for studying the action of quenched gap circuits. For this purpose the secondary circuit coupled to the quenched gap circuit, is in turn coupled very loosely with a measuring circuit (Fig. 152) and the resonance curve of the current effect is plotted in this way. a. If the resulting curve has the form of curve b in Fig. 182, the coup- ling oscillations are present and there is no quenching action. If, however, the result is like curve a of Fig. 182, this indicates complete quenching and only the natural oscillations of the secondary cir- cuit are present. A resonance curve formed like curve C of Fig. 182 shows that in addition to the natural oscillations of the secondary, the coupling oscilla- tions are also present. This may be due to any of three causes. Either the coup- ling oscillations occur at one discharge, and quenching occurs at another. Or the oscillations in the secondary circuit are always of the same kind, but the quenching action is not complete, " impure," i.e., the primary oscillations are not quenched until more than half an oscillation is completed [Art. 64a]. These two cases can be distinguished by coupling the secondary very loosely with a resonance circuit tuned to the natural frequency of the secondary circuit. A small spark gap is connected in parallel with the condenser in the resonance circuit .and adjusted so as to respond regu- larly. This small gap is then placed alongside of the main quenched gap in the primary and both are observed in a rotating mirror. If the quenched spark gap is seen in the mirror first alone and then together with the small gap and so on, we evidently have the first case (H. RlEGGER 140 ). The third possibility is the existence of a thorough quenching action, but a very loose coupling. Then the duration of half an oscillation and hence the time during which two coupling oscillations exist together [see foot-note to Art. 78c] is so great that the latter become apparent in the resonance curve. O FIG. 182. RESONANCE CURVES 149 b. From Art. 64 it follows that such observations as have just been described in a, offer a direct means of answering the question of the critical degree of coupling so important in practice, and thereby also the question of which of two spark gaps has the better quenching action. The coupling is made closer and closer; the critical and therefore the best degree of coupling is that at which the coupling oscillations have not yet appeared but are just about to become noticeable in the resonance curve. It was stated in Art. 646 that, under certain conditions, several critical degrees of coupling, at which complete quenching is obtained, may be found. In such cases a comparison of various gaps as to their quench- ing action is very difficult. c. The resonance curve also offers a simple means of determining whether a given method of increasing the quenching action (e.g., air blowers, magnetic blow-outs, the use of hydrogen instead of air, etc.), is really effective 141 ). First a condition of impure quenching, in which the coupling oscillations are evident in the resonance curve in addition to the natural oscillation of the secondary circuit (curve c .of Fig. 182) is inten- tionally obtained. If then application of the method to be tested causes the indications of the coupling oscillations to disappear from the reso- nance curve, this is proof of an improved quenching action. CHAPTER VI THE ANTENNA 91. General. Just as in ordinary wire telegraphy, so in radio- telegraphy, a system of communication is essentially comprised of two stations, viz., the "transmitting" and the "receiving" station. Similarly, the collection of apparatus used for sending off telegrams is called the "transmitter" or "transmitting set," while the corresponding apparatus at the receiving station is termed the "receiver" or "receiving set." Every radio station has an open oscillator, the "antenna," that part of the antenna which is suspended in the air being called the "aerial." The transmitter induces electromagnetic oscillations in the aerial, whence electromagnetic waves are radiated in all directions; upon reaching the antenna of the receiving station these waves again produce oscillations in it, thereby causing the receiving apparatus connected to it, to respond. If a dot of the Morse Code is to be telegraphed, the electromagnetic waves are sent out for only a very short instant, while if a dash is to be telegraphed, the waves are sent out for a somewhat longer period. 1. THE VARIOUS KINDS OF ANTENNAE 92. Form of the Aerials. a. The simplest form of antenna consists of a vertical wire suspended from an insulator: "simple antenna." This is nothing more nor less than a straight lineal oscillator. These simple antennae are no longer used except in special cases, as, e.g., with portable stations, on airships [Art. 96] and aeroplanes and where balloons or kites are used to carry the wire, thereby allowing the use of great lengths. The successful use of the streams of water as simple antennae (R. A. FESSENDEN 142 ), the stream being maintained by a pump, is mentioned mainly as a curiosity. Such antennae, while very disadvantageous because of their high ohmic resistance* may nevertheless be useful in special cases of emergency (e.g., in a fort or on a battleship whose normal antenna has been destroyed by the enemy's fire). 6. The use of a large number of nearly vertical wires results in such * A transmitter having worked 480 km. with a wire antenna 40 m. high, worked over a distance of 160 km. with a stream of water of about the same height as the wire antenna. 150 THE ANTENNA 151 forms as the "harp" or "fan" aerial of Fig. 183* and the conical or pyramidal form of Fig. 184. f Fig. 185 shows a cross-section of a double cone or double pyramid antenna. FIG. 1"83. FIG. 184. * The Italian Battleship "Carlo Alberto," with which Marconi made long distance tests in 1902 (see p. 117, ZAMMARCHi 1 ). The large Eiffel Tower Station has a harp- shaped aerial, stretched from the top of the tower. Of course this may also be con- sidered merely as a sector of an umbrella antenna. t The early Poldu Station of Marconi for long distances. This antenna was long in use for transmitting telegrams to vessels plying between Europe and America (see p. 105, ZAMMARCHi 1 ). 152 WIRELESS TELEGRAPHY c. Antennae having very great capacity at their upper end 1 * 3 '* are now widely used, especially the so-called "umbrella antenna." In its simplest form this consists of a vertical wire or bundle of wires, from the upper end of which wires radiate downward in all directions, sometimes ex- tending quite near to the ground. The form used by the TELEFUNKEN Co. 144 in 1910 for the construc- tion of the Nauen Station is shown diagrammatically in Fig. 186. A tower FIG. 185. 100 m. high, terminating below in a carefully insulated ball, serves partly to support the entire antenna and partly as a current carrier in conjunction with a bundle of wires with which it is connected. In 1911 this tower was increased to a height of 200 m. (Fig. 187) the form of the second antenna being shown in Fig. 188. In April, 1912, during a severe storm this tower collapsed. Shortly after the construction of an entire new antenna and towers was undertaken. FIG. 186. A similar antenna is that of the NATIONAL ELECTRIC SIGNALLING Co.'s high-power station at Brant Rock 147 (height 130 m.) Its um- brella consists of eight cage-like wire structures 91 m. long and 1.2 m. in diam. Another similar antenna is that of the high-power station at Eberswalde, Germany (C. LORENZ). * Probably first used by O. LODGE and A. MuiRHEAD 145 ) [Translator's Note]. The "flat top" antenna, such as that of the U. S. Naval Station at Radio (Arlington) also belongs to this class. THE ANTENNA 153 However, umbrella antennae are also often used for portable sta- tions. Many ingenious collapsible masts 148 of light weight have been devised for these, so as to be easily carried on pack-animals, wagons, etc., and requiring only a few minutes for erection and taking down. FIG. 187. .d. Antennae consisting of vertical risers which are then prolonged horizontally at the top, usually as several parallel wires (Fig. 189: so- called "F" or more often, "L"-antenna; Fig. 190: so-called "^''-antenna), 154 WIRELESS TELEGRAPHY 18 Masts ., FIG. 189. FIG. 190. THE ANTENNA 155 should also really be classified, within certain limits, among the antennae having large end capacity. They are especially adapted for use on board ships, as the horizontal portion can be conveniently stretched between the masts. Battleships are now often equipped with the form shown diagrammatically in Fig. 190$. A number of other forms, which may be considered as combina- tions of two of the forms already described, have also been proposed or actually used. FIG. 190a. 93. Comparison of the Different Forms of Aerials. a. It is evident that the effective capacity is greater in all of the various complex antennae described than in a simple single wire antenna of the same height. It is greater according as: 1. The distance between (spacing of) the wires is greater in the vicinity of the potential anti-node and as 2. The distance from the wires to the ground is less at this part. Both these factors give the umbrella antenna its very great capacity in comparison to other forms.* 6. The natural frequency of the complex forms described, in view of their greater effective capacity, is much lower, the natural wave- length therefore much greater than for a simple antenna of the same height. c. As to the current distribution, there is usually a current anti-node at the base of the antenna. Thus Fig. 44 shows the current distribution for a simple antenna oscillating at its natural frequency; a current node is at the tip, and the current distribution is sinusoidal. If the wave- length of the oscillation is materially increased by inserting a coil (self- induction) near the base, only the upper and nearly straight portion of the sine curve remains [Art. 3 la]. For conical and harp- or fan-shaped antennae, the current distribution curve is not sinusoidal, but generally similar to the heavy broken line curve in Fig. 191. * e.g., the antenna of the Nauen Station, Fig. 186, had an effective capacity of about 0.018 MF. 156 WIRELESS TELEGRAPHY A characteristic property of the umbrella type of antenna is the fact that the current, while flowing up through the vertical risers, flows down through the inclined radial wires. The current distribution in the vertical part is about as shown by the heavy broken line curve on the FIG. 191. right side of Fig. 192, while in the inclined wires it is in general similar to the curve on the left side of Fig. 192;* in fact these forms remain practi- cally unchanged, whether the oscillations are at the natural frequency or at a reduced frequency (increased wave-length) due to added coils. FIG. 192. d. Comparing the various antennae as regards distance effect, the height being the same for all, the factors which enter for consideration are [Art. 25]: * The latter constructed by the method of Art. 25d. THE ANTENNA 157 1. The frequency (wave-length) of the oscillation. 2. The current amplitude at the current anti-node.* 3. The current distribution, hence the form factor. Low frequency (great wave-length) is unfavorable for good radiation, but favors the propagation of the waves through space [Art. 139/]. A large effective capacity, giving a large current amplitude at the current anti-node* and a current distribution favorable to good distance effect is advantageous. In the umbrella antenna, however, the portion which is useful for producing distance effect is relatively small, as the currents in the vertical and the inclined portions tend to neutralize each other's effect, leaving only the shaded area shown in Fig. 192. 2. GROUNDING 94. Ground and Counterpoise. Effect upon the Current Distribution. If an aerial, say a simple, single wire antenna, were left with a free lower end, it would have a current node at the lowest point. This would make it quite difficult, to say the least, to produce strong oscillations in the antenna by charging it or by coupling it to a primary circuit, f Moreover, the conditions would in general be unfavorable. Two methods for avoiding this are in use, viz., 1. Direct ground connection and 2. "Counterpoise," i.e., a wire network, connected to the lower end of the antenna, parallel to, but insulated from the ground. 149 a. The result of a ground connection as explained in Art. 33, is the formation of an anti-node of current at the base of the antenna, if the ground is highly conductive. This is what virtually occurs with the waves of wireless telegraphy when sea water or very moist ground exists at the base of the antenna for a considerable distance around it. This is by no means true, however, if the station is erected upon very dry, e.g., sandy ground or upon non-conducting rocks, { ground water being absent or existing only at a great depth. In such cases the anti-node of current will occur higher than the base, the height increasing as the conductivity of the ground circuit decreases. 6. The effect of the counterpoise upon the current distribution is not materially different from that of a direct "ground," no matter what the nature of the soil. We must distinguish between three cases : * At a given potential amplitude the same coupling with the same primary circuit. t The spark gap [Art. 102a] or the primary circuit [Art. 536] would have to be placed at a considerable height above the ground. t One need only consider the fact that marble and slate are used in commercial light and power circuits as good insulating materials for potentials of several hundred volts. 158 WIRELESS TELEGRAPHY 1. The ground is a very good conductor. 2. The uppermost portion of the ground is a very poor conductor, but underneath this at a slight depth there is conductive ground water. 3. The ground is a very poor conductor and either there is no ground water at all or only at very great depth. We are justified in assuming that a condenser of considerable capacity is formed in the first case by the counterpoise and the surface of the earth facing it and in the second case by the counterpoise and the surface of the ground water facing it. The insertion of such a capacity, however, either does not change the current distribution at all or it may raise the current anti-node somewhat [Art. 30]. The third case is identical with that discussed in Art. 29 : an insulated conductor of great capacity is connected to the end of the antenna. c. The practical method of grounding is particularly simple in the case of ships, which, in nearly all instances which come into question, are con- structed of metal, so that connection to any portion of the ship's body usually suffices. For land stations, there are two methods mostly in use, viz., 1. A metallic plate or cylinder (having a surf ace of , say, a few square yards) is buried in the ground or ground water; or 2. A very large wire network, circular or square shaped is laid either on top of or in the ground.* The counterpoise, for both stationary and portable stations, is usually provided in the form of square or circular wire networks or radial wires in the shape of a star, fastened to poles a few feet (say 2 or 3) over the ground and insulated from it. For portable sets a rectangular strip of wire net- work, which can be alternately rolled up and unrolled and fastened to poles, is often used as a counterpoise. 95. Energy Consumed by the Earth Currents. 150 The electric field surrounding an antenna exists not only in the air, but also partly in the ground. Hence there must be currents in the ground which dissipate energy. It is impossible to generalize as to the extent of this energy consumption, which depends largely upon the form of the antenna, the frequency of the oscillations and the nature of the soil. The main points which come into consideration are about as follows : a. Figs. 193-199f illustrate a number of cases diagrammatically, it being assumed throughout that there is a current anti-node at the base of * The old counterpoise of the Nauen Station was a circular wire network 400 m. in diam., placed under ground at a depth of 0.25 m. f These figures are not based upon any exact calculations, but are drawn from a general consideration of each set of conditions; therefore no reliance should be placed upon their precision. H. TRUE ISO has investigated the course of the ground currents in several cases. If the ground has extremely low conductivity, the electric lines of force may be inclined at an angle to the surface of the earth for a short distance just above the ground [Art. 139e]. THE ANTENNA 159 FIG. 193. the antenna. Comparing the use of a conducting network as " ground" and as " counterpoise," i.e., Figs. 195 and 197 with Figs. 196 and 198, we find that it makes no difference qualitatively* so far as the course of the current lines is concerned, whether the network is buried in the earth as a " ground" or used as a counterpoise. Both cases, however, are dis- tinctly different from the case of grounding by means of a metal plate (Figs. 193 and 194). In the latter case all the lines of force pass from the antenna into the earth, each producing cur- rents in the earth, while, where a network of conductors is used, a large number of the lines of force between the antenna and the net- work pass entirely or almost en- tirely through the air, thereby dis- sipating no energy, f Only those lines of force which reach the ground outside of the conducting network, produce currents in the earth. Hence, in this respect such networks are far superior to the use of relatively small metallic plates as grounds, the advantage increasing as the area of the net- work is increased. From this point of view the umbrella antennae (Fig. 199) are particularly advantageous. If the network is made large enough to extend considerably beyond the ends of the inclined radial wires of the umbrella, practically all the lines of force will pass * However, there are quantitative differences. t That is if we neglect the heat de- veloped by the currents in the coun- terpoise, or in the case of a grounded network (Fig. 195) by the currents between the network and the surface of the earth. FIG. 194. FIG. 195. 160 WIRELESS TELEGRAPHY from the antenna to the network FIG. 196. FIG. 197. N "N \ \ \ X \ \ \ ~~~x \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ I \ \ ] 1 \ \ \ \ FIG. 198. without first passing through the. earth. b. In Figs. 193, 195 and 197 it is assumed that the soil is homo- geneous and has very low con- ductivity,* while Figs. 194, 196 and 198 are based on the assump- tion that a layer of ground water of relatively high conductivity is present at a short distance below the surface, underneath an upper stratum of very low conductivity. The difference lies mainly in that the lines of force choose the greater part of their path through the conducting layer of ground water, only a relatively short por- tion of their path being in the upper non-conducting layer. When grounding by means of a metallic plate, it is of great im- portance that the plate is placed at a sufficient depth to really reach the ground water (Fig. 194.) c. It is also important that the lines of force are not crowded into a very narrow space at any point of their path, as this always in- volves a relatively great dissipa- tion of energy. For instance if an antenna is grounded through a single vertical wire, the current field, as seen from above, would be of the form shown in Fig. 200. If now, as is frequently done, the ground wire is replaced by a metallic plate, the current field assumes a far more advantageous form, about as shown in Fig. 201. If the wire network is laid on the ground so as to be in (conduc- * That is, ground water absent or present only at great depth. THE ANTENNA 161 tive) contact with it, the lines of force follow approximately the course shown in the cross-section drawn in Fig. 202 A i. If portions of the net- work do not make intimate contact with the ground, but are very close to it, the path of the currents is not materially altered (See Fig. 202 A 2 ). Fig. 203 shows the approximate course of the electric lines of force and FIG. 199. the currents, when a counterpoise is used instead of a direct ground. It certainly is the more advantageous method ; the flow into the ground oc- curs just as if the network were replaced by a sheet of metal which is conductively connected to the earth at all points. There is no crowding of the current lines anywhere. \ \ / \ f ^ \ \ / /' A 1 1 ( ^-- :> > } \ ^^^. \ \ s / \\ / \ / \ FIG. 200. FIG. 201. It is essential, however, that the conducting network which forms the counterpoise is really insulated from the ground. Faulty insulation at any point may come under either of two classifications. If the faulty insulation still offers a very high resistance (e.g., a damp porcelain insu- 11 162 WIRELESS TELEGRAPHY lator), the general conditions will be affected but very slightly, although of course, there will be an additional loss of energy in the high resistance. But if the resistance is very low where the fault occurs (e.g., a spark dis- charge to ground) a very considerable portion of the current may pass to ground at this point, under very unfavorable conditions, similar to those shown in Fig. 200. d. The conductivity of the soil plays an important part in determining the course and density of the ground currents as well as the energy they Ai I / V I / v Az I | / * \ I / V y n FIG. 202. dissipate. In general, for a given form of antenna and a given frequency, there exists a critical value of the conductivity, at which the energy loss is a maximum. For any other conductivity, be it greater or lower than the critical value, the energy dissipation will be less. A change in the conductivity of the ground as, for example, may be caused by varying weather conditions is apt at times to result in a change in the course of the earth currents, in the damping and possibly even in the frequency of the oscillations. The earth, therefore, introduces a variable factor into the entire system, no matter whether we use a direct \ / i FIG. 203. ground or a counterpoise. Only where the earth possesses very high conductivity (sea water, very wet soil) does the effect of the weather become negligible. e. If we let R e represent the equivalent resistance, of such value that Relief f is the energy consumed per second by the earth currents, I being the current amplitude at the base of the antenna, then it follows directly from a and b, that this ground resistance, R e , must depend not only upon the nature of the soil and the method of grounding, but also upon all THE ANTENNA 163 those factors which determine the electrical field in the earth, and hence particularly upon the antenna form and the frequency of the oscillations. As regards the determination of R e see Art. lOOd. Tests with an umbrella antenna have shown that R e increases to- gether with the frequency of the oscillations, but decreases as the height above ground of the counterpoise (when the latter is used) is increased. In all cases R e was lower when using a counterpoise than when a direct ground of the form shown in Fig. 193 was used (H. TRUE 150 ). 96. Ungrounded Antennae for Airships. 151 a. The following forms of antennae have been used among others, for airships, where grounding of any kind is entirely out of the question : The antenna is a wire suspended from the car (or basket), which latter with its metal parts (motors, etc.), serves as counterpoise, insulated from the bal- loon body (the bag).. 2. Similar to 1, except that the counterpoise includes the metal ribs or frame of the balloon (as in the Zep- pelin airships) or a conducting sheath of the balloon in addition to the car. 3. The antenna consists of two wires* of unequal lengths, somewhat on the order of LECHER'S arrangement (Fig. 204) (H. BEGGEROW): e.g., one wire (a i &i) is made equal to one-fourth of the wave-length of the oscillations, the other (a 2 D) = % wave-length. The oscillations are produced at A in the car; nodes of potential will then occur at A and C (see Arts. 72c2 and 24a). So far as radiation is concerned, only the part BD, which forms a simple lineal antenna, is effective, as the portions ai&i and a 2 6 2 neutralize each other. 4. For directive antennae, adaptations of Fig. 416 have been suggested; the horizontal part is stretched out underneath the dirigible, parallel to its axis, while the two vertical antennae are suspended downward. 6. With balloons there is ever present the danger of sparks between any parts of the balloon having considerable potential differences caused by the oscillations in the antenna when transmitting, f Such differences of potential may be the result of various causes, thus : 1. If any part of the airship is conductively connected to the antenna, the oscillations of the latter are spread out over the entire metal structure of the airship. This results in differences of potential between individual * Which are kept at the same distance apart throughout by insulating spacers. f Normally, of course, the oscillations during reception are entirely harmless. 164 WIRELESS TELEGRAPHY parts of the airship. To avoid these, it is advisable to join all neighboring metallic parts of the airship by the shortest possible connecting leads or bonds. 2. The electric field of the oscillations may produce differences of potential (" influence" action) between individual parts of the airship. This danger is greatly lessened by keeping the anti-node of current far from the airship. 3. The magnetic field of the oscillations may induce currents in the metal parts of the airship. In this connection a node of potential or anti-node of current in the antenna is particularly dangerous 152 . In short there are so many possible tendencies for the production of a spark, that it is probably impossible to state in advance that any particu- lar arrangement is spark proof. On the other hand, none of the arrange- ments described in A need be feared as placing an airship in any con- siderable danger. In general, it may be stated that short wave-lengths are usually more dangerous than long ones, and that the danger diminishes with decreasing current and potential amplitudes. Accordingly arrangements involving a comparatively small amount of oscillating energy but having a high discharge frequency are advantageous when compared to those of equal total energy but using larger amplitudes (energy) for each oscillation at lower frequencies. 153 c. There is of course also the danger of gas, which has escaped from the balloon, becoming ignited by the sparks in the gap of the primary circuit. It is obvious that only completely enclosed gaps [e.g., see Art. Ill] should be used. 3. THE OSCILLATIONS OF ANTENNA 97. Frequency, Capacity and Self -induction. 154 a. To measure the natural frequency of an antenna, cause a loop or coil of wire inserted in the antenna to act inductively upon a measuring circuit, then pro- ceed according to any of the methods already described [Art. 71]. A small spark gap in the antenna or preferably a quenched gap circuit or impulse excitation [see Art. 78] serves to produce the oscillations. Or a primary circuit having a known and variable frequency, is loosely coupled to the antenna and its frequency adjusted until a measuring instrument in the antenna gives the maximum deflection. 6. Frequency measurements may also serve for determining the effective capacity, C, and the effective coefficient of self-induction, L, of the antenna [Art. 27a], for example, as follows: 1. A coil of known self-induction, L , is inserted at the current anti- node. This will change the frequency N to N f , the wave-length X to X'. Then N , 2 X2 L = L O Z _ , 2 = Lo /8 _ 2 approx. (1) THE ANTENNA 165 Applying this value of L to the equation [Art. 27a] N " = or X = 2irV L VLC (2) the value of C is obtained. . 2. A condenser of known capacity, C OJ is inserted at the current anti- node. Then if the new values of the frequency and wave-length are N" and X" respectively, we have, JV"2 N 2 X 2 X" 2 C = Co -- ^2 -- = Co /T^ approx. (3) Again applying the value of C to equation (2) we obtain L. It is advisable to apply both methods 1 and 2, and use the average of the two values obtained for L and C. The greater the difference between N and N' or N and N" the less will be the danger of inaccurate re- sults, while if these differences are small, N 9 N f and N" must be deter- mined with great precision.* c. Another method (C. FISHER), which however is neither so con- venient nor so accurate, 155 consists in the insertion of a resistance R at the anti-node of current in the antenna and measuring the decrement before (di) and after (dz) inserting R. Then we have for the difference of the two decrements d = d 2 - di = irfl-Jjr [Art - 27a l Combining this with equation (2) which gives the product C X L, we obtain both C and L. 98. Regarding the Effect of Coils and Condensers in Antennae. a. The insertion of coils (inductance) lowers the frequency, hence increases the wave-length, f decreases the form factor and, with a given potential * The following are typical antenna capacities mentioned in the literature of wireless telegraphy. Torpedo boat antenna C approx. 1 X 10~ 3 MF. Cruiser, battleship antenna C = approx. 2 X 10~ 3 MF. Brant Rock high-power station C = approx. 7 X 10~ 3 MF. Nauen high-power station C = approx. 18 X 10~ 3 MF. Eiffel Tower high-power station C = approx. 7. 3X 10~ 3 MF. The self-induction used in antennae is usually much greater than that of the aerial proper due to inductive coils in the antenna circuit. The self-induction of the Brant Rock aerial is given as 55,000 C.G.S. units, that of the Eiffel Tower as 196,000 C.G.S. units. fThis is often expressed as "lengthening the antenna" (aerial) by means of a coil and "shortening" it by a condenser. 166 WIRELESS TELEGRAPHY amplitude, decreases the current amplitude [see Fig. 47, Art. 31]. All these effects tend to reduce the radiation and also the radiation decrement. The insertion of a condenser at the base of an antenna has the oppo- site effect, in so far as it increases the frequency, hence decreases or short- ens the wave-length* and at the same time the anti-node of current is raised upward from the base of the antenna [Art. 30]. The form factor is thereby made more favorable for distance effect. As to the change in the current amplitude with respect to the poten- tial amplitude and as to the resultant change in the distance effect and radiation decrement, it is hardly possible to draw any conclusions to cover all the various forms of antennae. By inserting both a coil and condenser in series, these can be so chosen for any given aerial as to avoid any change in the wave-length, only greatly reducing the radiation decrement ("Antenna with reduced radiation damping" [see Art. 326]. b. Instead of using just coils, the wave- length of an antenna can be greatly increased by means of the arrangement shown in Fig. 205. f A coil, L, whose self-induction is very great as compared to that of the aerial and the connection to ground is inserted in series with the antenna and the condenser, C, is joined in parallel to it. We are justified in considering this arrangement, as used in practice, as form- ing a condenser circuit whose self-induction is practically that of the coil L and whose capac- ity consists of the condenser C in parallel with the capacity formed by the aerial and ground. A little consideration will make it evident that such an arrangement has a materially lower radiation decrement than the antenna alone. c. These arrangements have found practical application as follows: 1. In order to make the advantage of long waves for propagation available, it is customary to use coils of considerable self-induction in antennae, either alone or in conjunction with condensers (" Lengthen- ing Coils") [see, e.g., the coil marked 28 in Fig. 236]. 2. Coils of adjustable self-induction at times together with condensers, or condensers of adjustable capacity alone are universally used for tuning * This is often expressed as " lengthening the antenna" (aerial) by means of a coil and "shortening" it by a condenser. t This is sometimes called the "fly-wheel" method. FIG. 205. THE ANTENNA 167 antennae to a desired wave-length ("tuning coils," "tuning condensers," "aerial variometers"). 3. To obtain different wave-lengths with the same antenna, a conden- ser, at times with a coil in series, is so connected that it may be cut into (short waves) or out of (long waves) the antenna by means of a switch. Or a switch is arranged by means of which the condenser is connected in series with the aerial for short waves (Fig. 206) and in paral- lel to the coil (Fig. 205) for long waves. A 99. The Damping of Antennae and Its Causes. a. Only that portion of the energy which, during the oscillations of an antenna, is sent out in the form of electromagnetic waves, may be considered as useful energy. If then we wish to speak of the "efficiency" of an antenna, meaning thereby the relation of the useful energy to the total energy supplied, at the fundamental oscillation, this would be i.e., the ratio of the radiation decrement to the total decre- ment.* b. All other losses of energy which occur during the oscillation, are more or less necessary evils. These include : 1. Joulean heat in the antenna. 2. Joulean heat of the earth currents. 3. Joulean heat of the induced currents. 4. Losses due to brush (leakage) discharge. 5. Circuit losses. The development of heat (Joulean) in the wires of the aerial, in the tuning and lengthening coils, in the ground cir- cuit, in the counterpoise and in the various leads, has a considerable effect upon the decrement of such antennae whose radiation decrement has been much reduced. Hence for well-designed antennae it is customary to use braids of very fine, individually insulated wires, or bands or strips consisting of several such braids in parallel and interwoven, in order to reduce the ohmic resist- ance to a minimum, f * COUNT v. ARCO 157 estimates the efficiency of a properly constructed ship antenna at 50 per cent., if the wave-length is increased by the factor 1.3 by means of inserted coils. t COUNT v. ARCO 160 gives the following data: 2 kw. station: effective current at base of antenna, 13 amp.; antenna resistance, 6 ohms; 480 single wires in parallel. 8 kw. ship station: effective current at base of antenna, 35-40 amp.; antenna resistance 3 ohms; 3000 wires in parallel. To provide the necessary tensile strength copper-sheathed steel wires [Art. 36c] and also bronze wires are often used. 168 WIRELESS TELEGRAPHY The portion of the total decrement due to the earth currents may at times be as large as the radiation decrement. Even in spite of the greatest precautions in grounding, in the attempt to keep this portion of the decrement at a minimum the results will depend ultimately upon the nature of the soil. Such results as can be obtained at sea are probably never attained over poorly conducting ground. 153 Induced currents come mainly into question in guys, stays, iron masts and similar metal parts on board ships, and in the towers supporting the antennae and their guys in land stations. Experience has shown that these currents, which always mean a waste of energy, may harm the radiation considerably and be generally detrimental. A method of coun- teracting the bad effect of these currents is to insert insulating links in the conductors affected, or, in any case, insulating them from ground. This was very well provided for in the old Nauen antenna [Art. 92c]; the only conducting parts in which currents could be induced were the three guys holding the tower and these were well insulated from the tower at their upper ends and from the ground below. 159 It is well known that the brush or leakage discharge, which at night is visible over a large part of the antenna, has a very bad effect upon the * ?> a FIG. 207. decrement; it is therefore important to avoid sharp points and edges in the aerial. As increased surface (larger radius of curvature) for the conductors tends to reduce the brush discharge, it has been proposed to surround the antenna wires by metal piping or tubing joined conductively to the wires (Fig. 207), or else to use metal bands or strips, preferably having rounded edges, wound around rope, as the aerial conductors. The use of well-insulated high-tension cable instead of bare wire is perhaps even more effective. Specially designed insulators 1 to prevent brush discharge are frequently used at the ends of the wires. Circuit losses* may of course occur in any oscillator such as, e.g., a condenser circuit. They have not been previously discussed for the rea- son that they are easily prevented in all other forms of oscillators and hence are of no importance when ordinary precautions are taken. With antennae as used in radio-telegraphy, however, the prevention of circuit losses, in view of the high potentials involved and the severe weather effects 156 is a much more difficult matter. 100. Determination of the Decrement. a. Any of the methods already given may be applied to find the total decrement of the 'natural oscillations; a quenched gap circuit offers a suitable means for excitation. * This is intended to include losses due to spark discharges (to ground, etc.). THE ANTENNA 169 The value of the total decrement for various forms of antennae under normal conditions (good grounding, thorough insulation), there being no coils of great self-induction inserted, runs about as follows: Simple antenna (single straight wire, airship antenna) . . 25-0 . 3 Harp- or fan-shaped antenna 0.2 Conical or double-cone antenna . 16-0 . 18 Umbrella or ship (T) antenna . 12-0 . 16 As a matter of fact, inductive coils are always inserted. If their coefficient of self-induction is not sufficiently large to materially affect the frequency of the oscillations, the decrement, for umbrella and ^-aerials will be about 0.1. But if the wave-length is increased to three or four times its original value by means of induc- tance inserted in these forms of antennae, the total decrement can thereby be reduced to 0.05-0.03. b. The effective resistance, R, can be cal- culated, if the total decrement, d, is known, from the equation d = irR > [Art. 27a] if C and L are also known. R may also be found by causing an un- damped primary circuit to act inductively upon the antenna and then proceeding as per Art. 766. 161 The following (" artificial aerial") method has also been widely used: A primary circuit (quenched gap circuit or undamped oscillations) can be loosely coupled by means of the switches U\ and U 2 (Fig. 208), either with the antenna (E S 2 aerial) or (dotted position of switches) with a condenser circuit S 2 S' 2 CR, having the same capacity and self- induction as the antenna, but, in addition, a variable resistance, R. The latter is adjusted until the ammeter, A, gives the same reading (current effect) with either the aerial or the condenser circuit. Then the resist- ance of the condenser circuit = the desired effective resistance of the antenna. In order that the resistance of the condenser circuit may be easily determined, it. is advisable to so construct it that its resistance shall be very small as compared to that of the variable resistance R, the latter being made of such wires and so designed that its effective resistance = its D.C. resistance [Art. 366], so that it may be measured with direct current. c. It is particularly interesting to separate the radiation resistance from the other parts that make up the total resistance. FIG. 208. 170 WIRELESS TELEGRAPHY If the form factor of an antenna has been found by current measure- ments and the antenna stands on soil of good conductivity, the radiation resistance can usually be calculated with sufficient accuracy. In this case the field of the grounded antenna (height, h) over the surface of the earth is identical with the field which would result from the antenna and its "image, " i.e., an oscillator whose total length I = 2h, there being no ground present. The only difference is that the energy radiation of the grounded antenna is only one-half the radiation of this oscillator, as in the former the lower half is missing [Art. 33]. Hence if, according to Art. 286, the radiation resistance, R? } of this oscillator of length I is given by then for the grounded antenna this must be = 1607r 2 /ah\ (T) in which a has the values given in Art. 25c. With sinusoidal current dis- tribution (simple antenna) R% = 36.6 ohms [Art. 266]. The following is an experimental method (A. ERSKINE-MURRAY, M. REicH 162 ) for approximately determining the radiation resistance of a transmitting antenna. At a distance, r, of at least several wave-lengths from the transmitting aerial, a tuned receiving aerial is erected and the current effects /i 2 e // and /2 2 C // determined at the bases of the sending and receiving antennae respectively. The height of the transmitting aerial is then altered a little (say, by simply raising or lowering the aerial wires slightly by means of ropes) and the measurements repeated, giving the new values, I'i 2 e // and / Ve//- The discharge frequency and wave-length must be retained constant. This method assumes that the electric field at the distance of the receiving antenna, and also the ground resistance near the transmitter are not appreciably changed by the change in the height of the aerial wires an assumption which of course is not always correct. Under these conditions we then have the following relations:! Conceive a sphere of radius, r, surrounding the transmitting antenna. Then the energy which passes through a square centimeter of the surface of the sphere per second ex E 2 eff , E being the electric field strength at the point in question [Art. 26] ; it also oc E 7") It /to " ^// 1 eff (1) d(di + d 2 ) . d(d( eff /I 2 ejf (2) * a, 6, d and p are factors of proportionality. This d should not be confused with the decrements. 172 WIRELESS TELEGRAPHY Subtracting one equation from the other, we obtain b or d and thereby the radiation resistance, 7 s , having previously determined the total effect- ive resistance, R and R r and also, when damped oscillations are used, the sum of the decrements [(di + ck) and (d\ + cZ 2 *)]. A test of whether the assumptions upon which the preceding equations were based hold approximately true in a given case can be obtained by repeating the measurements at one or two different antenna heights; the additional equations so obtained should give the same resulting values for b and d. d. Having determined the radiation resistance R? and the effective resistance Rj of the aerial wires, as well as the total antenna resistance R, then from R R? + Rj + R ej the value of R e follows. This, for antennae on firm ground, seems to amount to at least several ohms, 150 but depends entirely upon the form of the antenna, the frequency of the oscillations, the nature of the soil and the method of grounding. * If di and d'^d^ equation (1) may be applied to damped oscillations also. CHAPTER VII TRANSMITTERS OF DAMPED OSCILLATIONS 101. The Different Types of Transmitter. There are two methods, customarily applied for producing the oscillations in the antenna, viz., a. A spark gap is inserted in the antenna and the latter is charged by means of an induction coil or its equivalent. The antenna discharges across the gap, during which discharge the antenna oscillates in its natural period. This is the "simple" or "Marconi transmitter.' * b. The antenna is coupled to a condenser cir- cuit. This gives rise to two possible cases, viz., . 1. Two coupling oscillations are produced in both the condenser circuit and the antenna the "Braun transmitter," or 2. The oscillations of the condenser circuit are quenched after a few cycles and the antenna con- tinues to oscillate with its own damping the "quenched spark gap" or "Wien transmitter." 1. THE SIMPLE (MARCONI)* TRANSMITTER 102. General. a. The antenna has a spark gap, F (Fig. 209) at the bottom. It is advantageous to have an anti-node of FlG 2 Q9. current at the foot of the antenna, for, with a given voltage, this will make the current amplitude of the fundamental oscil- lation a maximum and the spark damping a minimum, with the spark gap lying in an anti-node of current. This condition is no doubt always obtained in practice by grounding. b. The combined or multiple forms of aerials [see Art. 92] increase the effectiveness of the MARCONI transmitter. For the same height, their effective capacity is much greater and if their form is properly chosen, the current distribution along the antenna is much better than with the simple aerial. Both these differences are factors favoring increased distance effect at a given voltage [Art. 93d. * MARCONI now also uses the coupled BRAUN transmitter exclusively or at least, mainly, for damped oscillations. However, it was with the simple form of transmitter that he attained his first successful results and demonstrated the possibility of wireless telegraphy by means of electromagnetic waves over great distances. 173 174 WIRELESS TELEGRAPHY 103. The Damping. a. Any of the spark methods of excitation inher- ently involve a consumption of energy in the spark in addition to the energy losses occurring in antennae without spark gaps. Accordingly, the efficiency is not as high as for an antenna without a gap. If we define the efficiency, as in Art. 99cc, as the ratio of the energy radiated by the fundamental oscillation in useful form to the total energy consumed by it in the same time, we have d? where d is the decrement of the antenna without the gap and d is the spark-gap decrement. If, however, we conceive the efficiency as the ratio of the useful energy, radiated at the fundamental oscillation to the total energy supplied to the antenna by charging, the result is even less favorable. In the MARCONI transmitter there necessarily exist at the start not only the fundamental, but also a series of partial oscillations. These are of no use so far as the distance effect is concerned, as the receivers are always tuned to the fundamental oscillation. Hence the energy consumed in one form or another by the upper partial oscillations represents a further loss which causes an additional decrease in the efficiency. 6. From observations made to date, it appears that with a given antenna, the effect of the oscillations and also the distance effect do not increase as the spark length is increased beyond a certain point, in fact, they decrease beyond this point.* Apparently this turning point occurs earlier, according as 1. The effective capacity of the antenna is smaller (hence in this respect a multiple. antenna is preferable to a simple antenna); 2. The radius of curvature of the gap electrodes is smaller (hence large spheres or plates are better than small spheres). c. In tuned telegraph operation, the MARCONI transmitter is at a dis- advantage on account of the great damping of the oscillations, although this is in part only a factor of strong distance effect. MARCONI trans- mitters can be constructed with decreased radiation damping [Art. 98] and a total decrement of about 0.1. But the weak distance effect of such a transmitter requires a very high potential if long distances are to be attained in telegraphing. This leads to such insulation difficulties, that the reliability of such transmitters becomes uncertain for regular operation, even though they may have shown good results under first tests. * It is questionable whether this is due solely to the influence of the changed gap length. Presumably brush discharge, and circuit losses played an important part in these experiments. TRANSMITTERS OF DAMPED OSCILLATIONS 175 2. THE BRAUN* TRANSMITTER 104. Nature of the Coupling. The coupled (BRAUN) transmitter consists of a condenser circuit, the u excitation circuit" as primary and the antenna as secondary circuit. The primary and second- ary circuits are tuned so as to be in resonance. a. The coupling may be purely magnetic, inductive, or it may be direct, conductive [Art. 526]. The former is shown diagrammatically in Fig. 210, the latter in Figs. 211 and 212. With direct coupling the secondary circuit is comprised of the aerial proper (plus BA in Fig. 212), a portion (BE in Fig. 211 and AE in Fig. 212) of the condenser circuit and the line to ground, or, if a counterpoise is used, FIG. 210. 1.0 FIG. 211. this and its leads. In addition to these, mixed or combined forms of these arrangements have been used or are still partly in use (e.g., Fig. 213). _^ 6. The direct and the in- ^ -^ p ductive connections do not give materially different re- sults. The direct coupling has the advantage of simplic- ity; above all it avoids the necessity of insulating the primary and secondary turns from each other, which to say the least involves considerable inconvenience for the inductive arrangement. Direct coupling is now in very wide use; the in- ductive or mixed form is possible only for very close coupling. 105. Coupled Transmitter for Antennae having High Damping. Very Loose Coup- ling. Under highly damped antenna are to be understood such whose decrement FIG. 213. * F. BRAUN was probably the first to introduce the coupled transmitter into the practise of radio-telegraphy. His patent is dated in 1898. In the same year E. DucRETET 164 made some tests in France using an "Oudin resonator" (in its essentials arranged like Fig. 212) and thereby already recognized the importance of tuning. 176 WIRELESS TELEGRAPHY (as that of a simple antenna) is 0.2 or more. Very loose coupling we may define as such that the complications discussed in Art. 58 (two frequencies even with tuned primary and secondary) are not noticeable.* a. So far as the frequency is concerned, the primary and secondary cir- cuits must be exactly in tune. b. The time change of the oscillations in the aerial is similar to that shown in Fig. 123. Only one oscillation (i.e., one frequency), whose amplitude first increases and then falls off, exists. The looser the coup- ling, the more nearly the rate of the decrease or falling off of the amplitude is that which would be obtained with an oscillation having the decrement of the condenser circuit, i.e., 0.06 to 0.1. The current distribution along the antenna is the same as for the natural oscillations of the antenna. It may at times be advantageous to so shape the current distribution, by inserting a condenser, as to bring the current anti-node quite high. The fact that this reduces the degree of coupling between the primary and secondary circuits as compared to having the coupling at the point of the anti-node of current [Art. 536] is not detri- mental in this case. c. Very loose coupling is used when it is essential to produce very slightly damped oscillations. In practice, however, there is always the accompanying requirement that the distance effect should be as great as possible without seriously increasing the damping. Now from Art. 88 it appears that the current effect in the receiver (circuit III) at very loose coupling is rapidly increased by making the coupling slightly closer. Hence for good distance effect it is important to make the degree of coupling as high as the sharpness of resonance will permit. 106. Coupled Transmitter for Antennae having High Damping. Close Coupling. When the greatest possible distance effect is desired with- out regard to high damping, close couplingf is in general advantage- ous because of the increased current amplitude it provides. a. We then obtain, whether the primary circuit is in tune with the antenna or not, two distinct oscillations of different frequency and hence [Art. 24] different current distribution along the antenna, different current amplitudes at the current anti-node and different decrements. As a mat- ter of fact the primary is probably always tuned to the secondary circuit. Then, from Art. 58, et. seq., we may conclude: the oscillation having the higher frequency (shorter wave-length) has 1. A greater current amplitude at the anti-node of current. 2. More favorable current distribution along the antenna; the cur- * That is K 2 < ( dl 2 ~^ 2 ) 2 [Art. 57]. f That is K*> (^F^- 2 ) 2 [Art. 58]. TRANSMITTERS OF DAMPED OSCILLATIONS 177 rent distribution curve for a simple antenna will be of the form of curve /, Fig. 214, for the shorter wave oscillation, but like curve II for the longer wave oscillation. With antennae having increased end capacity the two curves are not much different from each other. 3. A decrement which may be either greater or less than that of the longer wave oscillation, but never much different from it. Hence the effect upon a receiver is better, in fact much better in some respects, if the receiver is tuned for the oscillation having the higher frequency (shorter wave-length). b. Whether it is best to tune the con- denser circuit exactly to the antenna appears questionable according to Art. 89. It is probable that a better effect of the more rapid oscillation upon a receiver is obtained by giving the condenser circuit a frequency slightly higher than that which the antenna had before coupling. The author does not know whether this has been tested in prac- tice. An increase in the effect of more than a few per cent, can, however, hardly be ex- pected, judging from the laboratory results. c. According to Art. 88 the current effect in a receiver (which here corresponds to the measuring circuit) tuned to the higher oscilla- tion, is greatest when the coupling is between 4 per cent, and 10 per cent. However in Art. 88 the tests were made with a condenser circuit as secondary, while in practice we have an open circuit transmitter, whose current distribution must be considered. With antennae having increased end capacity this can make but little difference, but with others, the current distribution for the more rapid oscillation is improved as the coupling becomes closer. Hence we may conclude that a fairly close coupling is advantageous for antennae.* d. The current amplitude at the anti-node of current in the antenna, which largely determines the distance effect, is given, for the shorter wave oscillation, by the expression [Art. 61a], FIG. 214. * Widely varying degrees of coupling have. been and still are used in practice. The TELEFUNKEN Co. formerly used up to 10 per cent, coupling, in special caseshaving obtained excellent results with much higher degrees of coupling. The Eiffel Tower transmitter operates at 4.7 per cent, coupling. 48 FLEMING 1 states the usual range of coupling to be from 30 per cent, to 70 per' cent. Presumably the degree of coupling is of very little importance as long as it is kept over a certain lower limit. It would seem as if the worst effects of very close coupling are compensated by the resulting advan- tages in other directions. 12 178 WIRELESS TELEGRAPHY Hence, the frequency (wave-length) being given, it is advantageous to use antennae having large effective capacity. Similarly the primary cir- cuit should have as large capacity as possible, compatible with the avail- able energy. A limit to the amount of capacity is encountered in that, with a given frequency, increasing the capacity requires a reduction in the dimensions of the current path, which may make it impossible to obtain a sufficiently high degree of coupling.* This same reason may at times make inductive or mixed coupling necessary, in such cases where it is impossible to make pure conductive coupling sufficiently close without sacrificing the advantages of large capacity. 107. Coupled Transmitters for Slightly Damped Antennae. a. The case of very loose coupling need not be considered for antennas whose decre- ment is less than 0.1; for the object is to produce oscillations in the an- tenna which have not the high damping corresponding to the natural oscillation of the antenna, but the low damping of the condenser circuit. In the case we are considering, in which the condenser circuit has the same or only slightly lower damping than the antenna, this would have no practical value. b. The effects of varied degrees of coupling are qualitatively the same as for more highly damped antennae. The tests of Art. 88, conducted with a condenser circuit as the secondary, indicate that 6 per cent, is about the best coupling for the current effect. However, for antennae whose current distribution varies widely at different points it must be remembered that the closer the coupling is made the more advantageous for good distance effect does the current distribution become. How much this factor tends to displace the best degree of coupling is a question which has probably not been answered to date by actual tests. Practical experience seems to lead to the conclusion that either the degree of coupling is quite immaterial or its choice depends upon the special conditions of the particular case to such an extent, that no generali- zations can be made. The TELEFUNKEN Co. reports excellent results with a 60 per cent, coupling, yet this same company used a 4 per cent, coupling at its Nauen high-power station. c. As regards the kind of coupling it should be pointed out that for umbrella type aerials having very large effective capacity the direct (conductive) connection can be applied even for very close coupling. This can be made clear from a consideration of the equation [Art. 536] K : If the entire condenser circuit is used for the coupling and the coupling * The arrangements which F. BRAUN 165 has devised to meet this condition, under the name of " Energieschaltungen" (i.e., " energy connections") obviate this difficulty. TRANSMITTERS OF DAMPED OSCILLATIONS 179 is located at the anti-node of current in the antenna, then Z/i 2 approxi- mately = LI and i.e., the greater the effective capacity, C 2 , of the antenna, the closer will be the coupling. 108. Commercial Form of the Braun Transmitter. a. Condensers. The requirements of the condensers used are : 1. High breakdown resistance. 2. Small volume, convenient form. 3. Low energy loss due to dielectric hysteresis. 4. No brush discharge. These requirements are best fulfilled by air, particularly compressed- air condensers. MARCONI formerly used air condensers at atmospheric pressure at the Clifden and Glace Bay transatlantic stations, which were equipped with a tremendous battery of air condensers totalling 1.6 mf., and which were charged to a potential of about 80,000 volts. Compressed- air condensers of the form shown in Figs. 68 and 69 have been in use by the Nat. Elec. Sign. Co., on the recommendation of FESSENDEN. The disadvantage of air condensers lies in the relatively large dimensions necessitated by the low dielectric constant of air. In this respect con- densers of good flint glass are preferable. These are used either in the form of plate condensers, which are submerged in oil to prevent brush discharge (DE FOREST, F. DUCRETET, E. RoGER 166 and apparently also now in use by MARCONI in his transatlantic stations), or else in the form of Ley den Jars (the battery of jars shown in Fig. 70 is part of a TELEFUNKEN station of about 500 km. range). The Eiffel Tower station has Moscicki condensers [Art. 396]. 48 In order to minimize the brush discharge the jars are sometimes (e.g., as by the Nat. El. Sign. Co.) immersed in oil, or at least they are de- signed so as to be long and narrow (Telefunken) and are arranged in series-parallel combinations. For example, the Nauen station formerly had three batteries, each consisting of 120 jars in parallel (each battery having a capacity of about 1.2 mf.), joined in series (Fig. 215). b. As to the design of the current path of the condenser circuit it is hardly possible to make any general statements. Fig. 216 shows a construction of the TELEFUNKEN Co. As formerly used for a station of 1000 km. range, it was made of silver-plated copper tubing, some of the turns being joined in parallel. In the arrangements illustrated by Figs. 211 and 212, the contacts A and B are made so as to be movable, allowing a convenient adjustment of the frequency and the coupling. Needless to state, the current path 180 WIRELESS TELEGRAPHY should be so designed as to avoid eddy currents in the condenser coatings and other metallic conductors as much as possible. c. The spark gaps may have either stationary or rotating electrodes. Gaps with stationary electrodes should be designed with as large a radius of curvature as possible so as to avoid undue heating. The ring- TRANSMITTERS OF DAMPED OSCILLATIONS 181 FIG. 216. FIG. 217. 182 WIRELESS TELEGRAPHY shaped electrodes introduced by the TELEFUNKEN Co. (Fig. 217), have apparently been very satisfactory. Frequently the spark gaps are en- closed in a case so as to reduce the terrific noise which accompanies the discharge of very large capacities at high potentials. Enclosing the gap furthermore permits the use of gases other than air and above all makes the use of pressures higher than atmospheric possible. 167 In regard to rotating gaps and the use of air blowers see Art. 118. 3. QUENCHED SPARK-GAP TRANSMITTER. WIEN'S TRANSMITTER 109. Impulse Excitation. If we understand this term as covering all those methods in which the action in the primary exciting circuit lasts a shorter time than the resulting oscillations in the secondary, we may distinguish between the following kinds of impulse excitation. a. Attention to one method has already been called in Art. 5662. If a relatively highly damped primary circuit is caused to act upon a less damped secondary circuit very loosely coupled to it, there will result in the secondary not only an oscillation of the same decrement as that of the primary circuit, but also one having the lower decrement of the sec- ondary circuit. This latter oscillation continues long after the highly damped primary oscillations have disappeared. This method, at one time, before other methods of impulse excitation were known, was proposed (E. NESPER 121 ) for use in connection with measurements. It should not come into question, however, in radio- telegraph practice. High damping, i.e., high energy consumption in the primary, in conjunction with .loose coupling, i.e., with only a small part of the energy transferred to the secondary circuit, must necessarily greatly reduce the efficiency. b. The quenched gap method is far more satisfactory. Here, after the primary circuit has had the opportunity to give up the most of its energy to the secondary, the oscillations in the primary are quenched. The principle of this method was discussed in Art. 62 et seq., its practical application in Art. Ill et seq. c. Regarding a kind of mechanical quenching by means of a rotating spark gap see Art. 1186 and d. d. An impulse excitation in the true sense of the word is represented in Figs. 218 and 219. If the circuit from the cell E is first closed and then abruptly opened by means of the interrupter A, the resulting current curve will be about like that marked / in Fig. 220. This will produce an e.m.f. of the form of E in Fig. 220, in the coil S. The current resulting from this e.m.f. charges the condenser, C, which discharges in its natural period (G. EicHHORN 168 ). Hence, while with the quenched gap method we have regular oscillations in the primary which are quenched fairly rapidly, we have in this case of true impulse excitation, an aperiodic action which produces the oscillations in the secondary. TRANSMITTERS OF DAMPED OSCILLATIONS 183 The advantages of this method for measuring purposes have already been pointed out in Art. 78d. A good interrupter is of course essential; those having the contact fastened to a stretched string or wire which is vibrated by an electromagnet are well suited to the purpose.* A practical application of this method is in the form of the so-called LfiflfiM E FIG. 219. "station-testers," condenser circuits of adjustable frequency, which can be made to oscillate with the aid of very low amplitudes. Fig. 221 illustrates one of these as made by the TELEFTJNKEH Co. The connections are those of Fig. 218. The "make and break" of the direct current is accomplished by an interrupter (at the right of Fig. 221) similar to the ordinary electric bell or buzzer. Frequently wave meters, as for instance that of C. LORENZ and the newer type of the TELEFUNKEN Co., are arranged for use as "station-testers" also. e. Pure impulse excitation (as described in d) or else a condition lying between the cases of the quenched spark gap and pure impulse excitation can also be obtained with a condenser circuit as primary, if its oscillations disappear FIG. 220. at the first passage through zero or at least after a very few oscillations. This condition obtains under certain circumstances with hydrogen spark gaps (B. GLATZEL 169 ) and especially with unsymmetrical gaps,f also to some extent with mercury arc lamps, in fact even when the condenser circuit is not coupled to any secondary circuit. The quench- ing of the spark therefore is not dependent on the reaction of the secondary. * These are made especially for this method by the TELEFUNKEN Co., C. LORENZ and ROB. W. PAUL (New Southgate, London). t S. EiSENSTEiN 170 heated cathode, cold anode; L. E. CHAFFEE aluminium cath- ode, copper anode in hydrogen, also at times with gaps having symmetrical elec- trodes. 93 184 WIRELESS TELEGRAPHY Furthermore, in this case the occurrence of pure quenching action is of course independent of the degree of coupling. The latter may be made as high as the arrangement of the circuits permits. FIG. 221. E. L. CnAFFEE 2 has succeeded in obtaining continuous oscillations with such a gap,* in fact even with a frequency of about 3 X 10 6 per sec. (X = about 100 m.). For this purpose he so regulates his primary circuit and his current supply that after say three cycles of the secondary oscillations there is a discharge of the primary circuit. The first primary discharge excites the secondary oscillations, which fall off slightly during the next two cycles, but are given a new impulse in the third period at the right instant (in phase). 110. The Connections. a. These are in general the same for the WIEN trans- mitter (Fig. 222) as for the BRAUN transmitter. The antenna is coupled either inductively or conductively with the condenser circuit containing the quenched gap. 171 But the degree of coupling, while it may be chosen between wide limits * R. C. GALLETTi 1700 seems to have obtained the same result by means of a peculiar combination of a number of condenser circuits and spark gaps which act successively one after another. FIG. 222. TRANSMITTERS OF DAMPED OSCILLATIONS 185 with the BRATJN transmitter, must in the case of the Wien transmitter be adjusted as nearly as possible to the critical degree of the particular spark gap in question in order to obtain the maximum efficiency. For good quenched spark gaps this critical value is usually about 20 per cent.* In order to obviate the necessity of varying the degree of coupling whenever the wave-length is changed, the TELEFUNKEN Co. has devised the following arrangement (Fig. 223) : In the condenser circuit there is placed a coil, LI, of much greater self-induction than the rest of the primary circuit. This coil is used for a direct coupling of the antenna. FIG. 223. FIG. 224. The wave-length is changed by varying the coefficient of self-induction of this coil (which in the actual construction is in the form of a Rendahl variometer), and the tuning of the antenna to the primary circuit is effected by means of the special tuning coil L 2 . Under these conditions the coefficient of coupling remains constant and independent of the wave-length. We have 'LiZ* as after tuning CiLi must = C 2 L 2 [Arts. 3 and 27]. As a matter of fact, it is advisable to make the coupling somewhat looser for the shorter waves. For this purpose an additional coil, L f (Fig. 224) , which is not used for the coupling and whose self-induction is of no consequence for the long waves (Li being very large) but comes * In Art. 646 it was brought out that some quenched gaps have several critical couplings. That one which gives the greatest current effect in the secondary is of course chosen. With the gaps described in Art. 109, the coupling may be increased far above 20 per cent., in fact to 40 per cent, or even higher. So far, however, it has not been demonstrated that this has resulted in higher efficiencies than have been obtained with ordinary quenched gaps. 186 WIRELESS TELEGRAPHY into importance for the shorter waves (Li relatively small), is inserted in the primary circuit. b. The oscillations sent off into space are virtually the natural oscil- lations of the antenna. Their damping should be kept as low as possible to secure sharp resonance in the receiver. Accordingly it is universal practice to use antennae with greatly decreased radiation damping [Art. 98] and make provision for reducing their losses as much as possible in order to maintain a good efficiency for the antenna [Art. 99]. If it were desired to use antennae with strong radiation and the attend- ant high radiation damping and yet keep the damping of the oscilla- tions radiated into space fairly low, this could only be accomplished by FIG. 225. means of the arrangement shown in Fig. 225. The primary quenched gap circuit, /, is coupled to a very weakly damped condenser circuit (the "intermediate circuit" //) which in turn acts inductively upon the antenna through a very loose coupling. The relation between the inter- mediate circuit, II, and the antenna, ///, is the same as for a loosely coupled BRAUN transmitter [Art. 105], so that the oscillations of the antenna can be made to have the same low damping as the intermediate circuit. The intermediate circuit, which adds a material complication to the equipment, has been dropped in radio practice. 160 It has been found preferable to secure low damping for the radiated oscillations by reducing the damping of the antenna itself. An intermediate circuit would then be of value only if its decrement were much less than 0.05, say at least 0.01-0.005. This to be sure can be attained in the laboratory, but hardly in practical installations. 111. Practical Construction of Quenched Spark Gaps. Their short life stands in the way of the commercial use of the mercury arc lamp and the quenching tubes mentioned in Art. 63, unless some more durable form be devised in the future. Such quenched spark gaps as have TRANSMITTERS OF DAMPED OSCILLATIONS 187 found application in practice (some however only for a short time), have all belonged to the class of very short metallic gaps. Some of these spark gaps a number of them have been called "high frequency generators" were believed by their inventors, who worked with a very high spark frequency, to produce undamped oscillations. As a matter of fact, however, under the actual working conditions these gaps all acted as quenched spark gaps. FIG. 227. FIG. 228. a. The gap made by the Badische Anilin und Sodafabrik (VON KocH 172 ) has two concentric metal electrodes (Figs. 226, 227), very close together, between which a whirling eddy of air is blown. This eddy is obtained by blowing the air into the cylindrical or con- ical space between the two electrodes tangentially, as shown diagrammat- ically for a simple case in Fig. 228. This whirling of the air has the following advantages: 1. Intensive cooling of the electrodes. 2. The spark is blown about all over the elec- trode so that the discharge is constantly taking place at new spots, not heated by the preceding discharge. Hence we obtain in this way much the same advantages as are obtained with ro- tating electrodes, viz., regularity of the dis- charge, increased breakdown potential and thereby increased energy. 3. The spark gap is rapidly deionized, thereby raising the breakdown potential and preventing the formation of arcs. Ordinary fans or blowers generally are not sufficient for increasing the quenching action (H. RAU SS ); the air velocities obtained, unless very special means are employed, are not great enough to renew the air fully in the short time available for deionization [see Art. 656]. To be sure, the use of extremely powerful blowers makes it possible to materially increase the quenching action and also to obtain quenching in gaps several millimeters long, under conditions which without a blower would give rise to the different coupling waves (B. GLATZEL, PicnoN 173 ). b. In the "plate spark gap" 174 ' devised by E. v. LEPEL 175 the electrodes FIG. 229. 188 WIRELESS TELEGRAPHY are two metallic plates having a very narrow and finely adjustable space between them. They are separated by a paper ring (Fig. 229) made of FIG. 230. FIG. 232. carefully chosen material. The spark which passes over the gap in the space left open by the hole in the paper ring, tends to choose points along the edge of the paper, which is gradually burned away by the spark. TRANSMITTERS OF DAMPED OSCILLATIONS 189 Presumably the paper provides against the non-passage or at least the irregular passage of the spark, which would otherwise result with the low JL* FIG. 233. potentials (sometimes only 220 volts, D.C.) used by v. LEPEL. Fig. 230 illustrates the exterior construction of one of these gaps as arranged for water cooling. c. Particular credit is due the TELE- FUNKEN Co. 176 for its work in the construc- tive development of the quenched plate gaps. Its gap, a diagrammatic cross-sec- tion of which is shown in Fig. 231, has electrodes of silver-plated copper. Be- tween the rims of these plates is a mica ring which serves both as an insulator and an air seal. The widened space AA, is intended to prevent the spark from dis- charging at the rim of the mica ring. The distance between the faces of the elec- trodes is very small, about 0.2 mm. As only a comparatively very low FIG. 234. FIG. 235. voltage, and hence low energy in the individual oscillations, can be ap- plied in view of the shortness of the gap, the TELEFUNKEN Co. always uses a number, say 10 or 12 of the gap elements or sections shown in Fig. 231, joined in series, so as to form "series spark gaps" (Fig. 232). 190 WIRELESS TELEGRAPHY Excellent results have been obtained with these quenched gaps. Nor is it surprising that this form of gap should offer such advantages. The plate gap probably provides the most favorable conditions of all quenched spark gaps, certainly of all those employing stationary electrodes, As the ions are emitted from the metallic circuit, they always find themselves FIG. 236, in the immediate vicinity of a conducting surface toward which they are driven either by absorption or by the action of the electric field existing between the gaps. Consequently the degree of coupling may be brought as^high as say 20 per cent, even with relatively large amounts of energy and thereby a high efficiency can be attained. These gaps moreover have the great practical advantage of requiring hardly any attention or re-adjustment. The spark moves about over the TRANSMITTERS OF DAMPED OSCILLATIONS 191 surface of the electrode so that the latter is worn down very evenly and very slowly. Hence it need be cleaned only at very long intervals. The FIG. 237. series of gaps, moreover, offers a simple and convenient method of varying the radiated energy; to reduce this simply requires the short-circuiting of 192 WIRELESS TELEGRAPHY one or more of the individual gaps as may be necessary, for instance, when suddenly called upon to work with a much nearer receiving station. d. The gap made by C. LORENZ (O. ScHELLER 177 ) (Fig. 233) instead of having flat plates, consists of two concentric or almost concentric spherical surfaces. In this construction the gap length is practically the same at all points. e. The gap devised by W. PEUCKERT 178 also belongs to the plate-gap class, differing from the others in that at least one of the two electrodes is rotated. Two forms of this gap are shown in Figs. 234 and 235. In the former the plates, A and B, are vertical. One of these, A, is stationary, and through this oil is kept flowing, which then spreads out in the space between the two plates. In the second form (Fig. 235) the plates are in a horizontal plane. An atmosphere containing hydrogen between the plates is procured by allowing alcohol to drip from a holder on top into the gap. The Peuckert gap, which for a short time was made by the so-called Polyfrequenz-Elektrizitatsgesellschaft, is distinguished by great regu- larity of the oscillations. 112. Commercial Construction of the Wien Transmitter. a. Figs. 236, 237 and 238 illustrate three quenched spark-gap stations of the Tele- funken Co. 179 The explanation* given below for Fig. 236 is probably * 2. =40 amp. D.C. fuses. 3. = D.C. switch. 4. = Voltmeter switch. 5. = Voltmeter 250 volts. 6. = Motor starter. 7. = Field rheostat (motor). 8. =4 HP, 110 volts, 1500 r.p.m., D.C. motor. 10.1 11. f = High frequency protective devices (condensers). 12. J 13. = 2 kw., 250 volts, 500 cycles. Alternator. 15. = Slide rheostat for alternator field. 16. = 30 amp. A.C. fuses. 17. = A.C. switch. 18. = A.C. ammeter, 50 amp. 20. = Key. 21. = Choke coil for main supply circuit. 22. = 220/8000 volts transformer. 23. = Quenched spark gap 8 sections. 24. = Primary capacity 27 X 10~ 3 MF. 25. = Primary self-induction. 26. = Aerial hot-wire ammeter, 20 amp. 28. = Aerial variometer. 30. = Antenna shortening capacity. 33. = Receiver. 34. = Primary transformer coil of receiver., 42. = Telephone. TRANSMITTERS OF DAMPED OSCILLATIONS 193 sufficient for an understanding of the other two figures. The simplicity of these stations is at once evident. One need only compare Fig. 238, illustrating the quenched gap transmitter of the Nauen high-power station with the former BRAUN transmitter (Fig. 215) of the same station to appre- 13 194 WIRELESS TELEGRAPHY ciate this. This simplicity of quenched gap sets is made possible by the comparatively low potentials, e.g., 8000 volts in the 2 kw. station shown in Fig. 236, which can be used; this removes the necessity of connecting the condensers in series. In fact the use of the extremely convenient mica or paper condensers has become possible. The one disadvantage of these last-named condensers, namely, their high energy dissipation, does not come into question so much here as it does in the case of the BRAUN transmitter. For, as the oscillations of the primary circuit are quenched after the first few cycles, it is not so serious that the condensers consume somewhat more energy per cycle, particularly as this loss is very small in comparison to that in the gap. Nevertheless, as long as there are no special limitations to the space available, good Ley den jars, air or oil con- densers are always given preference in order to keep the efficiency as high as possible; and this can usually be done, as the moderate potential makes it possible to keep the dimensions of such condensers comparatively small. b. The circuit connections have already been discussed in Art. 110; they are probably as shown in Fig. 224 in most cases. When changing of the wave-length between wide limits is desired, this is frequently obtained by inserting a condenser (the smallbattery of Leyden jars marked 30, in Fig. 236) directly in the antenna for the shorter waves and connecting it in parallel to an inductive coil in the antenna for the longer waves [Art. 98c]. 4. GENERAL CONSIDERATION OF TRANSMITTERS OF DAMPED OSCILLATIONS 113. Operation by Means of Interrupted Direct Current. The use of spark coils (induction coils) operated by inter- rupted direct current is quite frequent in small stations. The induction coil must be able to give a relatively large amount of electricity at moderate potential rather than very high potential. The requirements are therefore quite different from those for the operation of X-ray (ROENTGEN) tubes. The usual electromagnetic interrupter suf- fices for small currents. It is economical of both room and energy, and hence continues to be used for small portable sets and air- ships, and also for quenched gap transmitters on shipboard as shown in Fig. 237 (particu- larly in the form of emergency equipment, operated from a storage battery), in which case however a higher frequency of the interruptions is required. For larger currents the mercury turbine interrupters have given good FIG. 239. TRANSMITTERS OF DAMPED OSCILLATIONS 195 service. Fig. 239 shows one of these as made by the Allgemeine Elek- trizitatsgesellschaft. The advantages of the mercury turbine interrupter, particularly for use in measurements, are: 1. It interrupts relatively large currents with great regularity; 2. the speed of the motor and hence the frequency of the interruptions are independent of the amount of current to be broken. 114. Alternating-Current Operation. The disadvantage of induction coils with interrupted direct current lies mainly in the difficulty of ob- taining sufficient quantities of electricity to charge large condensers to a high potential. The use of alternating-current and commercial trans- former designs at once suggests itself. A.C. operation differs according to whether 1. Ordinary spark gaps (BRAUN transmitter) or 2. Quenched spark gaps (WiEN transmitter) are used. a. In the first case (BRAUN), where in general very high potentials are used, if we were simply to connect the primary (low-tension side) of the transformer to the generator and the secondary (high-tension side) across the spark gap, the following difficulties would result: FIG. 240. 1. The A.C. transformer would continue to deliver current after the oscillations in the primary circuit had died out. That leads to the forma- tion of arcs, which heat the electrodes, ionize the gap for an unnecessarily long time, thereby lowering the breakdown potential and the initial amplitude of the oscillations. 2. The high-tension side of the transformer is almost short-circuited by the spark. This may damage the winding and also cause a disad- vantageous reaction upon the primary of the transformer (low-tension side). This second difficulty can be overcome, at least in part, by inserting condensers (dC 2 Fig. 240)* or choke coils in the leads from the trans- * The connections shown in Fig. 240 are those formerly used in the Marconi Station at Poldhu. 196 WIRELESS TELEGRAPHY former secondary to the spark gap, also by placing sufficiently large choke coils in the primary side ($ 2 , Fig. 240) of the transformer. AS to the first-named difficulty, much depends upon whether a very high discharge frequency e.g., 500 to 2000 per second for "tone trans- mitters" or a low frequency, say 5 to 25 per second, is used. With the FIG. 241. higher frequencies, the use of rotating spark gaps [see Art. 118&] is probably not feasible, if the energy handled is large. With the low frequencies, the various operating difficulties can be very ingeniously overcome by the arrangement 98 employed by the TELE- FUNKEN Co. under the name of "resonance inductor" or "resonance transformer." Spark FIG. 242. The transformer (or induction coil) has an open core; its terminals are connected to the condenser circuit in the usual manner (Fig. 241). The spark gap, E, is so adjusted in length that the normal secondary potential is far below that necessary to jump across the gap. We then have the case discussed in Arts. 67 and 68: an undamped oscillating primary circuit (armature of the alternator, coil D, and primary coil, Si, of the TRANSMITTERS OF DAMPED OSCILLATIONS 197 resonance transformer) coupled to a condenser circuit (secondary coil, $ 2 , of the resonance transformer and ABCDE). If these two circuits are adjusted so as to be in resonance, then the oscillations which result upon completing the primary current path, will be about as shown in Fig. 242. The current amplitude and hence also the potential amplitude in- crease with each cycle, the potential rising far above the normal value which would correspond to the ratio of the transformer, until, after a number of periods, a spark jumps across the gap, F, thereby causing the condenser circuit FABCDEF to oscillate rapidly. Due to these oscillations, the energy which has been accumulated in the condenser circuit S^ABCDE, is quickly consumed. Consequently there is a rapid fall in potential, no arc is formed and there is no great increase in the primary current. FIG. 243. This series of events then repeats itself, beginning with the next period. No spark occurs until, after a number of cycles, sufficient energy has been pumped [Art. 61c] into the secondary circuit (S^ABCDE) to bring the potential to the necessary amplitude. The advantages of this method are, therefore: avoiding of arcs and of short-circuiting of the secondary coil, low spark frequency and much higher voltage than would correspond to the transformer ratio.* * With an initial frequency of fifty cycles per second the spark frequency, if so de- sired, can easily be reduced to five per second and the secondary potential raised to three times the normal transformation value. 198 WIRELESS TELEGRAPHY For a given transformer and at a given frequency there is a best degree of coupling. In order that this may be secured, it is advisable to introduce adjustable inductive coils (D, Fig. 241) in the primary (or secondary) circuit or to use a BOAS resonance transformer (Fig. 243) * so as to allow variation of the coupling between the primary and secondary circuits. In the Telefunken Station at Nauen (Fig. 215) which formerly was operated with a resonance transformer, the high potential was obtained by means of four transformers in parallel (right front of Fig. 215). Two inductive coils (left front of Fig. 215) were placed in the primary circuit. 6. With quenched spark gaps (WiEN transmitter) the deionization of the gap [Art. 65] is so intense that the danger of the formation of arcs is far less than with ordinary gaps. Hence with quenched gaps there is nothing to prevent the use of A.C. generators and transformers, nor of machines for tone transmission, whose frequency is between 250 and 1000 cycles per second. It is advisable to so regulate the current that at most two or three par- tial discharges, preferably only one takes place during each half period. 176 With several partial discharges the tone in the receiving telephone [Art. 165] is apt to lose all its purity. And while a flute-like tone free from upper harmonics has no special advantage (in fact a tone having some pure upper harmonics seems to be better for receiving 180 ), an impure dis- cordant tone, as produced by numerous irregular partial discharges is certainly not favorable for good results. Moreover such irregular, partial discharges tend to weaken rather than strengthen the effect upon the tele- phone diaphragm, as it may often not have time to return to its posi- tion of equilibrium and in any case is forced into extremely complex movements. If the current is further weakened, the same effect as is obtained with a resonance transformer can be secured, a spark passing only every second or third half period; the tone heard in the receiver is then at the corre- spondingly lower octaves. c. One danger to which the alternator, and, under certain conditions, also the motor driving it are subjected, consists in the high frequency cur- rents induced in the leads by the primary circuit or antenna currents. These may produce very high potentials endangering the insulation. To counteract this as much as possible it is customary to connect condensers (those marked 10, 11, 12 in Fig. 236) or sometimes non-inductive resist- ances (incandescent lamps) in parallel with the motor and generator. 115. Direct-current Operation. a. Direct current at high potential may also be used in the supply circuit for charging the condensers, espe- cially when quenched spark gaps are used. The arrangement for these conditions is very simple (Fig. 244). The quenched gap circuit, /, is connected to the D.C. generator through series resistance, R Q , and induc- * These are particularly recommended for measuring purposes. TRANSMITTERS OF DAMPED OSCILLATIONS 199 tance, Lo. If the voltage, the gap length and the supply current are properly adjusted with respect to one another, instead of an arc resulting across the gap electrodes we obtain a varying potential like that illustrated in Fig. 291 (lower part). The direct current charges the condensers and their potential rises until the breakdown potential of the gap, F, is reached. Then the potential falls through a series of oscillations reaching zero or almost zero, whereupon the process repeats itself. The number of discharges per unit of time depends upon the size of the supply current, as this determines the time necessary for bringing the condensers to the breakdown voltage of the gap. The discharge frequency can accordingly FIG. 244. be varied between wide limits by regulation of the current supply. In order to secure regularity of the discharges, it is important to bring the voltage up as high as possible, at any rate not lower than 1000 volts. The series resistance in the supply circuit should preferably have a rapidly rising characteristic. The Nernst resistances, consisting of iron wires in a glass bulb filled with hydrogen, are particularly well suited to this purpose; metal filament lamps may also be used. b. The inductance, L , should be wound without iron ; if for any reason an iron core is desired, it is advisable to keep the magnetization above the saturation point. The effects of the self -inductance are as follows: 181 1. It greatly reduces the fluctuations of the supply current. 2. The maximum gap potential is greatly increased by the inductance, at times even rising considerably above the dynamo potential.* 3. The increased maximum potential improves the irregularity of the discharges and, finally, 4. The series resistance, and hence the energy consumed in the supply circuit, can be less than would be required without the self-inductance, without endangering the generator. c. The supply source is usually a D.C. generator several of these being connected in series if necessary. In the Marconi transatlantic stations * In the transatlantic Marconi stations (Fig. 255), the actual maximum gap potential, with 12,000 volts normal generated (storage battery) potential, is stated to be about 18,000 volts. 191 This is apt to be dangerous for the generator. See Art, 114c for methods of protection against this. 200 WIRELESS TELEGRAPHY at Clifden and Glace Bay, a storage battery of 6000 cells, corresponding to a potential of about 12,000 volts (see Fig. 255), is in parallel with the dynamos or can also be used alone. 116. Measurement of the Energy Supplied; Determination of the Efficiency. To measure the efficiency of a radio transmitter, it is neces- sary to find the energy used in the secondary circuit on the one hand and that supplied by the current source to th.e primary circuit on the other. a. Measurement of Energy Consumed in the Secondary Circuit. If the kind of secondary circuit is optional, it is advisable to choose a condenser circuit and to determine its effective resistance, R, by measuring the decrement [Art. 77, et seq.]. It may be well to construct the current path of braided conductors having individually insulated wires and to insert in this a resistance, R, of very fine constantan wire (special resistance material) which has the same resistance for oscillating as for direct cur- rents [see Art. 366] and which should be so large that the resistance of the rest of the current path becomes negligible in comparison. If a hot-wire ammeter is then inserted in the secondary circuit and /2 2 e// is measured, then Rl^e/f is very nearly equal to the energy consumed per second in the secondary circuit. If the indications of the hot-wire meter are not con- sidered reliable, the resistance can be placed in an insulated (against heat) vessel, filled, say, with oil and the heat developed measured calometric- ally, 71 from which the energy consumption per second in the secondary circuit may be determined. If the secondary circuit is an antenna, the current effect, / 2 2 //, is measured by means of an ammeter inserted in the antenna and the re- sistance, R, of the antenna is measured by one of the methods given in Art. 1006. Then the product RI Z 2 // = the energy consumed per second in the antenna. 6. Measurement of the Energy Supply. 1. In the case of low discharge frequency, as, e.g., with a Braun transmitter when operated by an influence machine, an induction coil with D.C. in- terrupter or a resonance transformer, the amount of energy supplied is found as follows: if V = the discharge potential and C = the capacity of the condenser, then the energy in the condenser at the instant just pre- vious to its discharge is H CV 2 . If now there are f discharges per second, it follows that the energy which the condenser receives from the supply circuit in each second must be f .%CV 2 .* The discharge voltage, for slow (static) charging of the condensers, can be determined by means of a suitable electrometer, 182 or from Table VI if the spark-gap electrodes are spherical. This, however, applies only if the discharge potential is the same with a static charge as under operating conditions; this condition * The assumption is that the condenser actually gives up its entire charge. If the condenser coatings after the discharge still retain a difference in potential, Vi (residual charge), then the energy supplied per second is f . %C(V 2 ~ 7i 2 ). TRANSMITTERS OF DAMPED OSCILLATIONS 201 can be approximated by projecting ultra-violet light upon the gap electrodes and using a very low spark frequency. 2. With a high discharge frequency it can not be assumed in general that the discharge potential is the same as the breakdown potential ac- quired by means of a static charge. In this case, if the discharge poten- tial is not too high, the energy supplied to the primary circuit can be meas- ured by an electrodynamic wattmeter connected either at the point W or W (Fig. 245), according to whether or not the energy consumed in the series resistances and inductances is to be included. W FIG. 245. As the frequency of the supply current is usually much higher than that of commercial alternating currents and as its form is far from being sinu- soidal, the ordinary commercial wattmeters are not to be recommended for this purpose. They generally possess phase errors which, at the lower commercial frequencies, approximately sinusoidal currents and not too great a phase displacement between current and voltage, can be roughly corrected for by means of compensating coils of some sort, but which otherwise may become quite considerable. Hence, we are limited to the use of special wattmeters 183 in which such errors are care- fully avoided, unless it is preferred to use a laboratory wattmeter, consisting of a fixed current coil in which is suspended a potential coil, very small compared to the current coil, made of very fine wire and carry- ing a small mirror attached to it. The suspension and current lead are best made of a bronze strip, while a telescope and scale serve for reading the deflection. The usual commercial wattmeter multipliers (series re- sistances are not suitable, as they are not free from capacity and self-in- duction; incandescent lamps* or liquid resistances! may be used. 3. Another method is based upon the use of the quadrant electrometer. A known resistance, R (Fig. 246) which must have practically no induc- tance or capacity and be independent of the frequency (e.g., a straight, thin constantan wire, placed in a quartz tube filled with oil for cooling) is inserted in the supply circuit, and across its ends, P and Q, the two quad- * But they must be used far below the point of incandescence. t e.g., the special boracic acid mannite resistances of MAGNANINI, with platinum electrodes as large as possible. Formula: 1500 g. water, 181 g. mannite, 62 g. boracic acid; to this add a little potassium chloride, the quantity being such as to give the resistance a very slight temperature coefficient. 202 WIRELESS TELEGRAPHY rant pairs, aa f and bb' of the electrometer are connected. The needle is joined to T (Fig. 246). Then the theory 184 will show that the deflection of the needle, The proportionality factor, a, of the instrument having been de- termined by calibration with static potentials and the current effect J 2 e // by an ammeter / connected in the circuit, then from the deflection, #, we Electrometer To Generator obtain the energy consumption per second between the points Q and T, and hence the energy supplied per second to the condenser circuit/. 117. The Key. a. Just as in ordinary wire telegraphy, keys are used for telegraphing. In radio work, however, the difficulty of breaking cir- cuits carrying heavy currents and having high self-induction presents it- self. This may at times give rise to large sparks, which rapidly destroy the key contacts. FIG. 247. These sparks can be reduced by not entirely breaking the current path. This, for instance, is done in the arrangement of Fig. 240 in which closing the key short-circuits the inductive coil, Si, and opening the key again causes the full current to flow through Si; in short the current is alternately increased and decreased, but never entirely broken. This method can be reversed by leaving the inductance, in circuit (Fig. 241) and short-circuiting the primary side of the transformer with each closing of the key. 185 TRANSMITTERS OF DAMPED OSCILLATIONS 203 A simple solution of the difficulty for moderately large currents is the construction (shown diagrammatically in Fig. 247) formerly used by the TELEFUNKEN Co. and the MARCONI Co., as designed by F. BRATJN and A. GRAY, respectively.* Underneath the key proper (i.e., the lever arm of the key) a spring, F, is placed, to which an iron armature, E, and a platinum contact, Ci, are attached. If the key and hence the spring, F, are pressed down, contact Ci touches contact C 2 and the path of the primary current of the induction coil or transformer, as the case may be, is completed. The current passes not merely through the key, along the route AFCiC 2 , but then also flows through the winding S. If now the key is released so as to move upward, contact Ci will nevertheless remain touching C 2 , as the magnetic action due to the current in S continues to hold down the armature E and hence the spring with the contact C\. Not before the primary current has become reduced to virtually zero, is the armature and hence the contact Ci re- leased from Czj at which time of course no spark is formed. b. In large stations "key relays" are probably always used. An ordinary key, manipulated by hand, closes an auxiliary circuit which (similarly to the remote control switches used for commercial high-tension work) operates the key or contactor opening the main circuit. The con- struction of good key relays, or " relay keys," as they are sometimes called, is by no means a simple matter, owing to the very rapid and frequent interruption of heavy current demanded in telegraph service. c. Where extremely rapid operation is required, as, e.g., in the Wheat- stone rapid method, automatic keys can be used. The principle of these is essentially as follows : The message is punched in telegraph code into a strip or tape of strong paper or other insulating material. For instance, the letter a would appear as in Fig. 248. The strip so formed is then pulled between the contacts of a key suitably designed, thereby completing the circuit as each perforation passes through the key. The actual con- struction of such rapid telegraph apparatus and automatic keys is usually very complicated. 186 118. Spark Gaps with Rotating Electrodes. -,a. In gaps having smooth electrodes, as in the MARCONI gap shown in Fig. 249, the spark, which always chooses the shortest or approximately shortest path through the air, will constantly jump across from different points of the electrodes, B, A and 5'.f This prevents harmful local heating of the electrodes, which heating, with fixed electrodes, tends to reduce the breakdown voltage of the gap and causes a rapid deterioration of the electrodes * These designs of the key also prevent the formation of very high potentials at the sudden breaking of the current. t A is a very rapidly revolving disc, while B and B' rotate slowly. 204 WIRELESS TELEGRAPHY and consequent irregularities in the oscillations. Furthermore, the air currents caused by the rotation assist the deionization of the spark gap and the cooling of the electrodes. In the spark gaps of F. DUCRETET and E. RoGER 188 (Fig. 250) which have a tube, C, as one electrode and a rotating sphere, S, as the other, a special strong air current is provided by the ventilator {*, or blower, V. The advantage of such provisions becomes more evident as the spark frequency and the current are in- creased, i.e., as the tendency to form arcs increases. b. The action of spark gaps having small projections on the electrodes as, e.g., that shown in Fig. 251 (R. FESSEN- DEN, NAT. ELEC. SIG. Co.) or in Fig. 252 (MARCONI Co. 189 ) depends largely upon the shortest distance be- tween the electrodes, their width and their velocity. Let us first assume that the minimum distance between the electrodes is such that Bo FIG. 250. the highest potential occurring across the gap will be just sufficient to create a spark discharge. Then 1. With moderate speed and moderate width of the electrodes, the gap will have the advantage, for A.C. operation, of good cooling of the elec- trodes and the prevention of arcs, for the gap length grows so rapidly TRANSMITTERS OF DAMPED OSCILLATIONS 205 206 WIRELESS TELEGRAPHY after the oscillations have died out that there is no opportunity for an arc formation. In this case then, the spark gap is mounted directly on the shaft of the alternator (Fig. 251) or of a synchronous motor and the number and location of the projections or sparking points so chosen that they are nearest together (minimum gap length) at the instant when the A.C. voltage is at its maximum. With D.C. operation a gap of this type secures regularity of the dis- charges, the frequency being regulated by the speed control of the driving motor. If this frequency is high enough an audible sound or tone (tone transmitter) is obtained in the receiving telephone. 2. If the electrodes are very narrow and their speed very high, the result may be as follows: The oscillations of the condenser circuit commence at just about the instant when the electrodes are closest together. While the oscillations continue, the gap length increases very rapidly. At the same time the potential amplitude in the condenser circuit, if coupled at all closely to a secondary circuit, rapidly falls off [Art. 59c]. The effect of both these factors, i.e., increasing gap length and decreasing potential, FIG. 252. FIG 253. may be to disrupt the spark after a very few periods, even in such cases where, if the electrodes remained stationary, the spark would not be auto- matically quenched after half a cycle. (This is sometimes referred to as "mechanical quenching.") Whether or not a mechanical quenching results, depends largely upon the velocity, the wave-length and the coupling [Art. 59c] used. If the TRANSMITTERS OF DAMPED OSCILLATIONS 207 peripheral velocity of the electrodes is taken at 200 m. per second (a speed which is obtainable [see d]), and a 5 per cent, coupling employed so that the potential amplitude in the primary circuit will be zero or nearly zero after ten periods of the oscillations, then the movable elec- trode will in this time have covered a distance of 2 cm. with a wave- length of 3000 m., and a distance of 4.4 cm. with a wave-length of 6700 If the gap is properly constructed, however, the distance between the m. HHHHHHHHHHH Storage J3attery - Inductance FIG. 254. FIG. 255. electrodes will have become so large by that time that there is very little chance for the spark to form again during the next few periods, particularly if the minimum gap length is very small. In all cases the effect of increasing the gap length is assisted by the strong air currents formed. In order to fully take advantage of this, spark gaps with rotating electrodes have been constructed in the form of actual ventilators (Fig. 253, Balsillie System 190 ). c. Another possible case consists in having the gap length less than the length which the potential used can just jump across (Fig. 254, "short -circuit spark gap") and to revolve the electrode projections at very high speed. When a projection of the rapidly rotating electrode, F, approaching the stationary or slowly rotating electrodes FI and F 2 , comes sufficiently close, a discharge takes place. While this discharge is occurring the electrodes come nearer together and the gap length, the * MARCONI'S transatlantic stations [d]. 208 WIRELESS TELEGRAPHY gap resistance and the energy consumption are reduced to very low values [Art. lid]. Such spark gaps combine the two advantages of compara- tively high initial voltage with a relatively short average gap length and a very short gap length at the time when the oscillations of the discharge have already become very low in amplitude. The best results are ob- tained, other things being equal, with the highest speeds and, for a given speed, with the greatest retardation of the discharge [Art. 426]. Ac- cordingly ultra-violet light is to be avoided as much as possible. d. The spark gap shown in Fig. 255, which is use d in the transatlantic MARCONI stations at Clifden and Glace Bay, 191 is a combination of a short-circuit gap and a mechanically quenched gap. The electrode pro- jections on the wheel F are so shaped that the gap between the discs FI and F 2 is very small only while there is a projection between the two discs. On the other hand, the peripheral velocity of F is so great (about 200 m. per second), the degree of coupling so low it is reported as being approximately 5 per cent. and the width of the projections is so meas- ured, that the spark is disrupted after half a cycle and does not again form until the next projection comes into play. The discharge retardation seems to be used to excellent advantage in these gaps; at a potential of 15,000 volts and a frequency of 45,000 cycles per sec. (X = 6700 m.) it is claimed that the spark occurs only about one period before the instant at which the projections of F are at their minimum distance from the discs FI and F 2 . The result of this short spark gap combined with the heavy current due to the relatively high discharge voltage and capacity, is a very low energy consumption in the primary condenser circuit. The total decrement of the primary circuit, when not coupled to the secondary, is claimed to be only about 0.03 to 0.06.* 6. COMPARISON OF THE DIFFERENT TYPES OF TRANSMITTERS 119. Difference between the Coupled and the Simple (MARCONI) Transmitter. a. That the coupled transmitters are more complicated and that it costs more to construct them is self-evident. In addition, a relatively small induction coil with mechanical interrupter and a few stor- age battery cells usually suffices for a simple transmitter with its low capacity; the operating costs are therefore exceedingly low. Hence, if only comparatively short distances f are to be covered in telegraphing and if it is important to keep the energy consumptionf as low as possible, * The equivalent resistance of the entire condenser circuit with its spark gap is given in one case as 0.022 ohm. t Ranges of 100-150 km., with masts about 30 m. high have been attained, with the simple MARCONI transmitter. | e.g., stations which are difficult of access or light portable sets (such as light- ship stations or portable military stations). TRANSMITTERS OF DAMPED OSCILLATIONS 209 the simple MARCONI transmitter offers great advantages, which in fact have caused it to be retained in use to the present day and which will perhaps keep it in use for quite some time to come as an emergency transmitter. b. But it is in relation to the question of energy that the simple MARCONI transmitter is at its greatest disadvantage. To be sure, it does not require much energy, but the only way in which the energy can be increased is by raising the initial voltage (see e). On the other hand, it is possible to radiate much greater quantities of energy in the form of electromagnetic waves from a coupled transmitter, in view of the large capacity available in the primary circuit, than from a simple MARCONI transmitter at the same potential. Closely related to this is another advantage of the coupled trans- mitter, viz., the oscillations can be so modified, one may say, so formed, as to be best adapted for a particular receiver. If it is important to have great amplitude for the waves emanated, this can be secured without bringing the damping so high as to be prohibitive. Again, if very low damping is desired, this also can be obtained without causing too, great a reduction in the amplitude of the wave. It is for these reasons that in the one extreme case, where the greatest range is the object, as well as in the other, where the sharpest possible tuning is the desideratum, coupled transmitters are always used. c. As regards the energy losses it should be borne in mind that in the simple MARCONI transmitter, the spark gap lies in the antenna, while in the coupled transmitter the gap is transferred to the condenser circuit. As to which is the preferable location depends upon the particular circumstances. One thing is certain: when the object in view is to radiate oscillations having the lowest possible decrement from a given antenna, the coupled, and above all, the WIEN transmitter is far superior to the simple MARCONI transmitter. In both cases, with the simple as with the WIEN transmit- ter, we obtain practically the natural oscillations of the antenna; but in the case of the simple transmitter, the spark decrement, which in itself is about as large as that of a weakly radiating antenna having no spark gap, is added to the decrement of the antenna of the WIEN transmitter. d. The additional advantage of BRAUN'S coupled transmitter is that the upper partial oscillations of the antenna are not produced with an appreciable amplitude. This energy loss is therefore not involved. But in the BRATJN transmitter, at least when closely coupled, a second wave is obtained, which consumes energy, but is of no value for the distance effect in the customary receiving arrangements. e. A further material advantage of the coupled transmitter lies in the fact that the antenna is charged, not directly by the induction coil or 14 210 WIRELESS TELEGRAPHY transformer, as is the case with the simple transmitter, but only by the oscillations. The result is that the insulation of the antenna becomes a much simpler problem and that when slight faults occur in the insulation these have but little effect upon the oscillations [Art. 43]. In the simple transmitter, the slightest defect in the insulation suffices to endanger the entire operation. There is a natural tendency to belittle this last element and to assume that good successful insulation can present no serious technical difficulties. The fact remains that insulation troubles, especially in the tropics, are often so great, that they have caused the failure of entire installations 192 [see Art. 126]. 120. Comparison of the Braun and Wien Transmitters. a. The main advantages of the WIEN transmitter as compared to the BRAUN trans- mitter are as follows : 1. In the BRAUN transmitter the oscillations and hence the energy con- sumption in the primary circuit last as long as in the secondary. In the Wien transmitter they last only for a few periods. 2., The BRAUN transmitter produces two waves in practice, of which only one is fully made use of in the receiver. The WIEN transmitter emits practically only one wave from its antenna [Art. 78c]. As regards this second point it is of course possible to prevent the formation of two waves by means of very loose coupling [Art. 105], but then the transfer of energy to the secondary circuit becomes very ineffi- cient. The former frequent practice of making the coupling just suffi- ciently loose to make the two coupling waves only slightly different in their wave-length, is of little or no use. The two coupling waves in that case act upon the receiver like a single wave of much higher damping, which makes the advantage of the coupled transmitter more or less illusory. As to the first point, this would absolutely fix the superiority of the WIEN transmitter, so far as efficiency is concerned, if the same potentials and the same spark gap were used in the BRAUN and WIEN transmitters. But, as a matter of fact, a relatively long spark gap, with resultant low gap decrement [Art. lid], is used in the BRAUN transmitter, while either low potentials and short gaps or high potentials and a series of gaps^'.e., in either case, a high gap decrement [Arts, lid and 12] are used in the WIEN transmitter. Hence the energy consumed per cycle is much greater in the WIEN than in the BRAUN transmitter, so that the total energy loss in the gap circuit of the WIEN transmitter may be quite considerable in spite of the short duration of the oscillations.* * This no doubt is also why it is so important to have the coupling in the WIEN transmitter as close as possible, thereby minimizing the duration of the primary oscillations, without, however, making the coupling so close as to impair the quench- ing action. TRANSMITTERS OF DAMPED OSCILLATIONS 211 Nevertheless the modern WIEN transmitters seem to be much more efficient than the old BRAUN transmitters.* But in acknowledging this it must be remembered that formerly fundamental principles in the con- struction were often disregarded either for the sake of convenience or for other specific reasons; some of the old transmitters actually appear as if they had been intended not merely for telegraphing, but also for warming the room in which they were located by the heat developed by the eddy currents. In the meantime experience has taught the necessity of applying the principles discovered in the laboratory to commercial sta- tions so as to minimize all losses of energy. Hence we should not compare an old, poorly constructed BRAUN transmitter with a new, WIEN trans- mitter designed and constructed with the proper care. It is readily conceivable that the mechanically quenched gaps should have high efficiency, as the use of long sparks and therefore low gap decre- ments is here possible. In any case, MARCONI'S combination of mechan- ical quenching with short-circuiting the gap must be very efficient, for it unites good quenching action 191 with low energy consumption in the con- denser circuit. Quenching tubes also permit the use of long sparks with low gap decre- ments. Their use can give very good efficiency M. WiEN 92 obtained 80-60 per cent, efficiency (ratio of secondary to primary energy) at 30,000- 80,000 volts primary although the energy loss in the tube is added to that in the spark. b. The use of high initial voltages and long sparks in the BRAUN transmitter is not accidental. Nor does the reason lie solely in the lower decrements of the long sparks-. For in the BRAUN transmitter, of the two possible methods of increasing the energy of the transmitter (i.e., increas- ing either the initial voltage or the spark frequency) only the former can be done in a simple way, the latter involving considerable difficulties. If the gap has stationary electrodes, then, even with such large dimensions as belong to the spark gap shown in Fig. 217, local heating of the elec- trodes, with all its attendant disadvantages (arcs, reduced breakdown potential), is unavoidable, if the spark frequency is brought to the region of 1000 cycles per sec. The use of rotating electrodes, however, is in- * COUNT ARCO IGO claims that the efficiency (energy of the secondary divided by the primary energy) of the TELEFUNKEN quenched gap transmitters is about 85 per cent. 193 M. WiEN 17 obtained the following figures from a very carefully designed BRAUN transmitter at an initial potential of 72,000 volts: di = 0.034, d 2 = 0.175, K' = 0.032, 77 = 82 per cent, and for di = 0.034, <2 2 = 0.087, K' = 0.024, rj = 66 per cent. (i) = efficiency). Such high efficiency, however, is attained only by the most careful construction of the primary circuit; as soon as WIEN replaced the compressed gas condensers he was using by MOSCICKI condensers, the efficiency dropped from about 80 to about 69 per cent. The efficiency of the BRAUN transmitters which were used in practice was much lower. 212 WIRELESS TELEGRAPHY herently a complication, so that this means is hardly apt to be chosen except for large stations [see Art. 114a]. In the WIEN transmitter these conditions are more favorable. Here the oscillations in the primary circuit are only of very brief duration, so that the heat developed in the gap, at the same .current amplitude, is much less. Furthermore the deionization of this type of gap is in itself so rapid, that there is but little tendency to form an arc. Hence, the use of high discharge frequencies involves no difficulties in the Wien trans- mitter [see Art. 1146], and the result, namely, a high, pure note in the re- ceiving telephone and a lowering of the antenna potential, thereby reducing insulation difficulties, has proven its value in daily practice. c. In the WIEN transmitter, there is an important advantage, particu- larly for portable stations, in conjunction with the short duration of the primary oscillations, namely, the possibility of using condensers made of mica or similar dielectrics. The doing away of the series connection means a general simplification, not only for portable sets, but for all other stations as well [see Art. 112a]. The brush discharge on the condensers, which is very harmful in the BRAUN transmitter and necessitates undesirable complications [Art. 108a], is of no importance in the WIEN transmitter. CHAPTER VIII HIGH FREQUENCY MACHINES FOR UNDAMPED OSCILLATIONS 121. The Alexanderson-Fessenden Machines. It is reasonable to expect, a priori, that undamped oscillations of high frequency can be generated by a machine in the same way that commercial alternating currents of lower frequencies are produced. But it is an exceedingly difficult problem when frequencies of the order of 10 5 cycles per sec. are to be obtained. First of all, at these high frequencies, the hysteresis and eddy current losses become very large; a radical attempt to prevent the former by building machines without iron was soon abandoned as impractical. Then the structural difficulties increase very rapidly as the frequency is raised. Assume that a frequency of 10 5 cycles per sec. is obtained at the maximum allowable speed of 20,000 r.p.m.;* then if the diameter of the rotor is 305 mm.,t a path of only 3.2 mm. remains for the generation of each cycle, that is, 3.2 mm. is the maximum width available for a pair of armature coils with their insulation. And if this width is not to be further reduced, a high speed is unavoidable, thereby involving the well- known mechanical difficulties attendant upon rotation at such velocities. In spite of these difficulties, N. TESLA'S early efforts in this direction have been renewed again and again, particularly in America, 194 where FESSENDEN devoted himself to the problem. The high frequency alter- nators which through his influence were built for the NAT. ELEC. SIG. Co., by the GENERAL ELECTRIC Co. from E. F. W. ALEXANDERsoN's 195 designs, probably represent the best which has been achieved in this field of work in the past. a. The 100,000 cycle ALEXANDERSON alternator (A = 3000 m.) is of the inductor type. Fig. 256 is a diagrammatic cross-section of one of these alternators. The excitation is obtained by means of a single large field coil, S, which is wound around the entire machine and is supplied with direct current. The magnetic flux lines, M , of this coil pass through the iron cores, E\ and Ez, of the small armature coils, Si and *S 2 . The only movable part, Ji, has teeth or projections, Z, of iron, at its periphery. When one of these teeth is just between the armature coils, Si and $ 2 , the magnetic flux, M, has a path almost entirely through iron, excepting only at the very small air gaps between the teeth, Z, and the cores, Ei and E 2 ; in this position then, * Speed of ALEXANDERSON'S machine at a frequency of 10 5 cycles per sec. f Diameter of ALEXANDERSON generator. 213 214 WIRELESS TELEGRAPHY the magnetic reluctance is a minimum, the magnetic flux passing through the armature cores, EI and E z , a maximum. When, now, a space in- stead of a tooth lies between armature coils, the air gap, and hence the magnetic reluctance, are much larger, so that the amount of flux through the armature windings is very small. Hence as the movable part, J, rotates, the magnetic jflux passing through the armature coils varies peri- odically between a maximum and a minimum value, so that an oscillatory e.m.f., whose frequency = the product r.p.m. X number of teeth, is induced in the armature winding. The rotor of ALEX ANDERSON'S ma- chine is shaped like the cross-section, J, in Fig. 256 and has 300 teeth. The space between the teeth is filled with a non-magnetic material (phosphor- bronze) so that the surface of the rotor, J, is quite smooth, thereby preventing any material loss due to air friction (windage). The armature winding, in which the oscillatory e.m.f. is induced, does not, properly speaking, consist of coils, but of a single wire wound in a wave- shaped form (Fig. 257) ; any two such consecutive U-formed wires may be considered as a pair of coils of one turn each, joined in series but so as to oppose each other. Fig. 258 shows one-half of the completed armature. The capacity of the machine shown with its D.C. motor in Fig. 259* increases as the air gap between the armature and the rotor is decreased. It was 2.1 kw. in one machine having a 0.37 mm. air gap. The author has no record of its efficiency. b. A second machine with a frequency of 50,000 cycles per sec. (\ = 6000 m.) and a capacity of 35 kw. is shown in Fig. 260. This alternator was also designed by ALEXANDERSON and has a diameter of about 1 m. Further details had not been published at the time of writing. c. R. A. FESSENDEN 196 has described still another high frequency generator. Like ALEXANDERSON'S machine, it is also of the inductor type, but is characterized in that its movable portion (J, Fig. 256) acts at the same time as the short-circuited armature of an * High frequency alternator to the left, coupling in the middle and motor at the right. Diameter of alternator about 30 cm. FIG. 256. s v FIG. 257. HIGH FREQUENCY MACHINES FOR UNDAMPED OSCILLATIONS 215 FIG. 258. FIG. 259. 216 WIRELESS TELEGRAPHY A.C. motor (A.C. frequency = 500 cycles per sec.). This machine is very simple in construction and is claimed to have given 2.5 kw. at JV" = 1 X 10 5 cycles per sec. FIG. 260. 122. Goldschmidf s High Frequency Generator. R. GoLDscHMiDT 197 has attacked the problem of generating the high frequencies needed in radio-telegraphy along a different path. a. The basis of his method is as follows: If a coil, R (rotor), revolves in the magnetic field of a fixed coil, S (Fig. 261), through which a direct current is flowing, then the frequency, N, of the e.m.f. induced in R is equal to the number of revolutions of R per unit of time. But if an alternating curient of frequency N 1 flows through the coil S, it can be shown 198 that the e.m.f. induced in R may be considered as made up of one e.m.f., 8, of the frequency N + N f and another, >', of the frequency, N N'. What has just been stated in regard to the rotor with respect to the stator, must necessarily also hold for the stator with respect to the rotor; for as the induction depends only upon the relative motion of the two coils, the same result would occur if the rotor were held stationary and the stator rotated in the opposite direction. Hence we may state: If an alternating current of frequency N f is flowing through the rotor while the latter makes N revolutions per second, its field will induce an e.m.f., 8, of frequency N + N', and another, 8', of frequency N N', in the stator. HIGH FREQUENCY MACHINES FOR UNDAMPED OSCILLATIONS 217 6. Now consider the arrangement of Fig. 262. The storage battery, B, sends direct current through the stator winding, S. Then an e.m.f., 81, of frequency N, where N = revolutions per second of the rotor, is induced in the rotor. This e.m.f. sends an alternating current, /i, of the same frequency through the short-circuited rotor winding. Then, according to a, there is in turn induced in the stator an e.m.f., 82, of frequency N + N = 2N and another, 8' 2 , of frequency N N = 0; the latter, therefore, is not an oscillatory e.m.f. The e.m.f., 8 2 , induces a current 7 2 in the circuit comprised of stator winding, S, condenser, C, and the result of this current is that an alter- nating field, of frequency 2N, is superimposed upon the constant magnetic field of the direct current. This, according to a, results in an e.m.f., 83, of frequency 2N + N = 3N, and another, 8' 3 , of frequency 2N N = N in the rotor, the latter FIG. 261. FIG. 262. e.m.f., 8' 3 , adding itself to Si. The alternating current / 3 , due to S 3 and of frequency 3N flows through the rotor winding and, according to a, induces in the stator winding, S, an e.m.f., 8 4 of frequency 3N + N = 4/V and another, 8%, of frequency 3N N = 27V, the latter having the same frequency as 8 2 , upon which it is superimposed and so on. The result of the arrangement of Fig. 262 must therefore be the forma- tion of alternating currents whose frequencies are 2N, 4JV, QN, etc., in the stator and A 7 ", 3N, 5N, etc., in the rotor. c. However, what is needed for radio-telegraphy, is an oscillation of one single frequency in the antenna. To obtain this, GOLDSCHMIDT this comprises the second essential feature of his method makes use of the resonance principle, by means of which he brings the amplitude of the oscillation desired for actual service and of those oscillations which de- termine this useful oscillation, to such high values that the amplitudes of the other oscillations disappear by comparison. 218 WIRELESS TELEGRAPHY Fig. 263 is the diagram used by GOLDSCHMIDT himself to explain this method. The circuit RC B D 2 C^, which is tuned to the frequency N, serves to strengthen the current A; the amplitude of the rotor current depends only upon the ohmic resistance* of this circuit. At most only a very small part of the current, /i, flows through the condenser (7 6 , for at the frequency, TV, the inductance of the winding Z> 2 is made equal to the condensance (or capacity react- ance) of the condenser C 4 :* hence the impedance of the branch D^C^ becomes much lower than that of the branch containing CB, whose impedance is simply the conden- sance of condenser C$. The stator current / 2 , of fre- quency 2TV, attains a very great amplitude due to the fact that the circuit SCiDiCz is tuned to the frequency 27V. Furthermore this 263. current is prevented from flowing into the antenna because the in- ductance of the winding DI = the condensance of C 2 at the frequency 2N. The resonance circuit for the rotor current 7 3 , whose frequency is 3TV, is The circuit $C2-antenna-groundt is tuned to the frequency, 4N, of the useful current, 7 4 . The latter flows with any appreciable amplitude only through the antenna and not through the shunt D^C^, as the impe- dance of this shunt is much greater, at this frequency than the condens- ance of the antenna capacity. If it were desired to use the frequency 3N, the antenna would have to be connected to the rotor in place of the condenser Cs, and the condenser Cz and inductance DI could be omitted from the stator circuits, if Ci were * It is well known that the current, 7, in a circuit consisting of capacity, C, self- induction, L, and resistance, R, when the impressed or induced potential is F , is given by So that if coL = > we obtain coC i.e., only the ohmic resistance, R, determines I [see Art. 676]. t The antenna-ground circuit may, for the purpose in view, be considered as simply a condenser. HIGH FREQUENCY MACHINES FOR UNDAMPED OSCILLATIONS 219 properly dimensioned.. This would materially simplify the connections but, of course, the frequency would only be brought to three times the initial frequency, the latter being determined by the number of poles and speed of the machine. d. At the right of Fig. 264 is shown a GOLDSCHMIDT machine which was put into service by the C.LORENZ Co. at the EBERSWALDE station in April, 1910. The driving motor is at the left and in the center is a gear case, FIG. 264. needed to bring the comparatively low motor speed up to the high speed required by the generator. The latter, of course, is of multipolar con- struction and gives 12.5 kw. at a frequency of 3 X 10 4 cycles per sec. (X = 10,000 m.) with an efficiency of 80 per cent., and 8 to 10 kw. at 6 X 10 4 cycles per sec. (X = 5000 m.).* * In regard to the high frequency generator of COUNT ARCO (Telefunkeri) and his method of frequency transformation, 199 see the remarks at the end of the book con- cerning developments in radio-telegraphy during the last few years. CHAPTER IX UNDAMPED* OSCILLATIONS BY THE ARC METHOD 1. THE VARIOUS ARRANGEMENTS 123. The Problem and Its Solution by V. Poulsen. The requirements which undamped oscillations must meet in order to be of use for radio- telegraphy are as follows: 1. Their frequency must lie within the range used in wireless telegraphy (i.e., N must be between about 10 6 and 4 X 10 4 cycles per sec., corre- sponding to values of X from 300 to 8000 m.). 2. Their energy must be sufficiently great, and 3. Their amplitude and frequency must be nearly enough constant for radio purposes. The arrangement by means of which undamped oscillations can be produced in a condenser circuit is that shown in Fig. 244, in short it is the same as that by means of which a quenched gap circuit can be ex- cited with direct current. Whether this arrangement will give undamped or damped oscillations depends upon the construction of the condenser circuit, the nature of the gaseous gap, F, the dynamo voltage, the re- sistance and self-induction of the supply circuit and, finally, upon whether, and if so to what extent the condenser circuit is coupled to a secondary circuit. a. ELIHU THOMSON and N. TESLA, later also R. A. FESSENDEN, made early 200 use of this arrangement for the purpose of continuously exciting the natural oscillations of a condenser circuit by means of direct current [Art. 115]. It is highly improbable that either THOMSON or TESLA suc- ceeded in actually obtaining undamped oscillations of such frequency and energy as come into question for radio-telegraphy. THOMSON'S spark gap (the arc) had solid metallic electrodes in air at atmospheric pressure; with such electrodes, however, and potentials of not much more than 1000 volts, it is hardly possible to obtain undamped oscillations at * By "undamped oscillations," the author understands oscillations of which the amplitude remains unchanged from period to period (in the arc method, oscillations of type / or 77 [Arts. 130 and 131]). The designation "continuous" oscillations is also frequently used. But against the use of this term stands the existence of con- tinuous oscillations whose amplitude varies from period to period (see Fig. 290 and Art. 109e). The name ''continuous" is therefore not sufficiently specific for the case in question. 220 UNDAMPED OSCILLATIONS BY THE ARC METHOD 221 high frequency, of sufficient regularity to meet even the most modest radio requirements. TESLA made some use of carbon electrodes, which are far more apt to make the production of undamped oscillations feasi- ble; but even if he obtained undamped oscillations, their frequency can- not have been very high, as even his highest discharge frequency gave an audible tone. Then, somewhat later, W. DuoDELL 201 experimented with the ar- rangement of Fig. 244 using carbon electrodes and undoubtedly obtained undamped oscillations in this way. In fact he at that time discussed the essential condition for their production and through his experiments this method of generating oscillations became quite popular. Soon after this the action of damped oscillations as used in radio- telegraphy was investigated more thoroughly and the .advantages of low damping in the transmitter became evident. From then on un- damped oscillations were the sole aim of nearly all working in this field. But difficulties were encountered when it was attempted to bring the frequency up high enough for radio purposes with DUDDELL'S arrange- ment. Hence, after a long series of unsuccessful attempts it was con- cluded that it is impossible to obtain undamped oscillations of frequencies above 100,000 cycles per sec., with DUDDELL'S arrangement using carbon electrodes for the arc. However, WERTHEiM-SALOMONSON 202 disproved this theory by obtaining oscillations at 400,000 cycles per sec., using DUDDELL'S method. But the amount of energy which could be drawn from these oscillations was so slight, that his results could not yet be considered a practical solution of the problem of generating undamped oscillations for radio-telegraphic purposes. b. This problem was first solved by PouLSEN. 203 He soon showed that the arrangement of Fig. 244 would give undamped oscillations at radio-frequencies and sufficient energy, if modified as follows: 1. The gap (or arc, F, Fig. 244) is placed in hydrogen or a gas contain- ing hydrogen. 2. The positive electrode of the gap (or arc) is of copper, preferably cooled by circulating water, retaining carbon only for the negative elec- trode. 3. A magnetic field is caused to act upon the arc (magnetic blow-out). Furthermore, in order to improve the regularity of the oscillations, which is of great practical importance, we should add another require- ment, viz., 4. One of the electrodes (the carbon) is slowly revolved about its axis. The Poulsen arrangement ("Poulsen generator," "Poulsen arc") is therefore in principle that shown in Fig. 265, in which, however, the nec- essary auxiliary apparatus for rotating the one electrode is omitted. The two iron cores with direct current flowing through their windings provide the magnetic blow-out. 222 WIRELESS TELEGRAPHY The requirements as given by Poulsen are not of equal importance. A hydrogen atmosphere in the arc, perhaps in conjunction with the par- ticular materials chosen by him for the electrodes, suffices to secure the high frequency needed for radio-telegraphy with sufficient regularity of the oscillations. The magnetic field is required only and apparently is even then not absolutely essential if a very great amount of energy is wanted from the condenser circuit. c. The TELEFUNKEN Co. 204 arrived at a somewhat different solution of the problem through tests made at the sugges- tion of H. TH. SIMON. The Telefunken "high fre- quency lamp" is character- ized by the following points (Fig 266) : 1. As in the FIG. 265. POULSEN the negative arc is of generator, electrode at the carbon, the positive, of copper, the latter being water-cooled. 2. The arc burns in a hollow in the copper electrode, hence in the gases or vapors produced by the arc.* 3. A number of such arcs are joined in series. Fig. 267 illustrates the construction of one of these lamps; this form was used for a time in connection with wireless telephony, but is no longer in use. 205 124. Commercial Construction of the Poulsen Generators.! a. Figs. 268 (C. LORENZ Co.) and 269 206 show the earliest construction of the POULSEN generator with transverse magnetic field, according to the dia- gram of Fig. 265. The part con-' taining the heavy cooling vanes which is known as the "flame chamber" encloses the two horizontal electrodes which can be brought together for an instant at starting by means of a lever arm (at the upper right of Fig. 268). The large coils with their iron cores furnish the horizontal magnetic field across the inside of the flame chamber, and the small motor serves to revolve the carbon electrode. b. A second, considerably different form of construction used for radio- * Similar to the burning of flaming arc-lamps. t The credit for developing the construction of the POULSEN generators rests with the former AMALGAMATED RADIO-TELEGRAPH Co. (in particular with MR. RAUSCH YON TRAUBENBERG 207 ) and the C. LORENZ Co.'s telephone and telegraph works. FIG. 206. UNDAMPED OSCILLATIONS BY THE ARC METHOD 223 FIG. 268. 224 WIRELESS TELEGRAPHY FIG. 269. telephony and wherever there is no need of great amounts of energy, is shown diagrammatically in Fig. 270. The arc is vertical, the copper electrode, which is formed with large vanes (R), is at the top and the carbon electrode, which is made of a short piece of homogeneous carbon, is at the bottom. The magnetic field is produced by a single winding, S, having a vertical core, Eij and the course of the magnetic lines of force is guided by means of an iron ring, EZ, at the end of the copper electrode. The effect of the magnetic field is to cause the arc to move slowly about in a circle. An actual construction of this form of POULSEN generator is shown in Fig. 271. The principal object attained by the mag- netic field of this second form is that the arc is constantly moving about from point to point, so that rotating the electrodes becomes super- fluous. The disadvantage of this form, however, is that this arrange- ment makes it impossible to secure the same magnetic field strengths or UNDAMPED OSCILLATIONS BY THE ARC METHOD 225 to use them to their best advantage, so that it is not possible to obtain as high energy in the oscillations as with the form having a transverse magnetic field.* c. The hydrogen atmosphere was formerly obtained by causing a stream of hydrogen gas to flow through the case of the POULSEN generator. The hydrogen was either taken 1 directly from tanks, as marketed, or chemically prepared in special apparatus by the decomposition of water. The method lately in common use is much simpler. A small feed cup similar to lubricating oil cups (see Figs. 268 and 269) is located over the case and is filled with alcohol which continuously drips into the flame chamber where it is vaporized. 125. Use of the Poulsen Arc for Measuring Purposes. 115 For measurements, maximum regular- ity of the oscillations, rather than a great amount of energy is the Gas fas FIG. 271. FIG. 272. essential, does not Therefore a transverse magnetic field is undesirable as it tend toward constant regularity [Art. 136c]. Moreover * The KNOCKROE 208 Station, which was operated with 10-15 kw. oscillatory energy, had a transverse field of 10,000 lines of force per sq. cm. The CuLLERCOATs 208 Station (5 kw.) was equipped with a POULSEN generator of the second form. The energy which can be drawn from the oscillations is claimed to be about 19 per cent, of the total energy supplied by the D.C. generator. 207 15 226 WIRELESS TELEGRAPHY it is very important that the condenser is not of too great capacity [Art. 135c]. a. For some purposes the simple form of lamp shown in Fig. 272 (F. KIEBITZ) suffices: P is a copper plate or disc cooled by water on top and K is an adjustable carbon electrode. Hydrogen bubbled through acetone is recommended for the atmosphere in the arc chamber. FIG. 273. 6. The Physikalisch-technische Reichsanstalt has designed a POTJLSEN" generator for measuring purposes which gives particularly constant oscillations, but very little energy. "In this lamp the arc burns between a cooled outer copper cylinder of 23 mm. inside diameter and 30 mm. high and the surface of a homogeneous carbon, 22 mm. thick. A magnetic field in the direction of the axis of the carbon and whose strength is adjustable, keeps the arc in constant rotation, thereby prevent- ing the carbon from burning off unevenly. Three screws at the ends of the carbon serve for centering it with respect to the copper tube and are so arranged that this adjustment can be conveniently made even while the lamp is burning. The UNDAMPED OSCILLATIONS BY THE ARC METHOD 227 magnetic field is produced by a coil through which the arc current flows. The current is furnished by a storage battery at 240 volts. A suitable series resistance can be conveniently made from Nernst iron resistors." c. Pig. 273 shows a POULSEN lamp of the C. LORENZ Co., designed espe- cially for measuring purposes. It is provided with an automatic regulat- ing device (which can be seen under the lamp proper in Fig. 273) for the arc and is claimed to give very constant oscillations for long periods. 126. Circuit Connections of the Poulsen Transmitter. a. Coupled Poulsen Transmitter. In the first period following the discovery of POUL- SEN, the same method of working as in the BRATJN transmitter was used, probably universally, i.e., the POULSEN generator was connected into a condenser circuit and the antenna coupled thereto. To be sure, the condenser circuit of the POULSEN transmitter was quite different from that used by BRAUN. Of the requirements for the POULSEN circuit, viz., 1. Lowest possible damping, and 2. Comparatively little capacity and large self-induction, the latter is in direct contrast with those of the BRAUN transmitter, where the capacity is chosen as great as possible [Art. 106d]. The first require- ment caused the use of air or oil condensers, to prevent the loss due to dielectric hysteresis which occurs in solid dielectrics. Moreover, the use of air or oil condensers involves no such difficulties with undamped os- cillations as with the BRAUN transmitter, as in the former much lower po- tentials (at most a thousand, usually only a few hundred volts) and much less capacity* are used. The coupled POULSEN generator is still in use for wireless telephone work (Chap. XIV), in exceptional cases also for wireless telegraphy. The coupling between the primary circuit and antenna was inductive and loose in the POULSEN station at KNOCKROE. 208 Occasionally, how- ever, very close coupling was used. A medium degree of coupling is said to be undesirable, as this tends to make the frequency jump back and forth between two limiting values. f b. The uncoupled Poulsen transmitter. If antennae of relatively large capacity are used and coils of considerable self-induction are in- serted in these, then the ratio of supplied to useful (converted) energy and of capacity to self-induction are about the same for a POULSEN generator *The capacity of the POULSEN station at KNOCKROE, 208 intended for transatlantic service was only 0.03 mf ., while the BRAUN transmitter at NAUEN had 0.4 mf . and the MARCONI station at CLIFDEN had 1.6 mf. (air condensers) capacity [Art. 108a]. f It is usually stated that first one, then the other "of the two coupling waves'* appears. The author is not aware, however, whether it has ever been proven that the two oscillations, which are apt to occur alternately in the POULSEN transmitter, are identical with the two oscillations which occur simultaneously in the coupled transmitter producing damped oscillations. 228 WIRELESS TELEGRAPHY as for a normal condenser circuit. In this case, therefore, there is nothing to be gained by coupling the antenna to a condenser circuit and it is cus- tomary to connect directly into the antenna, which results in a particu- larly simple arrangement. If the antenna capacity is relatively small the arrangement of Fig. 205 [Art. 986] the POULSEN arc being placed between A andE is of ten used. This is frequently referred to as the "fly-wheel connection/' 209 127. Devices for Producing Signals. a. For telegraphing with damped oscillations a key which alternately makes and breaks the circuit is sufficient [Art. 116]. For undamped oscillations this is not so simple, for the following reasons: The distance between the electrodes which is the most favorable for the production of the oscillations, is generally greater than the gap length which the dynamo voltage would jump across and form an arc. Hence it does not suffice to simply close the supply circuit by a key in order to ignite the arc. This could be overcome in two different ways. Provision could be made by means of properly connected condensers, inductances and also transformers for producing a higher potential sufficient to form the arc, whenever the supply circuit is closed. Or again, the key could be so arranged that whenever it is closed the electrodes are brought into con- tact with each other or very close together. Both methods, however, have a great fault. It is comparatively difficult to keep the frequency and amplitude of undamped oscillations constant. Hence it is of the greatest importance to leave those condi- tions, which affect the oscillations, unchanged. It is evident that if the supply circuit is continually opened and closed, it becomes practically impossible to obtain fixed conditions and oscillations of the requisite constancy. It follows that in all devices intended for sending telegraphic signals, i.e., for alternately transmitting and suppressing the waves, provision must be made for a minimum effect upon the oscillations. b. The following are but a few of the many more or less successful arrangements which have been proposed for this purpose. The arrangement of P. O. PEDERSEN, 210 illustrated diagrammatically in Fig. 274, is intended for use in the coupled POULSEN generator and seems formerly to have been used in all the POULSEN stations. If the left end of the key, T, is pressed down, the aerial is connected to the coil /S 2 . Upon releasing T, the condenser circuit, SzCLR, is completed and oscilla- tions induced in it by the primary circuit, 7. The capacity self-induction and decrement of the condenser circuit, S 2 CLR, are made the same as the corresponding values of the aerial. Hence the primary circuit finds exactly the same conditions in the secondary in either position of the key. UNDAMPED OSCILLATIONS BY THE ARC METHOD 229 For the case of direct connection of the antenna to the POULSEN generator, the C. LORENZ Co. 210 proposes the insertion of an iron resist- ance in series with a condenser circuit* in the antenna and short-circuiting both by means of a key when telegraphing. c. In PEDERSEN'S ISI apparatus for rapid telegraphing, a small portion 1 R FIG. 274. of the inductance inserted in the antenna is usually short-circuited so that the waves radiated by the antenna are somewhat shorter than the natural wave-length of the receiver. When signals are to be trans- mitted, this short circuit is then opened and the receiver responds to the waves. 128. The Multitone Transmit- ter of C. Lorenz. 211 Aside from the direct application of un- damped oscillations to radio-teleg- raphy, W. BURSTYN proposed the use of the undamped oscillations of a condenser circuit the "tone circuit" of relatively low fre- FlG 275 quency, for affecting the spark of a quenched gap circuit by the particular period of this tone circuit, so * The condenser circuit is connected into the antenna like that (ALBCA) shown in Fig. 205. The condenser circuit causes an increase in the wave-length of the oscilla- tions, whose amplitude is decreased because of the energy consumed in the added iron resistance. 230 WIRELESS TELEGRAPHY as to produce a tone of this period (or frequency) in the receiving telephone. (This is called a tone transmitter.) Fig. 275 is a sketch of the connections. CLTF is the quenched spark gap circuit. The supply circuit, LoR Q , is fed by the direct-current gen- erator, M. The tone circuit, CiLi, consisting of a large condenser, Ci, and an inductance, LI, is connected in parallel to the spark gap, F. This condenser circuit oscillates (undamped) and the effect is about the same as if the direct current supplied by the dynamo had an alternating cur- rent (as from an alternator) of the same frequency as that of the tone circuit superimposed upon it. FIG. 276. The C. LORENZ Co. has specialized in the construction of this ar- rangement under the name of "multitone" transmitter (Fig. 276). The condenser C of the diagram (Fig. 275) is the mica condenser seen at the lower left-hand corner of Fig. 276, and the spark is that shown in Fig. 233 [Art. 11 Id], The coil LI is built with an iron core and can be seen back of the spark gap in Fig. 276. The electric constants are so chosen that the discharges of the circuit ~CLTF (Fig. 275) are of the form dis- cussed in Art. 109e, i.e., we have a case of impulse excitation. By means of a keyboard (on the top of the case in Fig. 276) various numbers of turns of the winding LI (Fig. 275) can be chosen, so that the frequency of the tone circuit and hence the tone in the receiving tele- phone can be very easily varied in this way. This simple choice of tone UNDAMPED OSCILLATIONS BY THE ARC METHOD 231 forms an advantage of this method as compared to tone production by means of an alternator, although the latter of course offers a far greater range in the amount of energy used. 2. STUDY OF THE ACTION* OF THE ARC 129. Characteristic of the Arc. Under the term characteristic of the arc (or of some other current carrying conductor) we understand a curve whose abscissa are proportional to the current in the arc and whose ordinates are proportional to the potential difference between the electrodes. a. Experimental Determination. The direct-current characteristic (so-called "static characteristic") is obtained by simply measuring the current with an ammeter, the potential with a voltmeter and then plotting the values so ob- tained in curve form. If, however, the current varies with time as, e.g., an alter- nating current, the characteristic may be found by means of the " BRAUN tube, used as shown in Fig. 277.f The curve over which the spot on the screen of the BRAUN tube moves, is the character- istic for that particular variable current ("dynamic characteristic"). b. The Static Characteristic of the Arc. 107 In Art. 96, it was shown that with direct current, within certain limits the voltage across the arc, V, in terms of the current, /, is where a and b are constants. Hence, the characteristic is an equilateral hyperbola (Fig. 278). It is said to be a "falling" characteristic, as an in- crease in the current corresponds to a decrease in the voltage. For very large currents V = constant = a * The explanation of what takes place in the arc method is due primarily to W. DUDDELL, A. BLONDEL, H. TH. SIMON and H. BARKHAUSEN. 212 There seems lately to have been a widespread impression that the work of these investigators effected POULSEN'S invention, i.e., as if POULSEN had simply drawn more or less evi- dent conclusions from existant theories. This, however, is an anachronism. POULSEN applied for his patents in 1902 and 1903, i.e., 2-3 years previous to any of the theo- retical work which might come into consideration. t CiC 2 are small plates for the purpose of electrically deflecting the cathode rays; A is the conductor whose characteristic is being determined. 232 WIRELESS TELEGRAPHY For very small currents, the equation given above does not hold, par- ticularly when / = 0, V does not become infinity, but V = V z i.e., equal to the discharge, "ignition, " or breakdown potential which is just sufficient to jump across the gap. In Art. 42 it was shown that this value depends upon the shape of the electrodes and the distance between them, and also upon the nature of the gas in the gap. It is far greater than the potential which exists across the electrodes of the arc, while the latter is burning with even moderate intensity* [see Table V]. FIG. 278. FIG. 279. c. The dynamic characteristic for alternating current has the shape of the curve shown in Fig. 279. The following important points about it are noticeable : 1. The value of the potential corresponding to a given current value is not the same when the current is increasing as when it is decreasing. Also there is a phase displacement between the current and the voltage; the latter is not at its maximum at the same instant as the current, f 2. The discharge potential, V z , i.e., in this case, the potential at which * If the distance between the arc lamp carbons is ^ mm., then V z is in general more than 1000 volts, while the potential, at the time the arc is burning, is of the order of 50 volts. f As these relations are very similar to those which exist between magnetic force and magnetic field strength in iron, H. TH. SiMON 212 has given the phenomenon the name of "arc hysteresis." These phenomena are closely related to the fact that the number of ions existing between the electrodes depends upon the current and the temperature of the electrodes at the preceding instant, the number increasing as the current and electrode temperature increase. Hence with rising current the number of ions is less at a given current value than with decreasing current at the same current value, and the voltage necessary to produce a given current is greater in the first case than in the second for the same reason. UNDAMPED OSCILLATIONS BY THE ARC METHOD 233 the current passes through zero is comparatively very low, as the gaseous path remains ionized even after the current has disappeared. This is due to the fact that in an ionized gas a little time is always required before the ionization has entirely disappeared [Art. 65a], and also largely to the fact that the electrodes, as long as they remain in- candescent, emit electrons which tend to ionize the gas [Art. 42c]. 130. Type I Oscillations. 7i < 7 . Con- sider the arrangement shown in Fig. 280. The resistance, RQ, and the coefficient of self-induc- tion, 7/o, of the supply circuit are first chosen so great that neither the current in the condenser circuit, CLR, nor the conditions existing in the arc can have an appreciable effect upon the supply current, IQ, which may therefore be con- sidered as constant. In the case of type I oscillations, i.e., oscilla- tions in which the amplitude, 7i , of the alternat- ing current is less than the supply current IQ, the current curve* is of the form of the heavy full line curve shown in Fig. 281. It follows that we obtain an undamped, almost sinusoidal, alter- nating current in the condenser circuit with type I oscillations. The voltage across the arc is not sinusoidal, but varies about as shown by curve V in Fig. 289. The characteristic of the arc with these oscillationsf has the form of the heavy full line curve of Fig. 282. The values of the supply current, 7o, and of the D.C. voltage, VQ, corresponding to it| are shown as heavy dashed lines which divide the plane of the paper into four quadrants marked 7, 77, 777, IV. Now, not only for type 7 oscillations, but in all cases where a direct and alternating current are superimposed the following condition holds: As FIG. 280. A A W FIG. 281. long as the characteristic lies within the quadrants II and IV, energy is added to the alternating current, while when the characteristic * In Fig. 281 and the following figures the ordinates to the left represent values of /i, those to the right, of / (current in the arc = /o + /i). f And with homogeneous carbons and slow oscillations. } In the static characteristic. 234 WIRELESS TELEGRAPHY lies in the other two quadrants, 7 and ///, energy is taken from the alternating current. The diagram, however, gives no indication of the amount of the energy changes. 131. Type II Oscillations. / io > 7 ; No Re-ignition. As soon as the amplitude of the alternating current, Ji, becomes greater than the supply current, Jo, then during the half period in which /i, flowing through FIG. 283. the path AB (Fig. 280), has the opposite sign (direction) to that of / , the current/ = /i + /o in AB must = 0. Consequently the arc is extinguished and does not form again until the voltage, V, across the electrodes has reached the value of the breakdown potential, V z . a. Figs. 283, 284 and 285* represent a series of cases diagrammatically, under the assumption that the ignition of the arc takes place suddenly FIG. 284. FIG. 285. and that the voltage across the arc while burning, Vb, remains constant. Fig. 283 represents the case in which the current amplitude in the con- denser circuit, /!<,, is only slightly greater than the supply current /o; in * Figs. 283 to 287 and 290 are drawn from figures of H. BARKHAUSEN. 212 In these figures the full line voltage curve represents the voltage between the condenser coatings, while the heavy dashed curve gives the arc voltage. UNDAMPED OSCILLATIONS BY THE ARC METHOD 235 Fig. 284 7i is much greater than IQ. In both cases it is assumed that the damping of the natural oscillations of the condenser circuit is not appre- ciable (R is very small). The effect of having the natural oscillations more highly damped is shown in Fig. 285, which in all other respects rep- resents the same conditions as Fig. 283. In each period, T, there are two different portions, viz., the subperiod, Ti, called the "discharging stage," during which the arc burns (7 is not zero), and the subperiod, T 2 , the "charging stage," during which the condenser acquires its charge, the arc being extinguished and the current / = 0. During the first subperiod, TI, the curve of the current, /i, in the con- denser circuit is part of a sine curve* i.e., we have an ordinary alter- nating current. During the second portion, T 2 , the current is a direct current /i = IQ. The voltage, V, across the condenser coatings varies correspondingly: during 7\ it is oscillatory, during Tz it rises from the value F a , at which the arc was extinguished, to the value, V 2) at which it is again ignited, rising approximately in a straight line.f The voltage, F, across the arc falls abruptly from the value, V z , which it has at the moment of ignition, to the value, F&, which it has during the t \ ^4- ^ _^ *o J r o 7 +/]^/ V t 1 o /j, I FIG. 286. FIG. 287. time of burning and then remains constant during the entire time TI. During the time, T%, in which the arc is extinguished, and hence no cur- rent is flowing through the arc, the voltage V is practically identical with V c (voltage across condenser coatings) ; only with relatively high re- sistance, Ri, and the consequent high damping, does V differ somewhat from F c . b. The assumed conditions governing Figs. 283, 284 and 285, would give an arc characteristic of the form of Fig. 286. Actual experimental * With decreasing amplitude, if there is any damping. f It is assumed that 7 = const. The actual form of the charging curve depends upon the capacity of the condenser, the resistance, R , and the self-induction, L (the dynamo voltage being assumed constant). If L is very great then the charging curve is almost a straight line, 213 while if R is very large and L is very small, the curve is a more or less straight portion of an exponential curve. 214 236 WIRELESS TELEGRAPHY observation, however, produces the form shown in Fig. 287. It follows therefore that the assumptions made in a are not quite correct. The ig- nition is not so sudden and the voltage does not fall abruptly from V z to Vb, nor does it remain entirely constant while the arc is burning, but rises slightly just before the arc is extinguished. Hence the variation A A / V . / \ / ^ .^ / -** - -1 -_>^. 1^ . -f / / FIG. 288. FIG. 289. of voltage must be about as represented by the curve of Fig. 288. This compares well with the curve of Fig. 289 which A. BLONDEL 212 determined experimentally. There is another point in which the actual facts differ from the as- sumptions made in a, according to which the voltage V c , across the con- denser coatings could not rise above the terminal voltage of the dynamo. As a matter of fact it may under certain conditions rise to a much higher value. That this may be possible is understood if we consider that a change in Jo the previous assumption that /o is constant is not entirely correct may produce higher potentials be- cause of the self-induction, Z/o, by adding to the voltage across the con- denser terminals [see Art. 1156]. Whether this is always the sole ex- planation is a question which need not be further investigated here. 132. Type III Oscillations /i > 7 ; Re-ignition Present. Fig. 285 shows that at the moment the arc is extinguished, the voltage across the electrodes jumps from the normal value Vb to the value V a . V a is not as high as the ignition voltage V z , which is just sufficient to start the arc at the end of the charging stage, T 2 . Under certain conditions in fact, the gas between the electrodes may still be so largely ionized immediately after the arc is extinguished, that a much lower voltage, e.g., V a , suffices to at once re-ignite the arc ("re-ignition"). FIG. 290. UNDAMPED OSCILLATIONS BY THE ARC METHOD 237 If this is the case, then the oscillatory discharge of the condenser con- tinues, until finally the voltage V a becomes too low to maintain the arc which is then extinguished. Hence, we obtain oscillations of the form of Fig. 290 or 291. The latter form is practically a representation of a rapid sequence of the natural oscillations of the condenser circuit. It is noth- ing more nor less than the form of oscillation whose practical application was discussed in Chap. VII.* FIG. 291. 133. Energy of the Oscillations. a. Type I Oscillations. Experience has shown that these cannot be produced so as to give great energy by such means as are known, the difficulty of obtaining high power increasing as the frequency becomes higher. b. Type II Oscillations. Under the same assumptions upon which Fig. 285 was based (7o = const., V = const, and during time arc is burn- ing V = Vb) } the energy which is supplied to the oscillation during one period by the direct current, and hence the maximum which can possibly be drawn from the oscillation, is approximately R 2L Ti - e where C, R and L are the capacity, resistance and coefficient of self-in- duction of the condenser circuit. Hence with R and L and also Vb con- stant, the energy increases very rapidly with increasing ignition voltage. c. Type III Oscillations. In the pure form of these oscillations (Fig. 291) we deal practically with the natural oscillations of the condenser * The natural oscillations of a condenser circuit discussed in Chap. I are also practically the same as those described here. The difference is merely that in the case mentioned in Chap. I, the supply current is not constant or even nearly so, but varies widely with the time, being furnished from either an induction coil or an A.C. transformer. 238 WIRELESS TELEGRAPHY circuit. The energy which is transferred in one discharge, is approxi- mately [Art. 66]. at the same time the highest voltage, V z , which occurs across the con- denser, may, under certain circumstances, be greater than the dynamo voltage (see Art. 1316). The energy consumed per second by the oscilla- tions is r . \ cvs where is the discharge frequency. This depends upon the rapidity with which the condenser becomes fully charged again after discharging; for a given capacity, it increases as the supply current, /o is increased. 134. Frequency of the Oscillations. 215 a. The frequency of type I oscillations is determined partly by the self-induction and capacity of the condenser circuit, partly by the characteristic of the arc. The frequency is always somewhat lower than the theoretical value of the natural fre- quency of the condenser circuit as obtained by THOMSON'S equation from the known values of the coefficient of self-induction and the capacity, but the difference is never very large. 6. In type II oscillations the period T consists of two parts, T\ and T 2 . The length of the discharging period, TI, is determined first of all by the period of the natural oscillations of the condenser circuit and the ratio 7i :/o; secondly, by the damping of the natural oscillations (see Figs. 283 and 285). The second or charging period, T^, is the interval from the time the arc is extinguished to the time it is again ignited. The period of the oscillation, T = TI + T 2 , can therefore not be found even approxi- mately by means of THOMSON'S equation, as it depends materially upon the rapidity with which the condenser becomes charged, i.e., upon condi- tions in the supply circuit. A consideration of practical importance is that not only the amplitude but also the length of the period and hence the frequency varies if there is any slight change in the voltage at which the arc ignites. This is what gen- erally happens as soon as there is the least change in the electrodes. The extent of the change in 7 7 2 depends largely upon the manner in which the voltage rises to the ignition point after the arc has been extinguished and upon the manner in which the voltage, V, across the electrodes rises. * c. For the pure type III oscillations^ (Fig. 291), practically the same * At the points where the V curve (Fig. 283 et seq.) cuts the "ignition character- istic" (abscissae time, ordinates a V z ) the V curve must be much steeper than the ignition characteristic so that ignition always takes place promptly. The ignition characteristic becomes steeper, the more rapidly the ionization of the gas disappears. f Type II oscillations of the kind shown in Fig. 290 are in general entirely irregular and quite useless. UNDAMPED OSCILLATIONS BY THE ARC METHOD 239 may be stated as for the natural oscillations of condenser circuits pro- duced by an induction coil or similar device [Chap. I]. The effect of the arc on the period* is not appreciable, the frequency, therefore, is constant and determined by the self-induction and capacity from THOMSON'S equation; as long as the distance between the electrodes is at least 2 mm. If this distance is very small, so that the deionization becomes very rapid, then in this case also, the frequency may be considerably lower than would be expected from THOMSON'S equation [Art. 5c]. 135. Practical Conclusions! for Type II Oscillations. Type / oscil- lations, because of their low energy are of no practical importance. Only Type II oscillations are used for radio-telegraphy with undamped oscillations. In practice it is important to give the oscillations as much energy as possible % and to keep the frequency as nearly constant as possible. a. The requirement of maximum energy leads to the maximum igni- tion voltage [Art. 1336]. This can be provided for in two ways, viz., 1. By lengthening the charging stage, T 2 , as much as possible, so as to give the gas plenty of time to deionize. 2. By the use of special means for rapid deionization of the gas. The first method involves the danger of destroying the constancy of the frequency [Art. 136c]. Moreover, the longer T 2 is made, the more does the current curve tend to differ from the sinusoidal form (see Fig. 284), i.e., the upper partial oscillations come into prominence in addition to the fundamental. The energy of the partial oscillations is wasted, however, for in practice, when coupling or when using a tuned receiver, only the fundamental oscillation is effective. As a matter of fact, therefore, it is best to work with oscillations in which the subperiod, T 2 , is relatively short, and in which, therefore, / lo is not much different from 7o (Fig. 285). b. Then, however, it is especially important to obtain a very rapid 'growth of the ignition voltage by special means, i.e., to deionize the gas in the path of the arc as rapidly as possible. Necessary precautions for this result are as follows: 1. Removal of the ionized gas from the space between the electrodes. The spontaneous deionization of the gas in the path of the arc, due to the ions recombining, is in general too slow to be effective at the high * That is, the period of the damped oscillations (T in Fig. 291) which come into consideration for practical use. f Strictly speaking, conclusions may be drawn from what has preceded only if the condenser circuit is not coupled to some other circuit. If it is loosely coupled to another circuit, the conditions will presumably change but very little, but with close coupling they will change very much. Systematic investigation of the coupling of Type // oscillations has to date been made only at low frequencies (S. SuBKis 970 ). J The important thing, of course, is to take as much energy as possible from the oscillations. 240 WIRELESS TELEGRAPHY frequencies involved. Moreover, as the electrodes are not so very close together as with quenched spark gaps, deionization by absorption at the electrodes and by an electric field cannot amount to much. Diffusion into the outer space is far more effective, particularly if the coefficient of diffusion of the gas in question is high; hence hydrogen, having the highest coefficient of diffusion, gives the best results. The most effective means of removing the ions in the space between the electrodes is the use of a magnetic blowout, as this acts while the current is still flowing, driving the arc and the gas contained in the arc out of the innermost recesses between the electrodes. 111 The use of a mechanical air blower for this purpose hardly offers any advantages. To be effective the velocity of the current of air or gas blown must be such that each particle of the air moves through a distance of at least 1-2 mm. during the half period of an oscillation (at X = 1000 m., this would be 1.5 X 10~ 6 sec.). Such high velocities, however (600 to 1200 m. per sec. under the assumptions made), aside from the complications in- volved, would produce such eddies between the electrodes as to defeat the very object in view and make a complete removal of the ions in the path of the arc impossible, in spite of the high velocity. 2. Prevention of the ionizing effect of the incandescent electrodes, particularly of the anode. The following are various methods for pre- venting or at least reducing this effect. a. Cooling the anode, at which the development of heat is particularly great. Cooling the anode as a whole is relatively a simple matter. Water or air cooling, the latter preferably aided by ventilators or a ribbed con- struction of the anode, suffice. But it is much more difficult to prevent local heating at the point where the arc originates and at which the emis- sion of electrons continues after the charging stage. This detrimental effect can be mitigated by: a. Use of a metal having very high heat conductivity (as copper or sil- ver) for the anode.* /?. Surrounding the electrodes by a gas having very high heat con- ductivity; in this respect, hydrogen, which has the greatest heat conduc- tivity of all the gases, is best. 7. Hydrogen, moreover, has the advantage of preventing the forma- tion of metallic oxides and even reducing any previously existing oxides, which are particularly active in the emission of electrons when incandes- cent. This is also true to a certain extent of an enclosed arc-lamp. b. Rotating one or both of the electrodes tends to reduce local heating only if it is so rapid that the base of the arc is moved sufficiently far in each period as to occur at a point not yet materially heated in the succeed- * Homogeneous carbon is used universally for the cathode. This difference or asymmetry of the two electrodes is also of value in that it prevents re-ignition. UNDAMPED OSCILLATIONS BY THE ARC METHOD 241 ing period. This, however, would require a speed of rotation of a much higher order than is ever used. c. The subdivision of the arc into several partial arcs [Art. 123c] in series is often advantageous. With the same total voltage and the same current, about the same amount of heat is developed in all the partial arcs as in the equivalent single arc, but the heat given off is much greater in the combined partial arcs than in the single arc. The problem of deionizing the gaseous path of the arc during the charg- ing stage and keeping it deionized increases in difficulty, other things being equal, as the frequency of the oscillation is increased, the time avail- able for the deionization being correspondingly shortened, and as the cur- rent and hence also the heating of the electrodes and the number of ions formed are increased. Therein lies the main explanation of the relative ease with which undamped oscillations of low frequency and energy can be obtained, while for a long time no one succeeded in obtaining undamped oscillations of such frequency and energy as are needed in radio-telegraphy . c. Amount of Capacity Allowable. Let the dynamo voltage and the frequency of the oscillations be given. Then the energy supplied to the condenser circuit per second is proportional to the capacity in this circuit [Art. 1176]. Hence, from this standpoint, a larger capacity is advantageous. On the other hand a larger capacity necessitates a larger current amplitude, 7i , in the condenser circuit, and as this must not be much larger than the supply current, 7 [see a], the latter must also be larger. But the greater the current through the arc becomes, the more intense will be the heating of the electrodes and the ionization of the gas in the path of the arc and the less effective, therefore, will be the curative methods given in b. Consequently, indefinitely increasing the capacity soon becomes detri- mental to the best results, so that in generating undamped oscillations by the arc method we are obliged to work with relatively small capacity and large self-induction in the primary circuit. 136. Regularity of Type II Oscillations (K. VoLLMER 115 ). It is evi- dently extremely probable that the burning of the arc causes the electrodes gradually to change, much more so, in fact, than in an oscillating damped condenser circuit of low discharge frequency, in which the gap is without current the greater part of the time. Every change in the path of the arc, however, will alter the ignition voltage and thereby the frequency and wave-length, as well as the energy and amplitude of the oscillations. a. These fluctuations may be subdivided into the following classes : 1. Slight fluctuations in arcs without transverse magnetic field, these being either a. Rapid fluctuations, or (3. Slow changes. 16 242 WIRELESS TELEGRAPHY 2. Great fluctuations in arcs having transverse magnetic field and caused by this field. The cause for 1 in arcs with no transverse magnetic field is no doubt the following : The arc, while burning, eats its way into the electrode (or electrodes) thereby gradually lengthening the arc and increasing the igni- tion voltage, so that the frequency and amplitude of the oscillations are also gradually changed thereby. This continues until the arc finds more favorable conditions at some neighboring point, to which it then jumps with the result that the arc length, ignition voltage, frequency and am- plitude also take a jump in the opposite direction. (The individual de- FJG. 292. pressions or cavities which the arc had eaten out, on its way around the hollow cylindrical electrode are easily recognizable in the accompanying photograph, Fig. 292.) Tests have shown that changes in the wave-length and in the ampli- tude occur in conjunction with changes in the average arc potential, in fact the wave-length variation is directly proportional to the mean arc potential variation. Other things being equal, it (the mean arc potential variation) increases with increasing capacity, decreasing supply current and decreasing wave-length (increasing frequency). The extent of the fluctuations depends very largely upon the construc- tion of the lamp. 115 In a lamp made by VOLLMEU and copied from that of UNDAMPED OSCILLATIONS BY THE ARC METHOD 243 the Physikalische Reichsanstalt, he found that when the adjustment was particularly good, at X = 2000 m. the intensity variation was about 2 per cent., the frequency variation about 0.03 per cent., and at X = 700m. the frequency variation was about 0.18 per cent. b. The results of these fluctuations are disturbances which interfere with the practical application of the oscillations as well as with any measurements. The rapid intensity fluctuations which occur in lamps without a trans- verse field are harmless, as the measuring instruments or detectors used do not respond to them, but indicate the average value. The slow fluc- tuations, however, may at times be very annoying especially in connection with measurements. FIG. 293. The frequency fluctuations interfere particularly if the arc circuit is very loosely coupled to a secondary circuit, (1) by flattening the reso- nance curve, reducing the sharpness of resonance and, (2) by materially reducing the current effect at resonance. 1. In Fig. 293 the broken line curve is the resonance curve which ought to be obtained by the action of an undamped oscillation of constant fre- quency upon a secondary circuit whose decrement, di = 0.005. The full- line curve is the resonance curve obtained with the same secondary circuit, if the frequency (or wave-length) of the undamped primary circuit fluctu- ates back and forth at a uniform rate between the limits X + X' and X X', X 7 being only 0.05 per cent, of X. It will be noted that even with this small fluctuation the resonance curve suffers a considerable flattening and reduction in height from its ideal form. The resonance curves obtained from arcs with a transverse field are of the form of the full-line curve shown in Fig. 294 (d 2 = 0.012) ; here the difference from the dashed curve, which would be obtained with constant frequency using the same secondary circuit, is much greater. (The peaks of the two curves were drawn alike in height intentionally.) From the 244 WIRELESS TELEGRAPHY shape of the curve it is evident that the fluctuations which occur are not symmetrical with respect to a mean value, but (similarly to the brush discharge of condensers [Art. 86]) the frequency lies mainly near a certain value (corresponding to 2 = 1450) from which it gradually fluctuates to a lower value (corresponding to C 2 = 1480). 2. The changes in the current effect caused by the frequency fluctua- tions, may be quite considerable, as is shown by the following tabulation which is based on the assumption that the fluctuations are symmetrical, amounting to 0.03 per cent, on each side of the mean value to which the secondary circuit is tuned. d 2 X = 2000 X = 1000 X = 500 m. 0.01 3 per cent. 24 per cent. 63 per cent. 0.03 0.5 per cent. 4 per cent. 16 per cent. . 05 0.2 per cent. 1 . 5 per cent. 6 . 5 per cent. Qualitatively, therefore, the effect of the frequency fluctuations is the same as if the primary circuit had constant frequency but material damping. c. In Art. 125 it was already pointed out that a transverse magnetic field, which is very advantageous for the energy of the oscillations, is very disadvantageous for their regularity. * This is true not only of the trans- verse field, but more or less so of all agents which tend to increase the energy. The explanation of this fact is simple enough. The require- ment for maximum energy is the most complete deionization of the gase- ous path of the arc during the charging stage, while on the other hand a slight ionization or electrification of the gaseous gap is advantageous, in fact is essential with low potentials [see Arts. 426 and 78c] for a sure and accurate timing of the discharge. Hence we must fall back upon a com- promise. This explains in part the use of carbon as the negative electrode, in spite of the fact that its low heat conductivity lowers the ignition volt- age. This also explains why the strength of the transverse magnetic field is in general not made very great, sacrificing a further increase in the energy of the oscillations. Somewhat of an exception to this rule is encountered in the use of hydrogen, one of whose properties tends greatly to increase the regularity of the oscillations; namely, the relatively low breakdown potential of hydrogen corresponding to a given gap length [Art 42c]. The consequence thereof is that with a given voltage (ignition voltage) the distance between the electrodes can be made considerably greater when hydrogen is used than, for example, with air. Hence any change in the arc length (say due to eating away or volatilization of the electrodes) amounts to a lower per- * The extent of the fluctuations depends upon the construction of the lamp in this case also. Good regularity can be obtained with lamps having a transverse magnetic field (see Art. 1916), but this is a much more difficult attainment than with lamps having no transverse field. UNDAMPED OSCILLATIONS BY THE ARC METHOD 245 centage of the initial length with hydrogen than with air, so that the re- sultant change in the potential and hence also in the wave-length and in- tensity is less than with air. 137. The Terms "Spark" and "Arc." 217 Doubt has occasionally been expressed of late as to whether the phenomenon in the gap consti- tuted a " spark" or an "arc" in a given case. In the two limiting cases there is never any doubt. Everybody speaks of " sparks" when, as in the early construction of the BRAUN transmitter, only a few, say ten to twenty discharges per second occur. Here the periods during which current flows through the gap are separated by long in- tervals of currentlessness, so to say, and the total time during which there is no current in the gap is much greater than the total time during which current flows through the gap. Here, then, both eye and ear receive evi- dence of intermittent discharges (limiting case 7). Again, everybody would call the phenomenon obtained in the gap between the electrodes, with undamped oscillations of type / or II, an "arc." In type / the gap is never without current, in type // the periods with and without current alternate so rapidly that neither human sight nor hearing can distinguish between them (limiting case //). Between these two limiting cases, however, are various intermediate forms, e.g., the case of damped oscillations at a very high discharge frequency. Here the total time of currentlessness becomes about equal to or even less than the time of current in the gap; at any rate the eye can here no longer distinguish the individual discharges and the ear at best can discover the presence of intermittent discharges only in the tone or note emitted by the gap.* Whether, in this case, we speak of sparks on the basis that the form of the discharges is inherently the same as in lim- iting case /, or whether we speak of an arc, in view of the fact that in lim- iting case II, the duration of current is of the same order as the duration of currentlessness, is a matter of individual preference. At any rate, it is advisable in such a case to obtain by actual test (e.g., by means of a discharge analyzer or a BRAUN tube) an exact picture of the time varia- tion of the discharges, rather than to dispute the propriety of the name given to the phenomenon in question. * In scientific and patent literature it is often claimed for some particular device or arrangement that it will generate undamped oscillations. If such a claim is based solely upon the arc-like appearance or sound in the gap, it must not be accepted with- out further conclusive evidence. CHAPTER X PROPAGATION OF THE WAVES OVER THE EARTH'S SURFACE 1. OVER PLANE OR SPHERICAL HOMOGENEOUS GROUND 218 138. Ground Having Plane Surface and High Conductivity.* Of these two assumptions, the latter is approximated in sea water. Under both assumptions we would have the following conditions. mi i i I'liiiri MIII | i mm \\\\\ \ \\\\\\\ \ ! \\\VM\\\\\\ V; \\\\\\\ V\\ A\\ \ \\X\\X \\\\\\ *' \ \ g x \ \\\\\\> *" vvX FIG. 295. a. General Nature of the Field. With a transmitter placed above the earth's surface, the following rule gives approximatelyf the form of the waves. Consider the ground removed and replaced by the image of the antenna, with respect to the earth's surface, so that the antenna and its image form two symmetrical halves. It is also assumed that the distri- * That is the specific conductivity > 10~ 12 c.g.s. units. f The results would be absolutely exact if the conductivity of the earth's surface were infinitely great. 246 PROPAGATION OF THE WAVES OVER THE EARTH'S SURFACE 247 bution of current and potential is the same as exists in the symmetrical halves of the antennae shown in Figs. 23, 42, 45 et seq., i.e., at any point, P, and its image, P', the current must have the same direction but the po- tential is of opposite sign in each. The rule then is : The waves which the antenna with its image would radiate, if they were placed in free space, have the same course through the air as the waves which are actually radi- ated by the antenna placed above the earth's surface. 219 b. Effect of the Form of the Antenna. From the preceding it follows that for a simple antenna (single straight wire), the electric field would be FIG. 296. as shown by the upper halves of Figs. 295 and 296 [see Art. 20a], the former representing the instant of maximum charge, the latter, the instant of maximum current. As the distance from the antenna increases, the electric lines of force approach more and more the form of circular arcs. In Art. 20a it was stated that the magnetic lines of force are also circular. With other forms of antenna the shape of the wave in the vicinity of the antenna, up to distances of one or two wave-lengths, may be considerably different from that just described, though the general character of the field, particularly the snapping apart of the lines of force must be more or less the same in all forms of antennas. 220 The greater the distance from 248 WIRELESS TELEGRAPHY the point of origin becomes, the more will the shape of the waves resemble that produced by a simple antenna. C. The Field at very great Distances from the Antenna. From Arts. 20a and 25a, the following may be concluded in regard to the field imme- diately above the earth's surface, at distances very great in comparison to the wave-length: 1. The direction of the electric lines of force is approximately perpen- dicular to the earth's surface, that of the magnetic lines parallel to it; both are perpendicular to the direction in which the wave moves. 2. The electric and the magnetic fields are in phase [Art. 20d\. 3. The amplitudes of the electric and magnetic field strengths are ex- pressed by _ cA . |7.| 3 X 10 10 c.g.s. units 1^0 1 amp. X r cm . cms. ah |7o| = 4-jr -r- - - - c.g.s. units A T (i) where h represents the height of the antenna, / the current amplitude at the current anti-node of the antenna, a the form factor of the antenna and r the distance from it. Accordingly, the amplitude of the field at great distances is inversely proportional to the distance r. d. Penetration of the Waves into the Ground. As the waves spread out over the earth's surface, they penetrate to some extent into the ground, but in so doing their amplitude is rapidly decreased. Thus in sea water of good conductivity,* at a depth of 1 m. the amplitude is only about one- tenth of its value above the surface, with a wave-length of about 700 m.* 139. Over Flat Ground of not very High Conductivity (A. SOMMER- FELD). If the earth's surface at the location in question has relatively low conductivity, as is the case even with fresh water, but particularly with dry ground, f the results are quite different, the change increasing as the conductivity and the dielectric constant of the ground decrease. 221 The rule given in Art. 138a for the construction of the field then no longer applies at all. The appearance of the field in the vicinity of the trans- * Specific conductivity = 5 X 10~ n c.g.s. units. The amplitude, A, at a depth Z, is of the form A = A e rz (A = amplitude at the surface). t For qualitative consideration the specific conductivity, a and dielectric constant, fc, may be assumed to be as follows : 223 Sea water the amplitude therefore .- in the latter the energy c -, the amplitude = The fact that in the latter there is a decrease in the Vr energy as the distance increases, in contrast to the wave following a wire and in addition to and entirely aside from such absorption as occurs is explained by the fact that the energy is spreading itself out over ever- increasing circles, as the wave travels its course. Absorption of course occurs in addition to this reduction in ampli- tude due to the expansion of the wave in space. As each wave advances through the air it is accompanied by a wave in the ground. And as the ground always has more or less conductivity, the moving electric field, constituting the wave, results in the formation of currents, just as in the wires of the LECHER system. These currents consume energy, which is drawn from that of the waves radiated by the antenna, so that an absorp- tion occurs in this way. c. While at short distances from the transmitter, the waves are al- * The direction of the flow of energy is, as already stated previously, perpendicular to both the electric and magnetic field directions. 250 WIRELESS TELEGRAPHY most entirely of the nature of space waves, as the distance increases the surface component becomes more and more predominant, as its amplitude decreases more slowly than that of the surface component. That is, the nature of the wave constantly approaches that of a surface wave.* This change is the more rapid, the shorter the wave-length is and the lower the conductivity and dielectric constant of the ground are. A cal- culation of the distance at which the actual amplitude of the wave differs by 10 per cent, from the amplitude of the space wave, results in the fol- lowing figures: Sea water f X = 2 km. Distance = 20,000 km. approx. Sea water X = 1 km. Distance = 5000 km. approx. Sea water X = 0.3 km. Distance = 500 km. approx. Fresh water f X = 2 km. Distance = 4 km. approx. The distance becomes still shorter with dry ground. Hence, while with sea water for all distances which come into consider- ation 20,000 km. is half the circumference of the earth and for all wave-lengths over 1 km. the waves have the characteristics of space waves, J with fresh water and even far more so with dry ground, they assume the characteristics of surface waves at distances of only a few wave-lengths or even less than one wave-length. Hence the nature of the wave propagation in this case must not be conceived as being the same as that described in Art. 138 over sea water. d. The subdivision of the wave into a space wave and a surface wave and a wave within the ground [b] makes it possible to give a simple description of the phenomenon. Physically, there is of course only one single wave extant, which travels partly through the air, partly through the ground along its upper surface. The appearance of the electric lines of force of this wave in air at a given instant and a distance of 30 to 30.5 wave-lengths from the trans- mitter is shown diagrammatically in Figs. 297 and 298, which are taken from an article by P. EPSTEIN, 225 the assumption being that the wave- length is 2 km. and that the conductivity of the upper stratum of the * When the distance becomes very great, the surface wave may again give way to the space wave, as the former is more rapidly absorbed. It is questionable, however, whether this effect is of practical importance. f On the assumption that a = 10~ n c.g.s. units for sea water and 10~ 14 c.g.s. units for fresh water. J Herein lies the justification for the statements in Art. 138. The electric and magnetic field strengths in this case, taking consideration of the absorption, are given by [Art. 138c]: E = 47T ~ 1/ | 3 X 10 10 c.g.s. units A T , <**> r i ~ ^ MQ = 47T - l/o I c.g.s. units A / in which /3 is the coefficient of absorption. PROPAGATION OF THE WAVES OVER THE EARTH'S SURFACE 251 30.0 30J +1* y/////////////////////////////////////^^^ 3O.25 FIG. 298. W, 30L5 252 WIRELESS TELEGRAPHY ground is about midway between that of sea water and wet ground. The scale of the ordinates (heights above ground) is one-twelfth of that of the abscissae (distances from transmitter) in these figures. By way of comparison, Fig. 299 represents the lines of force which would correspond to an infinitely great conductivity of the ground, ac- cording to Art. 138, the same scale and distance being used as in the pre- ceding figures. It will be noted that there is no very great difference between Fig. 299 and the other Figs. 297 and 298; the latter, however, are based on the assumption of relatively high ground conductivity. With dry ground the differences would be much more marked. 30.25 FIG. 299. 30.5 e. The nature of the field of the wave immediately above the earth's surface at very great distances from the transmitter, is of special practical interest. If the earth's surface were as good a conductor as a metal, then [Art. 138] 1. The electric field would be exactly perpendicular, the magnetic field parallel to the earth's surface, and 2. Both would be in phase [Art. 137]. As a matter of fact these conditions are approximately true over sea water, but they do not hold for fresh water or dry ground (J. ZENNECK 22 ). However, the direction of the magnetic lines of force remains parallel to the earth's surface, but the electric field instead of being perpendicular to the earth's surface tends to follow the direction of travel of the wave.* Hence to the component, E z , of the electric field strength perpendicular to the earth's surface there comes an additional component, E XJ in the * This is already noticeable in Figs. 297 and 298, even though not very prominent, as the conductivity and wave-length were assumed to be relatively high for these figures. PROPAGATION OF THE WAVES OVER THE EARTH'S SURFACE 253 direction parallel to the earth's surface. The ratio between the amplitudes of these two components is shown by the full line curves of Fig. 300 for different values of the conductivity and the dielectric constant,* under the assumption that the distance from the transmitter is so great that the waves may be considered not merely as surface waves, but as plane waves. From the curves it is evident that when the dielectric constant is small, the horizontal component can assume quite large proportions. Log$=- -17 d W 9 IO 10 IO 11 W 12 IO 12 it 1 * IO 15 IO 16 IO 17 FIG. 300. In this case, while the magnetic field and the vertical component of the electric field are approximately in phase with each other, there is a phase difference,

S 2 . DETECTORS 275 This produces a clicking sound in the telephone, T, connected in circuit with $2, every time an electromagnetic wave strikes the receiver. 6. In practice MARCONI seems to have used only his second method, which is shown diagrammatically in Fig. 328. The iron wires are formed into an endless string or rope, D, running over two grooved pulleys and kept in motion by a clockwork. The magnetic field, M, in the iron wires, is induced by two horseshoe magnets, H, under whose poles the wires are passed. When the wires pass through the inside of the winding, Si, which is connected into the receiving circuit, they are subjected to the action of the oscillations. The latter are here also indicated by means of a tele- phone connected to the coil S 2 . MARCONI succeeded in operating a relay with this detector [Art. 167c] and thereby to automatically record the received messages, 253 but at present he seems to prefer the use of a suspension galvanometer [Art. 1676] for recording telegrams. 154. Other Forms of Magnetic Detectors. In another class of mag- netic wave indicators, the iron body subjected to the action of the re- ceived oscillations, is located in a rotating magnetic field, or is itself ro- tated in a fixed constant field. To this class belong the arrangements 254 of R. ARNO, J. A. EWING and L. H. WALTER, A. S. Rossi, R. A. FESSEN- DEN, W. PEUCKERT and another of L. H. WALTER. 276 WIRELESS TELEGRAPHY None of these arrangements seem to have entered into radio-prac- tice, for which they are hardly so much intended as for measuring purposes. For measurements, certain of these devices have the advantage that their action is not determined by a single oscillation (as is the case with MARCONI'S magnetic wave indicators) but the effect of a sequence of wave trains is summed up to a certain extent, as with the thermal detectors. 3. IMPERFECT CONTACTS 155. Metallic Granular Coherer. 255 In its original form the "coherer" consists of a tube of non-conducting material (e.g., a glass tube) with two metallic electrodes, EI and E 2 (Fig. 329), between which are a large number of very small pieces of some suitable metal (granules, shavings). This in its normal condition offers an almost infinite resist- ance. If, however, sufficiently strong oscillations are passed through this coherer, its resistance is greatly decreased, falling to several thousand or, in some coherers, even to a few hundred ohms or less. This low resist- ance is retained by the coherer after the oscillations have ceased. In order to bring it back to its non-conducting state, it is necessary to shake it, say, by tapping against the containing tube. DETECTORS 277 Since the time when BRANLEY showed that this simple device con- stituted a wave indicator of much higher sensitiveness than the other forms then known, the coherer was improved and developed along the following lines. a. The shape of the coherer was not changed much. MARCONI cut the electrode surfaces at an angle to the axis (Fig. 330) so that the space between them is wedge-shaped. In this way, upon tapping against the side of the tube where the electrodes are closest together, there is no danger of jamming the metal filings between the electrodes. 6. As to the material, very little of a general nature can be stated. MARCONI used silver electrodes in his early work, as it was easy to form an amalgam with the silver, and his filling was a mixture of 96 per cent, nickel and 4 per cent, silver. Similar coherers were long used by the TELEFUNKEN Co. (Fig. 331). Later SCHLOMILCH, of the TELEFUNKEN Co., devised a very sensitive coherer of gold and aluminium; one electrode being of aluminium, the other of gold, while the filling is gold powder. FIG. 331. A. KOEPSEL obtained a very reliable coherer by using highly polished and very hard steel plate electrodes and granules of glass-hard steeL In regard to the filling, the chemical constituency is by no means the only determining factor, and the shape of the granules is of at least equal importance. In general, high sensitiveness is secured by giving the granules sharp points or edges. The danger of jamming with such granules is minimized by eliminating all those having a long narrow shape. c. Coherers are frequently exhausted (vacuum), this practice having been originated by MARCONI. Complete dryness inside the coherer is assured in 'this way a requirement for reliable operation. 278 WIRELESS TELEGRAPHY d. Some coherers are arranged with adjustable sensitiveness. Where the space containing the metal filings or granules is decidedly wedge- shaped, as in some of the coherers formerly used by the TELEFUNKEN Co., this adjustment is attained by simply turning the coherer, the sensitive- ness being greater when the narrow portion of the wedge-shaped space points downward. In other coherers the distance between the electrodes is adjustable, as in those of A. KOEPSEL, also in those of H. BOAS (Fig. 332) * which latter are in a vacuum. Regulation of the electrodes through FIG. 332. the air-tight ends is made possible by a flexible metal diaphragmf which closes the tube at one end and against which one of the electrodes is pressed from within by means of a spring. If, then, the micrometer screw is turned from the outside, this .brings pressure against the dia- phragm, thereby moving one of the electrodes within certain limits. 156. Mercury Coherers. a. In some experiments of the Italian Navy, the coherer J shown diagrammatically in Fig. 333, was tested. Two electrodes, either both of iron or one of iron and one of carbon, are placed in a glass tube, and between the electrodes is a drop of mercury. This coherer, which was also used by MARCONI for a time in some of his long distance work, seems to be more sensitive than those having solid metal nr> FIG. 333. granules. Moreover, it differs from these in that after the oscillations have ceased, it automatically returns to its initial high resistance.! This form of coherer is no longer used in practice, however. b. Another form of mercury coherer has been devised, apparently inde- pendently by A. KOEPSEL on the one hand and by O. LODGE and A. MUIRHEAD on the other hand. 256 The construction adopted by the latter is shown diagrammatically in Fig. 334. A small steel wheel, R, to which current is brought through the brush, B, is rotated by clockwork * From a pamphlet of H. BOAS. t The metal diaphragm is soldered to a metal tube which, in turn, is soldered to the platinum coating on the glass. J Apparently the idea originated with an Italian Signal Officer name.d CASTELLI. It is "self-restoring." DETECTORS 279 or by a small motor. The wheel dips slightly into mercury, Q, which is covered by a thin layer of mineral oil. Normally the wheel and mercury do not make a conducting contact; a contact is formed, however, as soon as oscillations pass through this coherer and disappears again as soon as the oscillations cease. These coherers of LODGE and MTJIRHEAD seem to have given good service in practice. According to the investigations of W. H. ECCLES, the action of this detector, as well as of that described in a, seems to depend upon the nega- tive temperature coefficient of the iron oxide coating which forms on the iron or steel electrode. If the oxide at the point of contact becomes heated by the oscillations, its resistance is greatly decreased and the current from the battery supplying the detector circuit (see Fig. 329) rises considerably above its initial value. FlG - c. L. H. WALTER 257 devised a very useful and also self-restoring mer- cury coherer, the sensitive contact being made between a tantalum point (T, Fig. 335) and mercury (M). This detector is said to be not quite so sensitive as, e.g., the electrolytic, with very weak oscillations, but with strong oscillations it gives much louder sounds in the telephone. 157. Carbon or Graphite Coherers (Microphone Contact). Another class of coherers makes use of carbon or graphite. Two arc-lamp car- bons, one resting loosely upon the other, or either an arc-lamp carbon or a graphite rod together with a wire constitute the simplest, though im- practical forms of this type of coherer. They suffice for the detection of electromagnetic oscillations as well as any microphone, which latter in fact was used by HUGHES for this purpose as far back as 1879. These coherers, like those made of metal granules, change their resistance when oscilla- tions are passed through them, but differ from them, resembling the mercury coherers in this respect, in that they are self-restoring. This coherer was used in practice for quite some time in the form, de- vised by A. KOEPSEL, in which the imperfect contact consisted of a highly polished, very hard steel plate and a hard graphite rod. This combination is very sensitive but is not sufficiently reliable for regular practice. 4. ELECTROLYTIC AND OTHER DETECTORS 158. Anti-coherers. This name is frequently applied to those de- tectors in which the effect of the electromagnetic oscillations, instead of being a reduction is an increase in the resistance; these anti-coherers more- over are self-restoring. a. De Forest's "Responder." In a tube of non-conducting material there are, as in the ordinary coherer, two metallic electrodes, which some- times are hollowed out as shown in Fig. 336. The space between them is 280 WIRELESS TELEGRAPHY filled with a paste which in one case, e.g., consists of water and glycerine, metal filings and pulverized lead. DE FOREST gives the following explanation of its action. If this de- tector is placed in a battery circuit, a small current flows through it. The resulting electrolysis, causes the formation of very fine metallic bridges between the metal filings. The effect of the oscillations is to destroy these bridges. When they cease, however, the current immediately causes the bridges to form again so that the wave indicator resumes its nor- mal resistance. b. The detector of J. E. IvES 258 contains two crossed silver wires, which almost make contact in a solution of potassium bromide or iodide or of both. Here the formation of the bridges between the two wires has been observed under the microscope. 159. The Electrolytic Detectors of Ferrie, Fessenden, Nernst and Schlomilch. It seems that the electrolytic detector, to be described in what follows, was announced independently by FERRIE, R. FESSENDEN, NERNST and W. SCHLOMILCH after M. I. PUPIN, at a much earlier date (U. S. Patent, 713045, 1898) had used a similar cell for rectifying alter- nating currents. The following is a description of the form in which the electrolytic detector was used by the TELEFUNKEN Co., under the name of "SCHLOMILCH cell" (Fig. 337): FIG. 336. FIG. 337. FIG. 338. In a container filled with dilute sulphuric acid there are two platinum wire electrodes, one of which is very thin and covered by glass tubing, except at its end where the bare wire projects for a very short distance. This thin wire is connected to the positive pole, the heavier wire to the negative pole of a battery whose e.m.f. is only slightly greater than the e.m.f. which is produced by the polarization of the cell, platinum dilute sulphuric acid platinum. Consequently a very small current flows through the cell, so that a galvanometer connected in circuit would show a DETECTORS 281 slight deflection. As soon as oscillations act upon the detector, a consid- erable increase in the current results, so that the galvanometer in the cell circuit has a much greater deflection and a clicking sound is heard in a telephone connected in the circuit. The moment the oscillations cease, the current falls to its normal value. Fig. 338* is an exterior view of the detector. In FsssENDEN's 259 "liquid barretter" (Fig. 339) the point of a fine Wollaston wire (platinum wire coated with silver) just dips into the sur- face of the electrolyte (potassium nitrate solution) ;f here also the Wol- laston wire is joined to the posi- tive pole. A very fine adjust- ment makes it possible to secure the most efficient depth of sub- mersion and also enables prompt readjustment in case the point of the wire is harmed at any time by too heavy a discharge. a. The characteristic prop- erties of the electrolytic de- tector are: 1. The sensitiveness increases as the surface area of the posi- tive electrode decreases. Hence, an extremely small electrode is used for radio-purposes. In the SCHLOMILCH detector it is a platinum wire of about 0.03 mm. diameter, in glass, from which it projects only very slightly, while in FESSENDEN'S liquid barretter it is a Wollaston wire of still much -smaller diameter. 2. The normal resistance of the cell, when not excited by oscillations is only several thousand ohms, hence is of about the same order as that of the coherer when excited. 3. Other things being equalf the galvanometer deflection or the inten- sity .of the sound in the telephone increases as the amplitude of the oscil- lations is increased. 261 The investigations of G. W. PiERCE 1 (which, however, were made with low frequency alternating current) indicate that the electrolytic detector * From a pamphlet of the TELEFUNKEN Co. In this form the positive electrode is renewable. A later construction of the TELEFUNKEN Go's, electrolytic detector has three fine wire electrodes, which can be used alternately. 260 t According to J. E. IvEs 259 a solution of caustic potash (1 vol. saturated solution to 2 vols. water) increases the resistance of the detector, but also increases the range of its variation due to the oscillations. IVES used a Wollaston wire of 0.001 mm. diameter (of the platinum), submerged to a depth of about 0.1 mm. t That is, with constant decrement, as this determines the galvanometer deflection as well as the amplitude of the oscillations. FIG. 339. 282 WIRELESS TELEGRAPHY acts as a rectifier due to the polarization [see Art. 162a],the resultant cur- rent being unidirectional. b. FESSENDEN 262 found that with his liquid barretter the signals in the telephone became louder and sharper on applying a pressure of three to four atmospheres to the barretter. c. The customary method is to connect the electrolytic detector in series with a battery and a telephone. It has often been proposed to elim- inate the battery, by using for the non-sensitive electrode of the detector, a metal which, together with the sensitive electrode, will form a galvanic cell of suitable e.m.f. 160. Crystal Detectors. There are a number of crystalline substances which, when substituted for the coherer in the arrangement of Fig. 329, produce a galvanometer deflection or sound in the telephone, whenever oscillations are passed through the circuit. All these substances can, therefore, be used as wave indicators. a. The use of these substances as wave indicators probably originated with the experiments of F. BRATJN (1901) 263 in connection with psilo- melan (a complex mineral of irregular composition and containing man- ganin), also with galena (PbS),iron pyrites (FeS 2 ) and pyrolusite (MnO 2 ). At the suggestion of BRAUN, the TELEFUNKEN Co. developed the psilo- melan detector; its sensitiveness was brought to a degree about equal to that of the SCHLOMILCH electrolytic detector. The following substances have since then been proposed and widely used in practice: 263 carborundum (SiC) (DUNWOODY), titanium di- oxide (Ti0 2 ), molybdenite (MoS 2 ) (G. W. PIERCE), copper pyrites (CuFeS 2 ), also (Cu 3 FeS 3 ), chalcocite (Cu 2 S), manganese dioxide (MnO 2 ) , and iron pyrites (FeS 2 ). The usual method is to place a small piece of one of these minerals be- tween two metal electrodes (of almost any suitable material) under light pressure, and in series with a battery and telephone in the circuit receiving the oscillations. The use of a plate of the detector material in conjunc- tion with a metallic powder (thus, molybdenite powdered silver) has also been proposed. In this same class also belongs the detector of S. G. BROWN, in which lead peroxide is placed between a lead and a platinum electrode, the lead being connected to the negative, the platinum to the positive pole of the battery. b. In a second class of detectors either a combination of two minerals or of one mineral with some specific metal is used. To this class belongs, e.g., the "perikon" detector of G. J. PicKARD, 263 which is a combination of zinc oxide (ZnO) and copper pyrites (CuFeS 2 ). c. As to the nature of the action of these detectors, 264 " it suggests itself that this may be thermoelectric. In fact C. TISSOT has shown that this is very probably the case with a number of detectors the combinations DETECTORS 283 metal-copper pyrites, metal-chalcocite, metal-manganese dioxide, metal- tellurium. He proved that: 1. These detectors are sensitive only if the contact is limited to a point. 2. They operate without a battery in series, and when a battery is used the sensitiveness does not depend upon the value or direction of its e.m.f. 3. The direction of the current (direct) obtained under the influence of the received oscillations is always the same as 'the direction of the therrno-e.m.f. With another group, however carborundum, anatase (titanium diox- ide), molybdenite and the perikon detector TISSOT'S tests established that: 1. The form of the contact is of little or no importance, even relatively large polished plates placed between two metallic electrodes make sensitive detectors. 2. The use of a battery in series with the detector, with proper value and direction of the battery e.m.f., increases the sensitiveness. 3. The sensitiveness of these detectors bears no relation whatever to the value of their thermo-e.m.f. He, therefore, concludes that in this last mentioned group thermo- electric forces play no important part in their action as detectors. G. W. PIERCE, l as the result of extensive investigations, including oscillograph records made with the BRATTN tube, concluded that with carborundum, anatase, brookite (another form of TiO 2 ) and silicon, ther- moelectric forces were not involved, but that these detectors were better conductors in one direction than in the other, in short, act as rectifiers [Art. 162]. 161. Incandescent Lamp Detectors, Gas Detectors. a. J. A. FLEM- ING 265 observed the following phenomenon: An electrode (A, Fig. 340), say of cylindrical form, is fused into an incandescent lamp bulb, whose filament is made incandescent^ by means of a battery, B.* A circuit containing a galvanometer, G, (or telephone) and a coil,$ 2 , is joined to the electrode, A, at one end and to the lamp filament, at the other, K. The aerial coil Si is coupled with /S 2 ; hence if oscillations pass through $1, the oscillations induced in the circuit AGS^K will cause the galvanometer to deflect (or produce a sound in the telephone). The galvanometer needle will return to its zero position as soon as the oscillations cease, i.e., the arrangement is a self-restoring wave indicator. Several years ago C. TissoT 265 used this wave indicator for measure- ments over considerably long ranges but he complains of the irregularity * A choke coil [Art. 1656] should be inserted in the leads from the battery to the lamp. 284 WIRELESS TELEGRAPHY of the deflections.* But in the more recent form of FLEMING'S " oscilla- tion- valve," the anode being a cylinder of carbon and the cathode a tungsten wire, this detector seems to have met all reasonable require- ments as to sensitiveness and reliability. This is borne out by the fact that MARCONI has been using it, in conjunction with an EINTHOVEN string galvanometer, in his transatlantic stations. FIG. 340. The incandescent lamp may be replaced by the tube devised by A. WEHNELT 266 which operates in the same manner. The incandescent cathode of this tube is a wire coated with a metallic oxide, and the anode is a hollow aluminium cylinder, concentric with the cathode. b. H. BRANDES 267 has found that it is very advantageous to insert an auxiliary battery or cell, E, about as shown in Fig. 341,f when using * TissoT 265 describes a wave indicator using rarified air (as in the ZEHNDER tube), which he found to be less sensitive but better adapted for measuring purposes, t DI and Z>2 are inductive coils. DETECTORS 285 these wave indicators. The sliding contact, SC, is adjusted until the detector is operating at the best point of its characteristic [see Art. 162a], thereby greatly increasing the sensitiveness as compared to operation without the auxiliary cell. This method has lately also been adopted by FLEMING. c. This arrangement had in fact been proposed quite some time ago by DE FOREST in his so-called "auction" detector. The audion, as first constructed, was in the main identical with FLEMING'S construction, ex- cepting that DE FOREST made use of an auxiliary cell (E, Fig. 341) from the very first. Another construction of DE FOREST'S " audion," 268 which seems to be of particular excellence is that shown diagrammatically in Fig. 342. Here F is the metallic filament made incandescent by the current from the battery, EI, N is a wire grid or network and P is a disc-shaped electrode. All three electrodes are placed in an exhausted glass bulb. 5. GENERAL CONSIDERATION OF DETECTORS 162. The Nature of the Action in Various Detectors. a. H. BRANDES 267 has shown that the action of very many wave indicators may be gener- alized under a single, common point of view. All these wave indicators have in common the fact that they do not follow OHM'S law, so that their characteristic [Art. 113] instead of being a straight line, is an irregular curve. This variation includes two cases, viz.: 1. The curve is not symmetrical in the first and third quadrants (Fig. 343), i.e., the current is not the same for any two potentials of equal amplitude but opposite sign. Hence, if the potential is that of an alternat- ing current or oscillation, the resulting current is not the same in both directions. Consequently the currents in two directions do not neu- tralize each other in their action upon a galvanometer which shows a deflection without the insertion of an auxiliary battery; likewise a tele- 286 WIRELESS TELEGRAPHY phone in the circuit is caused to produce a "click."* The detector there- fore is said to be a "rectifier." f 2. The characteristic curve is symmetrical in the first and third quadrants (Fig. 344). In this case the oscillations will cause equal currents to flow in both directions, so that the detector, per se, does not act as a rectifier and a gal- vanometer in the circuit does not deflect. If now, however, the constant potential of an auxiliary cell is impressed across the poles of the detector, an increase in the total e.m.f. due to the oscillations causes a certain increase in the current flowing through the FIG. 343. FIG. 344. detector; an equal decrease in the e.m.f. due to the oscillations, however, does not produce an equal decrease in the current, because of the curva- ture of the characteristic. Hence the effect of the oscillations is to change the galvanometer deflection or produce a click in the telephone; the de- tector with a battery in series acts as a rectifier. The rectifying action has been shown by BRANDES to be greater (1) the steeper the characteristic is toward the axis of abscissae at the point corresponding to the e.m.f. of the auxiliary cell, and (2) the sharper its curvature is at this point. Hence, in using this class of detectors, the auxiliary e.m.f. must be so chosen or adjusted as to permit of operation at the most favorable point of the characteristic.^ b. Many investigations of the characteristics and action of the various detectors have been made. 269 It has been shown that with incandescent lamp, electrolytic and the various crystal and thermal detectors (car- borundum, perikon, graphite-galena, copper-molybdenite, anatase, brook- ite) the characteristic assumes entirely different shapes upon reversal of the current. Thus, Fig. 345 shows the characteristic of the highly sen- * This explains FERRIE'S observation of the fact that the electrolytic detector acts as a wave indicator without the presence of a battery in the circuit. f In some cases current passes in one direction only, the detector acting more or less as a valve. | The heating devices of the detectors described in Art. 150c have a similar pur- pose; they are adjusted to give the temperature at which the characteristic will have its most favorable curvature. DETECTORS 287 sitive perikon detector [Art. 1606] as obtained by the measurements of W. H. ECCLES. 269 ECCLES also investigated the relation between the sensitiveness of various detectors and the value of the impressed auxiliary e.m.f. and in almost all cases found that maximum sensitiveness corresponded to a certain value of the auxiliary e.m.f. in agreement with the conclusions of BRANDES. He also investigated the relation between the D.C. energy delivered by the detector to the telephone and the energy supplied to the detector by the oscillations and, in all cases coming under his observation,* he found that the curve representing the rela- tion between D.C. energy delivered and the high frequency energy sup- Red Zinc Oxide - Positive '-25 ', 0.2 f * -~>K-^ , 50 100 150 200 0.2 , 0.4 . 0.6 ^ 0.8 1.0 Amp. Volt FIG. 345. High Frequency Energy FIG. 346. plied, is a straight line which does not quite pass through the common zero point (Fig. 346). Hence there is an initial value above which the oscillating energy must lie to produce useful D.C. energy. The ratio of the D.C. energy to the high frequency energy supplied, i.e., the efficiency of the detector, was found, under the conditions of the tests made by ECCLES, to be 13 per cent, for the electrolytic detector (sensitive electrode 0.006 mm. thick, sulphuric acid electrolyte), 9.3 per cent, for the carborundum detector, 13 per cent, for the perikon detector and about 3 per cent, for graphite-galena, the figures being the maximum obtained in each case. These tests are of particular value, because the conditions (frequency, energy) encountered in the actual practice of radio-telegraphy were re- tained as closely as possible, which can by no means be stated of all the investigations which have been made in this field of work. 163. What do the Different Types of Detectors React Upon? 270 a. Assume that only a single oscillation (say a single discharge from a con- denser circuit) acts upon a wave indicator. Then: 1. The reaction of the thermal detectors, and to some extent also of mercury coherers, depends upon the heat developed within them, i.e., upon the current effect -T^TJ ' ^o 2 . 4 A' a * Iron-mercury, electrolytic detector, carborundum, perikon detector, graphite- galena. 288 WIRELESS TELEGRAPHY 2. For magnetic detectors, the amplitude (or the maximum amplitude [Arts. 56c and 61c]) of the current, / (or I max ) is probably the determining factor. 3. For rectifying detectors, the quantity of electricity passing in one direction in excess over that quantity passing through the detector in the other should be the determining factor. This excess quantity depends not only upon the amplitude of the potential occurring between the poles of the detector, but also upon the damping. Where the rectifying action is complete so that the resultant flow of current is unidirectional, this quantity is approximately* = TTJ I 0) hence like the current effect, it varies as v^? but it varies as IQ, not as /o 2 .f With the metallic coherers, a certain minimum potential difference be- tween the electrodes must exist to produce a reaction. However, in order to cause a large change in the resistance, as is required in practice, a certain current effect must be reached. In this respect, therefore, the coherer is not unlike the thermal detectors, for its action also depends upon the decrement of the oscillations as well as upon the amplitude. b. Assume that a very rapid sequence of oscillations (e.g., undamped oscillations or damped oscillations of very high discharge frequency acts upon the detector. Then there are two possible cases, viz., 1. The effect upon the wave indicator is determined entirely or almost entirely by the first oscillation. The oscillations which follow do not materially aid the action. This is the case, e.g., with the coherer and with MARCONI'S magnetic detector. 2. The effect upon the wave indicator is the sum of the effects of the successive oscillations in the series. This undoubtedly is the case with the thermal detectors, the magnetic detectors of the WALTER type [Art. 154], to a certain extent also with the electrolytic detectors of the SCHLOMILCH type and in general, with all those to which Art. 162 applies. In considering this question, however, it is important to distinguish sharply between the effect upon the detector and that upon the receiving apparatus. If, e.g., the discharge frequency is increased, while the amplitude and damping remain constant, there is no doubt that the effect upon a thermal detector, in other words, that the direct current delivered by it will also increase, other things being equal. Nevertheless the intensity in the receiving telephone [see Art. 165] may decrease with the increased discharge frequency. The amplitude of the oscillations of _d * The exact value is ~ T -; =, irN 1 e d t It seems that the action of certain detectors, which are generally considered as perfect rectifiers, depends upon the current effect. If that is the case, then these detectors can not be pure rectifiers. DETECTORS 289 the telephone diaphragm, depends not only upon the amplitude of the D.C. impulses but also upon their variation with time and upon the length of the intervals between them; if these intervals become too short the amplitude of the telephone diaphragm's motion may decrease. c. If the individual oscillations follow one after the other with rela- tive slowness (e.g., damped oscillations produced by means of a resonance transformer) then not only the wave indicator, but also the method of reception will determine whether or not the effects of the individual oscillations are summed up. Thus when receiving with a telephone only the effect of a single oscillation comes into question, whereas with a siphon recorder [Art. 167a] or similar devices the result is a summation of the individual effects. 164. Testing the Sensitiveness of Detectors. 271 A general statement as to the relative sensitiveness of two detectors is often impossible when the detectors are of a different type. a. The ratio moreover may vary as the character of the oscillations used for the test varies; thus it may be different for undamped than for damped oscillations, and, again, with damped oscillations it will depend upon the amplitude, the decrement and the discharge frequency. At twenty discharges per second, a good coherer will receive at practically the same range as a thermal detector, but at 1000 discharges per second, with the same transmitted energy and a corresponding decrease in ampli- tude, the range of the thermal detector will be over five times that of the coherer (COUNT Anco 160 ). Hence where it is desired to determine which of two detectors is the most sensitive for service from a given station, it is advisable to let a closed oscillating circuit whose oscillations have just about the same time variation and discharge frequency as the waves of the station in question, act upon the detectors. The alternative method of using any of the so-called station testers or, perhaps, as is frequently done, of using an interrupted direct current to act upon the detectors, can only give very questionable results. 6. Different arrangement of the receiving circuits may also greatly affect the relative sensitiveness of two detectors. Thus, the results may differ if the detectors are connected to a weakly damped receiving con- denser circuit as compared to inserting them in a closed (aperiodic) cir- cuit [see Art. 175 et seq.]. c. Finally the receiving apparatus itself may greatly influence the relative sensitiveness and hence the attainable range. Thus telephone reception may give quite different results than a recording receiver involv- ing the use of a relay. All these factors must be carefully considered in comparing detectors as to their sensitiveness. 10 290 WIRELESS TELEGRAPHY 6. RECEIVING APPARATUS 165. Telephone Reception. The receiving apparatus assumes its simplest forms with those wave indicators which are self -restoring upon cessation of the oscillations or which at least are able to immediately indicate a new oscillation (including all wave indicators with the excep- tion of the metallic coherer). With these a telephone can be used as the receiver. a. The simplest method of connection for wave indicators without an auxiliary cell, is shown diagrammatically in Fig. 347* and for wave indi- cators with an auxiliary cell, in Fig. 348. f In these J is the wave indicator, T the telephone, E the auxiliary cell and L the receiving circuit containing /. A telephone thus connected when held to the ear produces a clicking sound at each discharge of the trans- mitter, if the discharge frequency is low. If this fre- quency is high, either a pure tone is heard in the tele- phone in which case the discharge frequency is regular and within the range of audibility, or otherwise a buzzing discordant sound is heard. The letter "a" ( in the MORSE code) is heard in the telephone as a short (dot) click, buzz or tone followed by a longer sound (dash) of the same character. Tele- grams can therefore be received by means of the ear, just as with the sounders or buzzers used in wire telegraphy. &. For wave indicators with an auxiliary cell the arrangement requires a slight modification. Firstly, it is advantageous to impress the particular e.m.f. at which the indicator is most sensitive [Art. 162a]. To adjust for this, a high resistance, AB (Fig. 349), is con- nected across the terminals of the cell E, which should have a relatively high e.m-.f., and is provided with a sliding contact, K.% Then any desired voltage, up to that of the cell, E, can be obtained between K and B and hence across the terminals of the wave indicator, J. Furthermore, the connections of Fig. 348 would have the disadvantage of causing the oscillation currents in the circuit L to partly branch off * The receiving circuit, L, must of course form a complete closed circuit. If the telephone is connected in parallel to the wave indicator, then the circuit L must have a condenser (block condenser [Art. 41d]) in it, in order to prevent the direct current of the wave indicator from flowing into it. t But see 6. J This is simply a "potentiometer" connection. \\ FIG. 347. FIG. 348. DETECTORS 291 into the telephone circuit TE (Fig. 348) so that only a part would pass through the wave indicator. To prevent this two choke coils, DI and D 2 in Fig. 349, are inserted at the junction points to block the path of the oscillations through the telephone circuit [Art. 416]. FIG. 349. 166. Amplification of the Sound in Telephone Reception. Two devices which have been successfully used for intensifying the sound produced in the receiving telephone, are the telephone relay of S. G. BROWN and the so-called " sound intensifier" of the TELEFUNKEN Co. F IG . 350. FIG. 351. In both apparatus, the detector current first acts upon a kind of micro- phone and the microphone current then flows through the telephone. a. The Brown telephone relay 272 is illustrated in Figs. 350 and 351, the latter showing the connections. N and S are the poles of a 292 WIRELESS TELEGRAPHY horseshoe magnet, on which two soft iron cores whose windings are marked K and H rest. P is a steel tongue carrying a small osmium-iridium plate 0, which lightly touches a contact point M also consisting of an osmium-iridium alloy. C is a dry cell whose circuit contains the contact OMj the winding K and the telephone T. The current, whose effect upon the telephone is to be amplified, is sent through the winding H. This causes the steel tongue, P, which takes the place of the telephone diaphragm to vibrate, thereby alternately strengthening and weakening the current in the circuit COMKT in unison with the oscillations, with the result that a materially stronger effect is produced upon the telephone, Tj than if the current in H were sent direct through T. Tests made by the British Admiralty and Post Office Departments indicated that this telephone relay doubled the range. Messages whose existence could not be discovered with the ordinary telephone apparatus Detector Current FIG. 352. were easily received with the BROWN relay. During tests made between the Clifden and Poole stations in Ireland, by using two relays connected in series it was possible to clearly hear messages 2 m. away from the receiver, while in the ordinary receiver without relay the same message could just be discerned as a slight noise. b. In the Telefunken Tone Intensified which is adapted only for use with a tone transmitter, the detector current is conducted to a small electromagnet (Ti, Fig. 352) having a large number of turns. In the field of the magnet there is a small armature A i, whose natural frequency of vibration corresponds to the frequency of the detector current and hence to the tone of the transmitter. This resonant armature presses against a microphone contact, MI, which is in the circuit of a local battery, EI. This circuit also contains an electromagnet, Tz, constructed identically as TI. The current flowing through T% pulsates at the same frequency as the detector current flowing through TI, but has a much greater amplitude. Hence the armature, At, which is identical with Ai, vibrates more violently than A i, so that the pulsations in circuit II, which contains the microphone contact, Mz, local battery, Ez, and electromagnet, Ta, become still greater than in circuit /. Armature A s and microphone DETECTORS 293 contact M 3 then produce another increase in the pulsations. This three- fold amplification is sufficient to produce a current of about 10~ 2 amp. in circuit 777, which contains the receiving telephone, when the detector current is only from 10~ 7 to 10~ 8 amp. FIG. 353. In conjunction with the intensifier it is customary to use a special loud-speaking telephone (LT, Fig. 352), having an acoustic resonator at the opening of its mouthpiece, the resonator being tuned to the tone of the transmitter, thus causing a further amplification. 294 WIRELESS TELEGRAPHY This extensive application of mechanical and acoustic resonance together with the microphone amplification results in a very marked increase in the sound intensity. Fig. 353 shows the construction of this sound in- tensifier, wnich has found a place in many stations. 167. Automatic Recording of Messages. With certain detectors (e.g., thermal detec- tors, LODGE and MUIRHEAD'S mercury coherer) the receiving telephone can be re- placed by a well damped galvanometer. If connected in series with the wave indicator and a battery, the galvanometer deflects and, as soon as the oscillations cease, returns to its zero position. This makes a direct record- FIG. 354. ing of telegrams possible in various ways. LODGE and MUIRHEAD'S method* as applied to their mercury L_J~l a coherers is as follows (Fig. 354). f A pen or pencil is attached to the movable coil, $, of a galvanometer and touches a paper strip or tape which is moved by clockwork, as in the ordinary MORSE recorder. As the galvanometer coil rotates the pen or pencil is moved perpen- dicularly to the direction of motion of the paper. As long as the wave indicator is not subjected to oscillations, the galvanometer coil remains stationary and the record is simply a continuous straight line (a to b in Fig. 355). A brief excitation of the wave indicator i.e., a MORSE dot produces the effect shown at b in Fig. 355, while a dash appears as shown at c-d. b. Later the movable coil galvanometer used by LODGE and MUIRHEAD was re- placed by the much more sensitive and * The complete outfit is frequently called "Siphon recorder." f M is the horseshoe magnet of the movable coil galvanometer. DETECTORS 295 less sluggish EINTHOVEN string galvanometer and a photographic record of the message made. The EINTHOVEN string galvanometer (FiG. 356)* as is well known, consists of a fine wire (Wollaston wire, fine metal strip or conductive quartz fiber), which is stretched in the gap between the poles of a magnet perpendicularly to the magnetic flux lines (between F and M in Fig. 356). If current is passed through this wire, the latter will be displaced from its normal position in a direction perpendicular to its axis and to the mag- netic flux lines. The wire moves in front of a narrow illuminated slit (Fig. 357). A photographic reproduction of the slit and wire on a sensitive film passing perpendicularly across the slit, appears, on the negative, as a broad dark band, with a fine light line (the wire) through its center. If, however, the wire is displaced from its normal position, first for a brief instant, and then for a somewhat longer duration, the photographic record will appear as a light line similar to the line of Fig. 355, i.e., the characteristic dot and dash. A complete photographic recorder 275 is illustrated in Fig. 358. f At the left is the galvanometer, whose wire and slit are illuminated by a small incandescent lamp (of which the plug and flexible lead are visible). FIG. 357. FIG. 358. The micro-photographic lens is placed in the metal tube, at the right is the camera and in back of this is the case in which the strip of sensitive * The cut is taken from a catalogue of PROF. EDELMANN & SON (Munich) who make this and also more modern types of galvanometer. The firm of E. HuTH 274 makes still another construction of the instrument. f Construction of the C. LORENZ Co. Other receiving apparatus of the same kind are constructed along very similar lines. 296 WIRELESS TELEGRAPHY film which is moved by clockwork, is developed and fixed. It has already been mentioned that MARCONI also uses the EINTHOVEN string galvan- ometer in his transatlantic stations [Art. 16 la]. FIG. 359. c. In place of a galvanometer, a relay (RiRi, Fig. 359) which opens and closes the circuit of a MORSE recorder, M, and local battery, Ez (Fig. 359),* can be used. FIG. 360. The construction of a polarized relay which is customarily used for this purpose, is no doubt evident from the diagram of Fig. 360 (a, view from above; 6, view from side). M is a permanent steel magnet, with one * In regard to the resistance, w, in Fig. 359, see Art. 1686. The choke coils, which here also are inserted to protect the relay from the oscillations in the main circuit, L, have been omitted in the diagram. DETECTORS 297 pole at A, the other at B. On the latter are placed the iron cores, BiB 2 , of the coils, Si and S 2 , which are in circuit with the wave indicator and a battery through the leads i\. U is a movable armature which makes contact with C/i closing and opening the circuit i which, in addition to the MORSE recorder (M } Fig. 359) contains a battery of one or more cells. A relay of this kind is quite sensitive. Thus the TELEFUNKEN 276 relays of this type were stated to respond positively when operated with 1.4 volts and a series resistance of 100,000 ohms. So high a degree of sensitiveness is attainable only if the adjustments for the distance BiB- 2 and the contacts C/iC/2 are particularly fine. Furthermore, the armature U must be balanced with excep- tional care to prevent interference from outside disturbances or from the rolling of the ship when used at sea. The TELEFUNKEN Co. formerly used a magnetic adjustment 276 on its relays. By turning a piece of soft iron mounted on the casing of the relay, the magnetic field within was varied, thus providing the desired regulation. This, moreover, has the advantage of entirely enclosing the relay in its casing, so that its other -pio. 361. adjustments remain fixed once and for all. The external appearance of the relay is shown in Fig. 36 1. 276 The reason for not simply inserting the MORSE apparatus directly in the same circuit as the relay is as follows : On the one hand, the poten- tial existing across the terminals of any of the usual wave indicators in their unexcited condition is limited to a certain maximum value, above which (at most 2 volts, usually much less) it must not be permitted to rise. On the other hand, only very small currents (usually considerably below Jfooo amp.) can be allowed to flow through the majority of detec- tors during excitation without harming them. This combination of very low voltage with very small current is generally sufficient to operate a sensitive relay but not a MORSE recorder. d. Some detectors in fact can not stand a current sufficient to operate a sensitive polarized relay. Hence, when it was desired to automatically record telegrams with such a detector, it was formerly necessary to employ a photographic method with the aid of a galvanometer. With the sound intensifier of the TELEFUNKEN Co., however, as long as the essen- tial requirements of constant and sufficiently high discharge frequency in the transmitter ("tone transmitter") are filled, it is possible to use the more convenient MORSE recorder. 298 WIRELESS TELEGRAPHY The connections 273 for this purpose are sketched in Fig. 362. The microphone current of the third amplifier (/// in Fig. 352) consisting of D.C. with superimposed A.C., instead of being led directly to the loud- speaking telephone (LT, Fig. 352), is sent through a small transformer, TF (Fig. 362), by way of the throw-over switch, U. A pure alternating e.m.f. is induced in the transformer secondary. In the secondary cir- cuit, however, is a rectifier, V, so that current flows through it and To Loud-Si FIG. 362. through the relay, R, in one direction only. This unidirectional current, however, is strong enough to actuate the polarized relay. 168. Recording Apparatus for the Metallic Granular Coherer. The recording devices described in what has preceded, suffice for wave indi- cators which are self-restoring, but not for the metallic granular coherer. The latter, if connected with one of these devices, would become conduc- tive at the first oscillation and remain so indefinitely; the relay in circuit with the coherer would then remain closed and the record would be a continuous straight line. FIG. 363. a. Hence, it is absolutely necessary to have a so-called "tapper" to restore the coherer to its normal condition after each oscillation or wave train. As is evident from the diagram of Fig. 363, the construction of the tapper is simply that of the ordinary electric bell or buzzer. In the method of connection shown in Fig. 364,* which represents a standard coherer receiving outfit, current flows through the coherer immediately before it is tapped. As soon as the tapper strikes the * EiE 2 are galvanic cells, K is the tapper, R the relay, UUi the "make and break" contact of the relay, M the MORSE recorder, W a bell or sounder. P is a throw over switch for cutting in either the MORSE recorder, M, or the sounder, W, DETECTORS 299 coherer, this current is interrupted within the coherer. In spite of all precautions [see 6], this will be accompanied by minute sparking which causes deterioration of the granules; this tends to prevent easy restora- tion of the coherer and to reduce its life or duration of usefulness. To meet this difficulty, the TELEFUNKEN Co. and others arranged the tapper so as to open the coherer circuit just before striking the coherer, so that the tapping always occurs at zero current. 6. In the method of connection shown in Fig. 364 there are three places where circuits containing coils wound on iron cores, i.e., having high self-induction, are opened. Hence, quite high potentials arise and FIG. 364. sparks occur at the points of interruption. This may result in the forma- tion of electromagnetic waves which act upon the wave indicator. But even if this does not occur, the interruption of the relatively large cur- rents may induce an e.m.f. in the circuit of the wave indicator, sufficient to cause the latter to respond to it. This difficulty is usually avoided or minimized by placing in parallel with the break in the circuit and with the iron-core windings, non- inductive resistances [as, e.g., w in Fig. 359] of suitable ohmic value or polarized cells (e.g., two platinum wires as electrodes in diluted sulphuric acid), or also condensers of proper size, sometimes in conjunction with non-inductive resistances. 169. Call Signals. When it is desired to transmit a message instantly and perhaps to obtain an immediate reply (as in military work), it is essential to have some method of calling the receiving station. This is also of great importance when ships at sea are in danger. Otherwise, for 300 WIRELESS TELEGRAPHY ordinary radio-traffic, a call signal is not absolutely essential. More- over there are recording receivers so arranged that their clockwork is automatically started when a telegram arrives and stops as soon as the telegram is completed.* a. Where a relay is provided to operate the MORSE recorder, it is com- paratively a simple matter to connect an electric bell (W in Fig. 364) as a call signal. b. However, when the use of a relay is objectionable or undesirable in view of its added complication to the equipment, so that a simple telephone receiver is used, the problem is somewhat different. The TELEFUNKEN Co. has found a simple solution for it 277 as follows: A moving coil galvanometer of high sensitiveness,! whose coil and the pointer connected thereto are very sluggish, is placed in the detector cir- cuit. When the pointer deflects up to a certain angle, it runs into a contact wheel which is turned by a small clockwork and which holds the pointer fixed. This closes a circuit containing the call bell and a battery, so that the bell rings until the operator at the receiving station breaks the contact. The sluggish motion of the movable coil and pointer makes it necessary for the transmitting station to send out a long dash lasting say 10-12 seconds, during which time the transmitter continues to send out waves having a cumulative action upon the galvanometer. This prevents atmospheric disturbances of short duration from actuating the call signal and unnecessarily calling the station operators to the receiver. 170. Comparison of the Different Kinds of Detectors. a. The main points to be considered in determining the practical usefulness of various detectors are as follows: 1. Sensitiveness. 2. Reliability in operation. 3. Simplicity in operation. 4. Simplicity of the necessary auxiliary apparatus. 5. Possibility of using a call signal. 6. Possibility of using a recording receiver. 7. Rapidity of telegraphing attainable. b. As to the sensitiveness, a practical consideration of particular importance is whether the action of a series or sequence of successive waves (wave trains) is cumulative or not [Art. 163]. This is not the case with the metallic granular coherer. As it is cus- tomary in modern practice to work with a relatively high discharge fre- quency and relatively low energy per discharge, this alone has been sufficient to displace the coherer from practical use.f * The use of such automatic recorders is greatly limited in wireless telegraphy, as atmospheric disturbances constantly actuate the clockwork. t 1 scale division = 10~ 7 amp. t Except in certain special cases [e]. DETECTORS 301 c. To this, however, is added another undesirable property of the coherer. It seems that with the carbon and graphite coherers, as well as with the metallic granular coherers, the reliability or certainty of operation becomes greatly reduced as the sensitiveness is increased, a change which is not nearly so marked in other wave indicators. Operators have always suffered from the capriciousness, one might say, of any very sensitive coherers. High sensitiveness is of practical value only when combined with suffi- cient reliability in operation. Little can be stated on this subject as to the various wave indicators, as this depends not merely upon the par- ticular type of indicator, but to a great extent upon the care taken in the construction of the individual indicator. Non-sensitiveness to mechanical jarring and above all to momentary overloading caused by atmospheric disturbances or the proximity of a powerful transmitter is essential to reliability. Accordingly, wave in- dicators having a point contact, as, e.g., certain of the thermal detect- ors, are dangerous in both respects, while electrolytic detectors like that of SCHLOMILCH are non-sensitive to jarring, but very sensitive to overloading. d. Operation is simplest with those detectors which, when once adj usted, require no further regulation (some of the crystal detectors, magnetic detectors, incandescent lamp detectors). Wave indicators, whose sen- sitiveness depends largely upon the pressure at the point of contact, are apt to require frequent readjustment.* For this skilled operators are needed and it is often the cause of poor service. As to the handling of the receiving apparatus, the use of a polarized relay involves considerable skill and care in making the adjustment and readjustments. In this respect the photographic recording receivers have the advantage; but with these, the string of the EINTHOVEN galvan- ometers requires equally careful adjustments. e. As to simplicity of the receiving apparatus, it is evident that receivers involving any moving parts operated by clockwork are at an inherent disadvantage. MARCONI'S magnetic detector has the additional disadvantage of occupying a large amount of space, which however is not so important in large land stations. The number of necessary apparatus, however, aside from the case of the coherer, which is disadvantageous from this viewpoint also, depends not so much upon the type of wave indicator as upon the object in view. If only telephonic reception is desired, the apparatus becomes as simple as possible, but recording the messages and using a call signal always * The fact that readjustment for maximum sensitiveness is possible with these wave indicators comprises an advantage from another point of view. For those indicators in which such readjustment is impossible may become worthless after a sufficiently severe atmospheric disturbance. 302 WIRELESS TELEGRAPHY complicates the receiving apparatus, no matter what kind of a wave indicator is used. The great simplicity and sensitiveness of the telephone receiver explains why this has become the rule, the recording receiver the exception, even though the latter has certain decided advantages. The sensitiveness and reliability of telephonic reception is largely dependent upon psycho- logical factors in the operator and is easily interfered with by external noises;*' 278 the reliability of a recording receiver depends only upon the excellence of the apparatus and it always furnishes a positive document of the received message. /. Formerly the possibility of applying a call signal and a recorder drew sharp lines between the various wave indicators. This distinction, how- ever, has gradually disappeared. The method of calling [Art. 169] intro- duced by the TELEFTJNKEN Co. seems to be adaptable to nearly all wave indicators of practical value. And as to recording received messages, there appear to be two methods applicable to all wave indicators, either by means of an EINTHOVEN galvanometer (photographic recording) or by means of the TELEFUNKEN sound intensifier, which latter, however, presupposes tone transmission. g. All the various wave indicators and recording devices are capable of responding to the speed of telegraphing obtainable by manual operation of the transmitting key or relay key. With the high speeds attained by means of automatic keys and rapid telegraph devices [Art. 117 c], the use of the metallic granular coherer, which involves the setting into operation of a series of mechanical apparatus is of course out of the question. So far as the author knows, the limit of permissible speed in transmission has not been reached with any of the other wave indicators used in practice. However, the limitation in the permissible speed of transmission is encountered in the recording receivers, particularly in the EINTHOVEN galvanometer (photographic) method, which seems to have responded to the highest speeds f used. * On airships, aeroplanes, etc., the attendant noises make telephone reception very difficult. For this reason the coherer has been returned to in some instances, in conjunction with a relay controlling the circuit of a small incandescent lamp. Relatively short and longer periods of incandescence in the lamp represent dots and dashes. 278 However, telephone reception has also been used with considerable suc- cess on flying machines. f The rapid telegraph apparatus of P. O. PEDERSEN, operating between the POULSEN stations at Lyngby and Esbjerg, is said to have attained a speed of 300 words per min. ; the normal speed of the POULSEN stations is given as 150 words per min. (The CULLERCOATS station transmits 200 words per min. over a distance of 800 km.) A speed of 100 words per min. has been achieved with MARCONI apparatus; the transatlantic MARCONI stations are said to operate at "quelques dizaines" words per minute. CHAPTER XII RECEIVERS 171. The Aerials at the Receiving Stations. The waves sent out by a transmitter result in an electromagnetic alternating field, which may be a rotating field, at the location of the receiver. Consequently if a conduc- tor is placed within this field oscillations are generated in it. The conductor for this purpose is the antenna, which also serves for transmitting, a complete station being equipped for both sending and receiving, 'just as in ordinary wire telegraphy. Usually a throw-over switch is provided for connecting the aerial either to the transmitter or to the receiver, as may be desired ; otherwise, it is arranged that the re- ceiver is automatically connected to the aerial whenever the station is not transmitting. 280 Following a suggestion of O. SauiER, 281 trees have been successfully used as receiving aerials for distances of about 50 km. The method is to hammer a nail into the tree a few yards above the ground and to connect the receiving apparatus between the nail and the ground. FIG. 365. FIG. 366. As to the direction of the aerial, it is advantageous to have this the same as the direction in which the electric field has its greatest ampli- tude. This depends upon the ground on which the station stands. In the one limiting case (namely, sea water), in which the field of the trans- mitted waves is a vertical alternating field (according to Art. 138), a vertical aerial is by far the best. In the other limiting case, with the station standing on dry ground, the direction in which the amplitude of the electric field is a maximum is apt to be at a considerable angle to the vertical (Art. 139, et seq.) ; hence an aerial inclined in the direction shown 303 304 WIRELESS TELEGRAPHY in Fig. 365 (arrow indicates direction of wave propagation) is materially more efficient than a vertical aerial or one inclined as that of Fig. 366.* 172. General Consideration of the Receiving System. a. The elec- tric field surrounding the receiving antenna produces an e.m.f., 8, along its length. This e.m.f. 163 8 = E . a 2 h 2 = Eh' 2 (1) where E is the component of the transmitted field strength in the direction of the aerial, h 2 the total and h' 2 the effective height of the aerial, a 2 being its form factor [Art. lOOc]. Consequently this e.m.f. for a given field, i.e., for the same transmitter, increases as the height and the form factor of the antenna are increased. In this respect, therefore, great height and large form factor offer the same advantages as for transmission; moreover, antennae which radiate freely are also advantageous for reception in this respect. b. However, from another standpoint high radiation is a disadvantage for reception; for, as soon as the receiving antenna begins to oscillate it radiates energy at a rate which increases as the radiation resistance increases. This radiated energy is lost to the receiver, which can make use only of such energy as is carried over to the detector. The conditions 282 existing in the receiving antenna are similar to those in any oscillator acted upon by an external e.m.f. [Arts. 56 and 67]. If the oscillator is tuned to the frequency of the external e.m.f. which is the only case we need consider in practice it will oscillate at its natural fre- quency, the amplitude of the oscillations growing constantly until the point is reached where the energy consumed in the receiver during one period = the energy supplied to it by the external e.m.f. (the transmitter) during the same period. Hence the maximum amplitude decreases as the energy consumption increases and therefore also as the radiation resistance of the antenna is increased. With undamped oscillations [Art. 676] the current 7 2 in the receiver is given by R 2 being the total resistance of the receiving antenna. Consequently the heat developed in a detector of resistance R, \ it'*) * Assuming the same constants as those on which Fig. 303 (page 254) is based, the aerial of Fig. 365, if brought to a vertical position would reduce the amplitude by 18 per cent, and if brought to the position of Fig. 366 would cause a reduction of about 66 per cent, (see the inclined aerials discussed in Art. 205). RECEIVERS 305 (#' 2 = effective resistance of the antenna without the detector). If the transmitted oscillations are damped (decrement = di) it follows from r> Art. 70, by substituting OAT 2 T [Art. Sd] for d 2 therein, that i__ . JL _!_ . s 2 R',Y 4N ^ So Now let us assume that the receiving antenna is so well constructed that the (JOULEAN) heat loss in its wires and in the ground is negligible compared to the radiation losses. Then R f z = R%. Moreover, let the detector resistance be at its best value, i.e., the value at which the heat developed within the detector and hence also the range are at their maximum. With undamped oscillations this value is R'z; with damped oscillations it would be rR'z, where r is somewhere between 1 and 2 for all important conditions encountered in practice. (If the transmitting and receiving antennae as well as the ground conditions at both points are the same, T = \/2 = 1.41.) Then the heat developed, Rdl Accordingly the greatest heat development attainable in the detector is entirely independent of the form and height of the antenna with undamped oscillations and is affected only very slightly, namely, through the value ~r, by these factors in the case of damped oscillations. It is increased, however, as the wave-length of the oscillations increases. 20 306 WIRELESS TELEGRAPHY c. Maximum heat development in the detector is a requirement for obtaining maximum range. The resultant increase in the decrement of the receiver may, however, be undesirable from other viewpoints (as for sharp tuning). Hence the energy consumed in the detector is usually left considerably below the possible maximum. Assuming it to be so low that Rd <^ R?, then from b we obtain that for undamped oscillations Rdl^eff = rA ' s 2 approximately K^ and for damped oscillations or, for undamped oscillations RJ^eff = 2(167T 2 X 10 10 ) 2 ' (W^T 2 " 8 2 C - ' S * UnltS and for damped oscillations d - f-rs EC? c.g.s. units. (167T 2 xlO 1 ") 2 4 X 3 X /, , A dl I 1 + dj In this case an antenna of low height and low form factor is at a great advantage and the importance of long wave-length becomes very marked (R. RUDENBERG 163 ). d. Nevertheless we must remember that the discussions in c and b take only incomplete consideration of the influence of the wave-length, in that the receiver only is considered. If we take the transmitter into account as well, the conditions become altered. According to Art. 138c and the second foot-note in Art. 139c Substituting this value in the equations obtained above, the heat devel- opment in the case of maximum range (Rd = R% and Rd = fRz resp.) becomes X10 10 for undamped oscillations, and * ai = form factor and hi = height of transmitting aerial; ft = coeff. of absorp- tion [Art. 1396] + stray field coeff. [Art. 1406]; /i = current amplitude at base of transmitting aerial. RECEIVERS 307 for damped oscillations ; while in the case of the best possible sharpness of tuning (Rd^ RZ) these values become 2 4W 2 ' 7 ' " /1 " 2 for undamped oscillations, and 3 4X3X10 10 /, , dA l I 1 + dj for damped oscillations. In the first case (maximum range) long wave-length with undamped oscillations is important only in that it is advantageous in regard to absorption [Art. 139/] and stray field [Art. 140]. In the second case (maximum tuning sharpness), however, long wave- length offers considerable additional advantages. Moreover in this case the combination of a freely radiating transmitting aerial with a weakly radiating receiving aerial would be materially superior to two similar aerials. e. According to d } with damped oscillations of constant frequency, the current effect in the receiver c ^ - The current effect Ii 2 e // a ^ di (1 + -r ) \ 2/ I\ 2 the base of the transmitting antenna <* - Hence the current effect in the receiver c e - L -r' , i It follows that, in making long distance tests "under the same condi- tions," it is essential that not only the current effect at the base of the transmitting antenna but also the decrement of the transmitter oscil- lations remain constant. It is not sufficient to simply keep the current effect at the base of the transmitting antenna constant. 1. THE ORIGINAL MARCONI RECEIVER 173. The First Arrangement. a. Fig. 367 shows the simple arrange- ment used by MARCONI in his first experiments. It is the exact counter- part of the original transmitter shown in Fig. 209, the spark gap of which is replaced by the wave indicator, which, in the original MARCONI equip- ment, was a metallic granular coherer. This arrangement, even if the coherer were replaced by, say, a thermal detector of very high resistance, would have the great disadvantage of too great resistance in the receiver and too large a decrement in the 308 WIRELESS TELEGRAPHY receiving antenna. If the equations of Art. 172 are applied to this case,* in which R d < R'z, we obtain approximately for undamped oscillations: 2R d So' and for damped oscillations RdPeff L. J R d 47V . ..o, feo i.e., the greater the resistance of the detector, the less heat will be devel- oped in it. Moreover, to the high resistance of the metallic granular coherer, there is added the difficulty that when unexcited it has a capacity effect, while when excited, it is simply a very high resistance. Hence the receiving antenna if tuned to the trans- mitter in one condition, can not be tuned for the other. The arrangement of Fig. 367 had still another \A disadvantage: It was easily affected by atmospheric disturbances. If the por- tion of the antenna above the coherer, through which it is insulated from ground, obtained only a slight static charge, this brought its potential differ- ence with the earth sufficiently high to break through and excite the coherer. 174. The Marconi Transformer. This last-mentioned difficulty was what chiefly induced MARCONI to soon remove the coherer from the aerial. He replaced it with a coil, Si, and caused the latter to act inductively upon another coil, >S 2 , having a much greater QJ number of turns and the ends of which ^ .^JL.^ wcre connected to a coherer (Fig. 368). f The transformer (SiSz) thus formed by * these two coils was called the S 2 / is formed. A large part of the energy in the antenna is then transferred to this * Assuming that natural oscillations of the antenna are still possible, t But see c of this article. RECEIVERS 309 To Relay FlG. 369. circuit S 2 Jj and the heat thereby developed in the coherer so reduces the latter's resistance that the relay responds. b. There is another point to be considered. With the coherer directly in the aerial, the use of multiple antennae gained nothing over the simple antenna. The use of several wires instead of a single aerial wire did not increase the potential across the coherer terminals, and the greater current ampli- tude, obtainable with the multiple aerial, did not help the coherer much. Now, however, it became possible to make use of the increased current amplitude of the multiple antenna, for with the transformer the increased current could be used to produce much higher potentials across the coherer than would be obtained in the antenna itself. To be sure, these advantages can only be secured if the antenna is tuned to the transmitter oscillations and the secondary circuit ($ 2 + coherer in unexcited condition*) is tuned to the antenna. The impor- tance of just this requirement was probably not recognized at the time; however, the fact that the entire arrangement operates satisfactorily only if certain requirements are filled, was recognized and pointed out by MARCONI from the first. The requirements were met by trying out in each station what was the best form of the transformer, which as a matter of fact consisted primarily in adjusting the primary and secondary fre- quencies (and perhaps also the degree of coupling). c. The arrangement of Fig. 368 can not be used j ust as shown there; for the coil $2 would close the relay circuit (see Fig. 359) even when the This is prevented by inserting FIG. 370. coherer was in its non-conducting state. a block condenser, C, (Fig. 369 or Fig. 370), which has no appreciable effect upon the oscillations if its capacity is sufficiently great [Arts. 30c and 41c]. The latter was not so essential, as a very close coupling was used. 310 WIRELESS TELEGRAPHY 2. RECEIVERS FOR TUNED TELEGRAPHY WITH DAMPED OSCILLATIONS The main object of tuned telegraphy is to have the receiver respond only to waves of a certain frequency (wave-length) and not at all, or at any rate only very slightly, to waves of any other frequency (wave-length).* The solution of this problem varies according as the receiving antenna is highly or slightly damped. 175. Receivers for Highly Damped Receiving Antennae. Such receivers are always constructed as to have a slightly damped secondary circuit coupled to the primary (antenna) circuit. The detector may be in the secondary circuit or it may be in either a condenser circuit or a closed circuit (detector circuit) coupled to the secondary circuit. To Relay FIG. 371. All these circuits are tuned to the transmitted frequency and hence are in resonance with one another. The following are a few of the many arrangements which are or have been in use, many of them being very similar in principle. a. Condenser Circuit Secondary; Inductive Coupling with the Aerial. This arrangement was used by MARCONI and with it he first demonstrated the possibility of tuned telegraphy, f It is shown diagrammatically in Fig. 371. Condenser C serves as a * This condition is more or less obtainable by simply loosening the coupling be- tween Si and 2 in the arrangements of Figs. 369 and 370 [see Art. ISOdJ; in fact these connections were used by MARCONI for tuned telegraphing. t Probably the first proposal to use tuned telegraphy was that of O. LODGE (Brit. Patent 11575 of 1897, applied for May 10, 1897). In this patent some of the re- quirements which an arrangement for tuned telegraphy must fill are clearly stated. LODGE, however, does not seem to have had any practical success until MARCONI completed his first successful experiments in tuned telegraphy. RECEIVERS 311 block condenser; as it has much greater capacity than condenser Ci, to whose terminals the coherer F is connected, the latter, Ci (in conjunc- tion with the coherer in parallel) determines the fundamental frequency of the condenser circuit [Art. 46]. FIG. 372. Of late, the MARCONI Co. makes use of a special tertiary circuit for the detector in its commercial stations. This so-called " multiple- tuning apparatus" of the MARCONI Co. is shown in Fig. 372.* The variable condenser at the upper left hand and the self-induction adjust- able in steps below the condenser, i serve for tuning the aerial. The variable condenser in the middle is part of the secondary or intermediate circuit, while that at the upper right belongs to the detector circuit. The self-induction of these two (secondary and tertiary) circuits is adjusted to the same step in both simultaneously. b. Condenser Circuit Secondary, Direct Coupling between Aerial and Condenser Circuit. This arrangement was used by LODGE and MuiRHEAD 283 with the granular coherer and by the TELEFUNKEN Co. (see diagram of connections, Fig. 373) with the SCHLOMILCH detector, when particularly sharp tuning was desired. The TELEFUNKEN Co. used a special tertiary condenser circuit (777, Fig. 373), as proposed by F. BRAUN, 284 containing the effective condenser C 2 and the block condenser of large capacity, C. * Courtesy of the MARCONI Co. To Relay FIG. 373. 312 WIRELESS TELEGRAPHY Fig. 374 illustrates a TELEFUNKEN receiver for thermal detectors on this same principle. The primary inductance (7 + 77 in Fig. 373) is divided into two parts. One part (marked "4" in Fig. 374) is coupled to the condenser circuit 777 (Fig. 373) containing the detector, while the other part (at the upper right hand in Fig. 374 and marked "EA") con- FIG. 374. sists of a coil of variable self-induction (RENDAHL variometer). The condensers (P, Fig. 374 = Ci, Fig. 373 and S, Fig. 374 = C a , Fig. 373) are variable plate condensers. The results obtained with the connections shown in Fig. 373 depend very largely on the relative amount of self-induction in 7 and 77 as com- RECEIVERS 313 pared with the effective self-induction of the rest of the antenna and of the " lengthening " or " loading" coils in it. If the self-induction of / and II is relatively small, as was the case in what has just preceded, the primary circuit must be considered as: Aerial, coils I and //, ground; while the secondary would be comprised of the condenser circuit Ci, coil I + //. But if the self-induction of coils / and II is relatively large, we have a case of the "fly-wheel" system, described in Art. 986, applied to the receiver. The primary circuit then consists of the con- denser circuit comprised by the inductance I + // (Fig. 373) and the capacity formed by the condenser Ci in parallel with the capacity an- tenna-ground. c. Single Coil Secondary. With this arrangement, in which the natural oscillations of coils [Art. 23] and not of condenser circuits is employed, A. SLABY and COUNT ARCO, following soon after MARCONI, succeeded in obtaining a tuned radio-telegraph system. It is now no longer in use. 176. Receivers for Weakly Damped Antennae. If the decrement of the antenna is not much different from that of a well designed con- FIG. 375. FIG. 376. 77777/777T/ FIG. 377. denser circuit without spark gap, then the use of a condenser circuit as secondary no longer offers the same advantages as with a strongly damped antenna [Art. 180d]. Hence, in this case, which applies to all quenched spark operation, the antenna is coupled to a closed detector circuit 285 containing the detector as shown in Fig. 375.* The coupling may be either inductive (Fig. 375) or conductive (Fig. 378). The TELEFUNKEN Co. 160 has applied this method of connection for use with transmitters arranged for two standard wave-lengths in the following manner. An inductance Si (Figs. 376 and 377) is always left in the receiving antenna, in which there is also a condenser C. When the transmitter is working on the short wave, Si and C are placed in series (Fig. 376), while for the longer wave they are connected in parallel (Fig. 377). In the latter case we again have the "fly-wheel" connection * C' is simply a block condenser of great capacity. 314 WIRELESS TELEGRAPHY [Art. 986]. A receiver built on this principle is shown in Fig. 236 (marked "33"); C is a variable plate condenser by means of which the receiving antenna can always be exactly tuned to the transmitted oscillations. 177. Tuning the Receiver for a Double Wave Transmitter. In Arts. 175 and 176 it was tacitly assumed that the transmitter furnished a wave of only one length. This is the case with the WIEN transmitter, but is true of the BRATJN transmitter only if the coupling between the primary and secondary circuits is very loose. If the coupling in the BRAUN transmitter is not very loose, two waves of different length are obtained. The question then at once suggests itself: Which wave shall the receiver be tuned for? 285a This question is justified from two standpoints, viz.: a. There is, firstly, the question per se as to whether it is better to tune the receiver for the longer or for the shorter wave. In Art. 106a, the reasons in favor of the shorter wave-length (higher frequency) were discussed. On the other hand, the fact remains that the shorter wave is more rapidly absorbed in the daytime than the longer wave [Art.139/] and that, moreover, the longer wave is more efficient in regard to produc- ing useful energy consumption in the receiver [Art. 1726]. As a matter of fact, however, it is universal practice to tune for the shorter wave, so far as the author knows. 6. Secondly, there may be some question whether it is best to have the receiver tuned exactly for the wave-length to which it should respond. If the receiver consisted of a single, slightly damped system, then [see Art. 87a] a certain definite small displacement from exact resonance (i.e., a slight dissonance) between the receiver and the transmitter oscil- lations should give the best results, at least in case the transmitter is quite loosely coupled so that its two waves are nearly of the same fre- quency. Even if the receiver consists, not of a single, but of two or three loosely coupled circuits or systems, it is very probable that the same holds true. Accordingly, it is not unreasonable that, with a not very closely coupled transmitter a slight displacement from resonance may be ad- vantageous, or, to put it more correctly, that well adjusted receiving stations really operate at a point slightly off exact resonance. 178. Adjustment of the Energy Delivered to the Receiver. Accord- ing to Art. 172, it is of great importance for the heat developed in the wave indicator and hence for the range of operation, that the energy delivered to the wave indicator has a distinct relation to the energy losses in the receiver. On the other hand, maximum sharpness in tuning [Art. 180] requires the lowest possible damping and hence minimum energy delivered to the wave indicator. Therefore, either one or the other requirement will be met according as the chief object in view is longer range or very sharp tuning. Or, otherwise, a compromise is made, the energy supply to the wave indicator being adjusted to give a good RECEIVERS 315 range, without allowing the sharpness of tuning to fall below the desired practical limit. The amount of energy delivered to the wave indicator is adjusted by varying the degree of coupling between the detector circuit and the antenna or the secondary circuit of the receiver. Figs. 378 and 379* show the method of arranging a conductive coup- ling of the detector circuit f direct with the antenna in Fig. 378 and with the secondary circuit, ABCi, of the receiver in Fig. 379. The coupling is varied by means of the sliding contact Sc. As the portion- A-Sc of C Telephone FIG. 378. FIG. 379. the coil AB is increased (or decreased), the current flowing through the detector Z and hence the action in the detector is increased (or decreased) while the damping is also increased (or decreased). For inductive coupling of the detector circuit, the arrangements shown in Figs. 375-377 can be used if the coupling between Si and S 2 is variable [Art. 54]. 179. Receivers for Two Different Detectors. In receiving stations where two different wave indicators (say, one for telephone reception, the other for call signaling or for recording) are to be used, it usually does not suffice to simply install a throw-over switch for connecting either wave indicator to the rest of the apparatus. Aside from the fact that this would limit the reception to one of the wave indicators at a time, it is advisable to have separate secondary circuits adapted to the individual requirements of each indicator. * This arrangement of circuits may be considered as dividing the current between the two parallel branches consisting of the self-induction A-Sc and the detector Z with its block condenser, C. f These connections were used by the TELEFUNKEN Co., in conjunction with the electrolytic detector. 316 WIRELESS TELEGRAPHY Recording The arrangement used by the TELEFUNKEN Co. for this purpose and illustrated in Fig. 380 will serve as an example. It requires little or no further explanation; the "tuning coil" and the variable condenser C serve for tuning the aerial. Fig. 381 82 shows the construction of the tun- ing coil, Figs. 382 82 and 383 82 are the coupling transformers for the record- ing and for the telephone re- ceivers respectively, arranged for adjustable coupling. Fig. 384 shows the entire outfit assembled as a unit. 180. The Sharpness of Tun- ing. If the frequency of the transmitter is changed, the effect upon the receiver will also change. Assume that a thermal wave indicator (e.g., a thermo- couple) is used in the receiver. Plot the deflections of the gal- vanometer in the circuit, which deflections are proportional to =_ ~\ To Telephone _~} To Relay FlG. 380. FIG. 381. the current effect in the detector, as ordinates and the different trans- mitter frequencies as abscissae. The resulting "resonance curve of the receiver" will be of the form of the heavier curve in Fig. 385; the effect is a maximum at a certain transmitter frequency, No, at which frequency the transmitter is said to be "in tune," while at any other frequency it is "out of tune." If now the galvanometer is replaced by a relay, the latter will not respond below a certain current. Thus, let us assume that under the conditions represented by the heavy curve in Fig. 385, the current at resonance is }{Q milliampere and that at least %Q milliampere is required to actuate the relay; then the relay will not respond at frequencies below 0.967 N Q or above 1.033 jV , i.e., at a dissonance of more than 3.3 per cent, in the transmitter. This 3.3 per cent, is RECEIVERS 317 FIG. 382. FIG. 383. FIG. 384, 318 WIRELESS TELEGRAPHY sometimes called the "necessary dissonance". Apparently, therefore, the "sharpness of tuning" varies inversely as the necessary dissonance.* a. The sharpness of tuning depends upon two factors, viz. : 1. The shape of the resonance curve (Fig. 385) and hence upon the sharpness of resonance [Art. 70c]. 2. The factor of safety [Art. 148] of the station. The relation to the form of the resonance curve is evident from Fig. 385. The steeper the curve, the sharper is the resonance, and hence the sharper will be the tuning. Thus, if, e.g., the resonance curve were the flatter, light curve in Fig. 385, then, under the same conditions as were 0.9 N \ , , 1.1 N _\ Dissonance in -10 98765432 1012345G7S 9 10 FIG. 385. assumed previously, the necessary dissonance would be about 6.5 per cent., the sharpness of tuning correspondingly less. As to the other factor which determines the sharpness of tuning, it was pointed out above that under the conditions assumed J^o milliampere was required to make the relay respond. When the station is tuned, i.e., under normal operating conditions yQ milliampere is supplied to the re- lay. Hence the station has a working factor of safety of \/3 H the working safety factor were lower, e.g., \/1.5 ; then under the conditions represented by the heavy line curve of Fig. 385 the relay would only re- spond within 2 per cent, of resonance, so that the tuning would be much sharper. From the preceding it is evident that record tests giving very great sharpness of tuning must not be considered as conclusive. By ad- justing a receiver so that the slightest deviation from resonance suffices to prevent the apparatus from responding as an indicator, the tuning appears to be, in fact really is, very sharp; but the station is entirely unfit for normal service. * The best measure of the sharpness of tuning is the reciprocal of the necessary dissonance value. RECEIVERS 319 &. As to the form of the resonance curve, this is easily determined for a receiver without secondary condenser circuit as used for weakly damped antenna oscillations. For if the conditions in the detector circuit are such that the current effect in it is proportional to that in the antenna [Art. 556], then the resonance curve is exactly the same as that corresponding to a primary circuit of decremented in the transmitter and a decrement dz in the receiving antenna and is determined by the sum of the decrements of the transmitting and receiving antennae. At the same time, the decre- ment of the receiving antenna of course depends also upon the amount of energy supplied to the detector. The resonance sharpness and, hence, also the sharpness of tuning in- crease as the damping of the transmitter oscillations and that of the re- ceiving antenna decrease. c. The resonance curve for receivers with secondary condenser circuit is easily calculated if the transmitter oscillations are undamped and if the primary and secondary circuits of the receiver are very loosely coupled. In this case, in a very short space of time only the impressed undamped 0,8 0.7 0.6 0.5 0.4 ' 0.3 0.2 '0.1 /I 7- \ \ \ 1.005 1.010 1.015 1.020 0.980 , 0.985 0.99 0.995 1 N/N t ~* FIG. 386. oscillations of the transmitter frequency exist in both primary and sec- ondary circuits of the receiver, and they almost solely determine the current effect [Art. 696]. A simple consideration of this shows that the resonance curve of the receiver is obtained approximately* in the following manner. Plot the resonance curve (the thin full line curve in Fig. 386) , which, according to the second foot-note of Art. 74a, corresponds to the * The exact equation for the resonance curve is : tor A/I + 320 WIRELESS TELEGRAPHY decrement of the receiving antenna with undamped oscillations (di = 0), the ordinates being the values of I 2 e f f /I 2 r effm Similarly, plot the resonance curve corresponding to the decrement of the secondary circuit of the receiver (dashed line in Fig. 386). Then find the product of the ordinates of these two curves corresponding to the same abscissa. This product is approximately* the value of the ordinate of the desired resonance curve (heavy full line curve in Fig. 386) at the same abscissa. In Fig. 386, d 2l (receiving antenna) =0.1, d 2z (secondary circuit of receiver) = 0.05. For d^ = 0.02 the dash-and-dotted line is obtained.* From the preceding, it follows that by the use of a secondary circuit a much sharper tuning is possible than without a secondary, the difference being the more marked the less damped the secondary circuit is. d. If the transmitter oscillations are damped, the conditions governing a receiver with secondary condenser circuit are quite different. In general two oscillations (of different frequency) are induced in the receiving antenna, one, the impressed oscillation, of the same frequency and dec- rement as the transmitter oscillation, the other the natural oscillation of the fundamental frequency and decrement of the receiving antenna and hence of the same frequency as the secondary circuit which is tuned to the receiving antenna. Consequently, even if the impressed oscillations have but little effect upon the secondary circuit, the natural oscillations of the receiving antenna will. 1 O.'J 0.8 0.7 0.6 0.5 X , 5 4 S 2 1 1 2 34 5 Dissonance > FIG. 387. The conditions encountered here are relatively complicated, as three damped systems (transmitter oscillations, primary circuit and secondary circuit of receiver) come into question, and moreover as two quite different requirements, viz., maximum resonance and sharpness of tuning on one hand, maximum range on the other hand, counteract each other in this case. * See foot-note on preceding page. RECEIVERS 321 So far as the resonance sharpness is concerned, we may assume that, other things being equal (equal decrements), it increases the looser the coupling between antenna and secondary circuit of the receiver is made. The ideal case, therefore, is that of extremely loose coupling. This has been theoretically investigated by H. RiEGGER; 286 some of his results are shown in Figs. 387* and 388A and B.* The conditions assumed for Fig. 387 are about those existing in the BRAUN transmitter with greatly damped antenna; decrement di of the exciting circuit in the transmitter and hence the decrement of the trans- mitted oscillations = 0.1 approximately; decrement J 2l of the receiving antenna = 0.3; decrement d^ of the condenser circuit in the receiver = 0.9 0.8 0.7 0.6 0.5 \ M 0.3 0.1 \ i o i Dissonance FIG. 388A. 0.03. The resonance curve (a) of the current effect in the condenser cir- cuit shows that the sharpness of resonance which can be attained (p = 50 approx. [Art. 70c]) is considerably greater than it would be without a secondary condenser circuit, with the antenna acting directly upon the detector. In this latter case the resonance curve would be as shown by curve b, the resonance sharpness would = 15.7, corresponding to di + d 2l = 0.4. The assumptions on which Figs. 388A and 3885 are based correspond to a quenched gap transmitter and two antennae with greatly reduced radia- tion damping: di = d 2l = 0.03. In Fig. 388A a relatively large amount of energy supplied to the detector by the condenser circuit (d2 2 = 0.03) is assumed, while in Fig. 3885, this is assumed to be very low (d^ = 0.01). As a means of comparison, the curve b, the resonance curve which would be obtained in the receiver without a secondary condenser circuit and cor- responding to di + d 2l = 0.06, p = 105, has been drawn in each figure. Here again it is seen that the secondary condenser circuit considerably * In these figures curve c is the resonance curve for d\ + d 22 ; it almost coincides with curve a. 21 322 WIRELESS TELEGRAPHY increases the resonance sharpness (p = 143 in Fig. 388A, p = 156 in Fig. 3885). Accordingly the use of a secondary condenser circuit even in con- junction with very slightly damped antennae, is justified when particularly sharp tuning is desired. 287 On the other hand, however, the examples illustrated show that the resonance sharpness attainable without a sec- ondary condenser circuit, suffices for all practical purposes and in fact would suffice even if the decrements of the antennae were double the decrements assumed for Figs. 388A and B. The range is determined on one hand by the energy supplied to the detector (and hence by the damping, d? 2 , of the secondary condenser cir- 0.1 V-& 54 2 21 012 345$ Dissonance > FIG. 3885. cuit) and on the other hand by the degree of coupling between the re- ceiving antenna and the secondary condenser circuit. The practical problem therefore is : how far may or must we go with both these factors to obtain maximum range without seriously reducing the sharpness of tun- ing?* Such investigations 288 as have been made to date do not suffice for arriving at a general answer to this question. Actual experience in prac- tice has shown that in those cases where there is any condenser circuit in the receiver, the coupling between condenser circuit and receiving antenna must in any case be very loose, if good tuning is at all required. f It has further been shown that this loose coupling may be adopted without mate- * A comparison of the two curves marked "a" in Figs. 388A and B respectively, is instructive in this connection. The sharpness of resonance is almost the same in both cases although the energy supplied to the detector in Fig. 388A was assumed about three times as great as that in Fig. 3885. t Recognition of this requirement originated in the theoretical investigations of M. WiEN 289 and in the experiments made by H. BRANDES and L. MANDELSTAM 289 at almost the same time. Close coupling is used almost solely for such cases where sharp tuning is of no importance and where it is desired to communicate with various stations of somewhat different wave-length, as for instance in coastal stations com- municating with ships at sea or again where it is desired to "listen in" to traffic between other stations [see Art. 184aj. RECEIVERS 323 rially sacrificing range, as long as the transmitter oscillations are not too strongly damped. The reason for this is that conditions in the loosely coupled system (secondary or tertiary circuit of the receiver) which acts directly upon the wave indicator, are about of the nature described in Art. 61 c more and more energy accumulates in this system during a series of periods or cycles, so that eventually quite a large amount of energy exists in the system, even if only very little is transferred to it in each period. This, however, is based on the assumption that all energy consumed in the secondary circuit (JOULEAN heat, eddy currents) without being useful in the wave indicator, is kept as small as possible. Otherwise, the use of a secondary circuit in the receiver may be detrimental to the range without being of much value toward sharpness of tuning. It may therefore be important to block the path of the oscillations into the circuits of the auxiliary apparatus where a part of their energy would be wasted, by means of choke coils [Art. 1656]. If this is done, however, it is essential that the choke coils themselves do not consume any energy;* hence they must have no iron cores. With iron cores, they would serve their purpose of keeping the oscillations out of the auxiliary apparatus fully as well if not even better, but hysteresis and eddy current losses in the cores would result. 181. R. A. Fessenden's Method for Maintaining Secrecy of Tele- grams. 290 The "secrecy sender" of Fig. 389 transmits waves uninterrupt- edly, but when the circuit of the wire loop, K, is closed, their wave-length, X', differs from X, the wave-length to which the receiver is tuned, by an amount given by FESSENDEN as Y per cent. If the circuit of this wire loop is broken by pressing the keyf the transmitter oscillations have the wave-length, X, for which the receiver is tuned. { At the receiver ("interference preventer") (Fig. 390) the oscillations in the aerial branch off between two paths, ACiSiE and ACzS^E. The former is tuned to the wave-length X, the latter being so dimensioned that with the wave-length X' the amplitude of the oscillations in C 2 S 2 becomes equal to that in CiSi* The coils S\ and S'%, which are coupled with Si and $ 2 respectively, are wound so as to oppose or "buck" each other, so that with wave-length X' the electromotive forces induced in S\ and $' 2 , practically neutralize each other. Hence, if the transmitter is operated without depressing the key in loop K (Fig. 389) and wave X' is sent out, no appreciable oscillations are induced in the circuit CS'zDS'i of the receiver. But if the key in loop K * The construction of really good choke coils is not so very simple a matter. Accordingly systems or methods of connection in which no choke coils are needed offer a practical advantage. t The key is not drawn correctly in Fig. 389. t Compare P. O. PEDERSEN'S method for undamped oscillations [Art. 127 c]. 324 WIRELESS TELEGRAPHY is closed and wave X is sent out, oscillations of very high amplitude are obtained in branch CiSi and of very low amplitude in C 2 $ 2 ; consequently the electromotive forces induced in S'i and S'z do not neutralize each other, and the wave indicator, D, responds accordingly. Undoubtedly this method makes the reception of telegrams very diffi- cult. Unless the receiving station is tuned exactly for the wave-length X and so sharply that a dissonance of % per cent, suffices to make the sig- nals disappear, the signals will be received constantly, whether the key at the transmitter is depressed or not. f FIG. 390. The practical tests conducted by the NAT. EL. SIG. Co. with this method were claimed to have given very good results, even in overcoming atmospheric disturbances [Art. 183]; its application, however, will prob- ably remain very limited to a few special cases. 182. Multiplex Telegraphy. The solution of the problem of construct- ing a receiver which will respond within certain limits only to a single wave-length, is at the same time a solution of the problem of multiple telegraphy receiving telegrams from two transmitters simultaneously on one antenna. a. Fig. 391 illustrates an arrangement of this kind used by MARCONI with considerable success. For the longer wave, the primary circuit consists of the aerial, coil S, primary coil I of the transformer and ground. The secondary circuit tuned to this is II. For the shorter wave, the pri- mary circuit consists of the aerial, condenser C, primary coil/' of the trans- former and ground. The secondary circuit tuned to this is II' . The RECEIVERS 325 conditions are such that the system to the right does not respond to the longer wave, that to the left, not to the shorter. Any other arrangement for tuned telegraphy can, of course, be simi- larly used if the tuning is sufficiently sharp.* b. The simultaneous transmission of two telegrams from the same an- tenna is also feasible. It is simply necessary to couple two different con- denser circuits with the aerial, each condenser circuit adjusted so as to be FIG. 391. in resonance with its own secondary. The connections used by MARCONI are shown in Fig. 392; as in Fig. 391, the portion to the left is for the longer wave, that at the right, for the shorter. c. In the multiple or duplex telegraph systems just described it is essential that the wave-lengths of the two transmitters whose telegrams are to be received on the same antenna, be different. Duplex reception at the same wave-length is possible if both the transmitters are tone trans- mitters and work with different tones. Tests of this kind have been made by the TELEFUNKEN Co.; 292 two of the sound intensifiers described in * Thus the TELEFUNKEN Co., e.g., has received telegrams from three different stations simultaneously on a single ship's antenna. 291 326 WIRELESS TELEGRAPHY Art. 1666 were connected to the receiver, each adjusted to the tone of one of the transmitters. Perfect duplex reception in spite of equal wave- lengths was possible; as soon as the two tone frequencies differed by 20 per cent. 183. Methods for Overcoming Atmospheric Disturbances. a. The atmospheric disturbances which are particularly frequent during the sum- mer months, and are especially noticeable in the hours from noon or from sunset to sunrise, even in stations where static charging of the antenna is FIG. 392. out of the question, seem to originate primarily in lightning discharges between two clouds or between a cloud and the earth. 293 This is not contradicted by the frequent disturbances experienced under a clear blue sky; the distance over which clouds can be seen from a point on the earth's surface is extremely short as compared to the distance at which a stroke of lightning can excite a wave detector. Hence an electric storm makes itself felt in a radio-receiving station at tremendous distances. The early wireless stations, as long as their receivers were arranged for relatively highly damped waves of great amplitude, suffered severely from these atmospheric disturbances, particularly in the tropics. Con- siderable improvement resulted as soon as the receivers were arranged RECEIVERS 327 FIG. 393. for less damped transmitter oscillations of lower amplitude (i.e., low an- tenna decrement, loose coupling with the secondary condenser circuit or detector circuit). Even to-day probably the best protection against atmospheric dis- turbances still is a powerful transmitter permitting ( the use of very loose coupling and a not too highly sensitive wave indicator in the receiver. a. MARCONI has devised a number of special arrangements for mitigating the effect of atmospheric disturbances. 294 1. The primary circuit of the receiver consists of the aerial and PCE (Fig. 393). The natural oscilla- tions of the aerial are so regulated* by means of coil S and condenser C, that the anti-node of current and node of potential occur at the point P [see Art. 31 et seq.]. Accordingly if waves of the same length as the natural wave-length of the antenna strike the latter it will oscillate with a potential node at P. If now a ground connection PE\ is made at P, no appreciable current will flow through it. But, if any other electromagnetic disturbance occurs, the greatest part of the current induced in the aerial will flow through PEi to ground as its impedance is lower than that of the path ~SCE. Hence the effect of the disturbance upon the secondary circuit (//) is greatly diminished. Accounts of how successful this arrangement is in practicef have never been published so far as the author is aware. 2. Another method of the MAR- CONI Co. is sketched in Fig. 394. Z>iZ>2 are two rectifying detectors, of opposite polarity, so that one allows the current to flow through it in one direction, the other in the opposite direction [Art. 162a]. For one of them, let us say Di, the size of the auxiliary battery (not shown in Fig. 394) is so chosen that the detector is very sensitive, while for the other, Z) 2 , it is so chosen as to make its sensitiveness very low. Consequently under normal operating conditions only DI responds and the telephone * The aerial is of course also tuned to the transmitter. t It seems probable that this arrangement would also be effective against electro- magnetic waves of another length, hence would increase the sharpness of tuning. - D FIG. 394. 328 WIRELESS TELEGRAPHY receives current in one direction only. But if a heavy atmospheric dis- turbance occurs, both detectors respond, the current flows through in both directions and the telephone is not affected. b. Those radio-systems which produce a tone of more or less purity in the receiving telephone have proven themselves as an excellent safeguard against atmospheric disturbances; for the latter are heard in the telephone as short dissonant crackling and can usually be easily distinguished from the tone signals. 184. Achievements of Tuned Telegraphy. The advantages* which make tuned telegraphy so decidedly preferable, are best expressed as the following disadvantages of untuned telegraphy: 1. The telegrams can be " picked up" by any and all stations within the range of the transmitter : no secrecy of telegrams. 2. Communication between two stations A and B can be crippled by constantly sending out signals from a station C within whose range A and B are located: deliberate intentional interference. 3. If A and A' on one hand and B and B' on the other hand are two sets of communicating stations, each of which lies in the range of the other three stations, then A and A' can not communicate while B and B r are exchanging messages : interference between stations. Whether or not tuned telegraphy entirely overcomes these obstacles can not be stated for all cases, as the distance between the stations in question and their ranges are very important factors. The question can only be: To what extent does tuning overcome these obstacles and are they entirely removed in any specific case? a. As regards the maintenance of secrecy of messages, let us consider the following possible case. A transmitting station, A, and a receiving station, A', are arranged for continuous communication with each other. Another station, C, is no further from A than A'. The question " can C be prevented from receiving telegrams sent out from A by means of the tuning methods previously described?" must be answered by a decided "no." If A and A' are arranged for constant operation their actual (ultimate) range must be much greater than the distance between them [Art. 148] and the wave indicator used must not be too highly sensitive. It follows that it will then be possible to receive the telegrams by means of a very sensitive wave indicator in an untuned closed detector circuit [Art. 176]. In general, the ordinary receivers will serve the purpose if the coupling is made closer. Thus, the TELEFUNKEN receiver described in Art. 176and illustrated in Fig. 236 is specially arranged for this. The coil correspond- ing to S 2 in Figs. 376 and 377 is movable so that its coupling with Si and hence also the coupling between the antenna and the detector circuit can be varied. In order to tune for any transmitter which is sending out * Aside from the increased range obtained by tuning. RECEIVERS 329 signals the procedure is as follows : Starting with very loose coupling, gradually make the coupling closer until a sound is heard in the telephone. Then adjust the condenser until the sound in the telephone is a maximum. Finally loosen the coupling again very gradually, readjusting the con- denser (C in Figs. 376 and 377) if this is necessary, so that maximum loudness is obtained in every case. 295 The picking up of messages by stations other than those intended to receive them, is made more difficult according as the amplitude of the oscillations required for the given distance is reduced, by decreasing the damping of the oscillations. b. Similarly in regard to intentional interference, 2950 assume the dis- turbing station C to be as near to the communicating stations AI and A 2 as these are to each other and that all are normal types of stations of moderate ranges. First, then, we must take for granted that station C can determine* the wave-length AI and A 2 are using and that C tunes its transmitter to give the same wave. C is then in a position to interfere with AI and A 2 even if its range is only one-half or one-third of that of AI and A 2 .f Eliminating this case, however, let us assume that C is unable to determine the wave-length used by AI and A*, so that C's wave-length differs considerably from that of AI and A 2 . Whether or not C can inter- fere in this case depends simply upon how far it can raise its amplitude. If the receivers at AI and A 2 have very loose coupling, C would not be able FIG. 395. to reach a sufficiently great amplitude in its transmitter oscillations to succeed in its purpose. { Under these conditions, therefore, tuned re- ceivers provide a much greater protection against intentional interference. c. In regard to the prevention of interference between a number of stations in the same general vicinity, let us consider the following extreme case. Assume two stations AI and A 2 very close together at one place, BI and B 2 similarly located at another place (Fig. 395). Then we must distinguish clearly between the following two cases: 1. The two stations at one place, say AI and A 2 operate as transmit- ters of equal strength, while those of the other place BI and B 2 are both receivers (Fig. 395). Then, by suitable tuning methods, it can un- * Any wave meter employing a wave indicator is suitable for this purpose. Most wave meters are arranged so as to be suitable for measuring waves coming in from a distance. t In view of the safety factor with which AI and A 2 must operate for constant service. t That is, -unless C could come very near, to either AI or A 2 . 330 WIRELESS TELEGRAPHY doubtedly be arranged that BI receives only AI'S telegrams and B 2 only those from A 2, even if the frequencies of AI and A 2 differ by only a few per cent. This in itself constitutes a great advantage for tuned telegraphy. 2. If, however, one of the stations must receive while its neighbor is transmitting, the conditions are quite different. Thus let AI and B 2 be transmitters, while BI and A 2 receive (Fig. 396). Everything now depends upon the distance of AI from A 2 and of BI from BZ* If this distance is only a small part of the wave-length, it will be impossible for A 2 to get the telegrams from B 2 without hearing the signals from its neighbor A i whose waves have a tremendous amplitude at so short a distance from the transmitter. But if the stations AI and Az, as well as BI and B 2 , are relatively far apart, then of course service between the two pairs, AiBi and B 2 A 2 , can be maintained without mutual interference. Just how far apart the neighboring stations must be depends upon the ranges of the stations, the difference between their wave-lengths, the sharpness of tuning of the receivers [Art. 180] and also upon whether the transmitters are single or double wave transmitters, the former being decidedly more advantageous.* If, even to the present day, frequent complaints of interference between stations are still heard, 296 imperfect design of the transmitters (high damping) and receivers is undoubtedly largely responsible for this. It must be remembered that with the great number of shore and ship stations now in operation it would have been impossible to maintain even a passable service using the old methods, whereas with modern systems the service on the whole presents no great difficulties. d. Stations arranged for tone transmission and operating on the acoustic or mechanical resonance principle [Art. 185] are least affected by interference. For here interference need really be feared only if the disturbing transmitter has the same tone as well as the same wave-length. 185. Methods for Preserving Secrecy of Messages. The fact that tuning does not in itself suffice to guard the secrecy of messages is a great disadvantage! for many purposes (as in army and navy work). * The NAT. ELEC. SIG. Co. (FESSENDEN) makes the following guarantee: Given three stations AI, A z and B 2 of equal range. If the distance AiAz is 1 per cent, of the distance A 2 -B 2 , and the wave-lengths differ by 3 per cent., A 2 will not be disturbed by AI. In fact with standard sets a difference of % P er cent, in wave-lengths is claimed to be sufficient to prevent interference. Reports of tests indicate that this company's apparatus really gives very fine results in this respect. 296 " t On the other hand this is a direct advantage for distress calls at sea, where it is important that as many ships as possible hear the call for help. RECEIVERS 331 The interception of messages by stations other than those called, can be prevented to some extent by telegraphing so rapidly that such relays as are customarily used will no't respond and only specially trained operators will be able to read the messages in the telephone.* Further- more the apparatus can be so arranged that the wave-length is easily and rapidly changed and then vary the wave-length in accordance with a prearranged program, perhaps automatically . f This method makes it very difficult for an uncalled listener to tune his receiver to the rapid variations, but it is of no avail against untuned, highly sensitive receivers. Probably all such methods as those described must be regarded as more or less makeshifts, to be used only when absolutely necessary and which are successful only in special cases. The following, however, are im- portant effective methods for providing secrecy. a. A galvanometer whose natural oscillations are slightly damped (about like the WIEN vibration galvanometer) or a telephone having a diaphragm whose natural oscillations are slightly damped or, again, a telephone combined with a closed spherical resonator 2970 is used in the receiver. These respond well only if the frequency of the interruptions in the transmitter is the same as their own natural periodicity. This is mechanical tuning. % The "sound intensifier" of the TELEFUNKEN Co. with its oscillating armature [Art. 1666] also belongs to this class of apparatus. In all such arrangements assuming that the oscillating mechanical system is tuned to the discharge frequency of the transmitter, the curve of the oscillations is like that shown in Fig. 135; i.e., the amplitude of the oscillations rises gradually, first reaching its maximum after several periods, the number of which depends upon the decrement of the oscillat- ing system; both this number of periods and the maximum amplitude increase as the decrement decreases. Herein lies the explanation of why in all cases of such mechanical tuning the sensitiveness of the arrangement depends upon the rapidity of operation (i.e., of telegraphing). For, in order to take full advantage of the sensi- tiveness, every signal must last long enough for the oscillating system to attain its maximum amplitude. If the telegraphing is done so rapidly that the duration of the individual signals is not sufficient to reach the maximum amplitude, the sensitiveness will be correspondingly reduced. The decrement remaining constant, the time required by the oscillat- ing system to reach its maximum amplitude increases as the period lengthens, i.e., as the discharge frequency is reduced. For this reason, such devices for mechanical tuning were of little practical use as long as it * This method was tried at one time by the MARCONI Co. and by the DE FOREST Co. t This method was adapted by the TELEFUNKEN Co. at one time. J The first proposal of such a method was probably made by A. BLONDEL. 332 WIRELESS TELEGRAPHY was customary to work with low frequencies, as this greatly limited the permissible rapidity of operation. The adoption of high discharge frequencies in the transmitter has made the use of mechanical resonance in the receiver possible without any great detriment to rapidity of signaling. Nevertheless, even to-day the use of mechanically resonant receivers in conjunction with automatic transmitters operating at very high speeds offers great difficulties. b. Another method has been proposed repeatedly from the earliest days of wireless telegraphy. It is based upon transmitting each signal, say each MORSE dot, not as a single discharge, but as a series of periodic discharges occurring at certain fixed equal intervals. The receiver is then so adjusted that it will respond only to oscillations occurring at these definite intervals. Probably the only apparatus of this kind which were used in practice were those of ANDRES BuLL 298 and of HovLAND. 298 They were rather complicated and will not be described in detail here. But it should be pointed out that in practical tests these apparatus gave good results. There can hardly be any question that these apparatus, when properly designed and constructed for reliability in operation, provide an almost perfect protection not only against the "picking up" of messages by stations not called or intended to receive them, but also against atmospheric dis- turbances; on the other hand it is just as true that their complication limits these apparatus to certain special work. 3. RECEIVERS FOR UNDAMPED OSCILLATIONS 186. General. For recording reception, for which thermal and crystal detectors and the EINTHOVEN string galvanometer (photographic method) are generally used, conditions are much the same for undamped as for damped oscillations. The secondary circuit of the receiver is made as slightly damped as possible, is loosely coupled to the antenna [see Art. 175] and may react in any way upon the detector. But for telephone reception a decided difference is encountered between damped and undamped oscillations; the arrangements for receiving damped oscillations, described in Art. 165, can not be used without modification for undamped oscillations. For in telegraphing a dash of the MORSE code, the excitation of the wave indicator would displace the telephone diaphragm from its normal position at the beginning of the dash, causing a click to be heard and nothing more, as the telephone diaphragm remains displaced in a fixed position just as long as the waves from the transmitter keep coming in and the wave indicator remains excited. Hence dashes and dots could not be distinguished as both would be heard simply as clicks. This difficulty can be overcome by sending the oscillations out in a RECEIVERS 333 series of "wave trains" obtained by means of a kind of interrupter in the transmitter. It is much simpler, however, to provide the interrupter at the receiving end, using it to alternately make and break the connection of the wave indicator to the oscillating circuit. Then the telephone diaphragm is displaced at each "make" and returns to its normal zero position at each "break." That is, the motion of the diaphragm has the same frequency as the interrupter. Consequently as long as waves strike the receiver, the tone of the interrupter is heard in the telephone, being audible for relatively long and short periods as MORSE dashes and dots are transmitted. In telegraphing with damped oscillations, an interrupter would like- wise be needed if the discharge frequency were above that (several thou- sand per second) of easily audible sounds. This condition is easily ob- tained with quenched spark gaps and D.C. operation, but to the Author's knowledge, has never been used in radio-telegraph practice, being re- stricted to radio-telephony. 187. Methods Employing the Ordinary Detector. a. Fig. 397 illus- trates diagrammatically one of the arrangements used for the reception of undamped oscillations (V. POULSEN, C. LoRENZ 299 ). The condenser circuit drawn in heavy lines is the secondary of the ^" receiver; it is as slightly damped as possi- ble and very loosely coupled to the antenna. The interrupter, U, which I I I C 7 operates on the principle of the electric bell or buzzer, alternately connects and JL disconnects the detector and its auxili- -p^ 007 i 1 JG. oy < . aries to and from the secondary circuit several hundred times per second. This arrangement can of course be varied in a great number of ways. 6. The interrupter used in this method is something more than a necessary evil. It has a certain decided advantage. It has been shown that it is of the greatest importance, both for the sharpness of tuning [Art. 180c] as well as for range [Art. 676] to have as little damping as possible in the secondary. As long as the detector is connected to the secondary, the energy (as long as it remains in the vicinity of the critical value [Art. 1626] of the detector) which is not converted or only partly converted into direct- current energy, is consumed in the detector. In the auxiliary apparatus and their connecting leads energy losses can hardly be entirely eliminated, in spite of the insertion of choke coils. But when using an interrupter, no such loss can occur in the wave indicator or in the auxiliary apparatus whenever these are disconnected. The amplitude of the oscillations in the secondary circuit and hence the To Telephone 334 WIRELESS TELEGRAPHY energy stored in it, rise to a very high value. Then when the interrupter connects the wave indicator into the circuit, it is acted upon by an oscilla- tion of very great amplitude and almost all the energy accumulated in the secondary circuit is used for exciting the wave indicator. POULSEN seems to have succeeded in reducing the decrement of the secondary circuit to 0.003 in this way. 300 This could not be attained if a wave indicator with its accessories were continuously in circuit. & A further advantage of the interrupter is the possibility of pro- ducing a musical tone in the telephone by suitable connections and so obtain to some extent the same advantages secured by means of a tone transmitter with damped oscillations. 188. The Ticker. 301 a. The so-called "ticker" method devised by POULSEN, in which none of the wave indicators described in Chap. X is used, is shown in Fig. 398. Here U is the interrupter and the secondary circuit proper is drawn in heavy black lines. C f is a very large condenser of several tenths of a microfarad capacity, while C has only a few thousandths of a microfarad T capacity. Many modifications of this ar- rangement have been devised.* b. The basic idea in POULSEN'S arrange- ments (Fig. 398) is as follows : As long as -p _ condenser C f is disconnected from the oscillating circuit CSz, the latter accumu- lates a relatively large amount of energy. Then, when the ticker connects the large condenser C' in parallel to the small condenser, C, C r takes the major part of the current and of the stored energy, so that it obtains a relatively high charge, which upon discharging through the telephone, T, causes a click to be heard in the latter. Even though the procedure may be somewhat more complicated in its details than here outlined, the essential features of what occurs are as just described. c. The sensitiveness of this arrangement for telephone reception seems to be greater than for methods using any of the best wave detectors described in Chap. X. The latter all have a low efficiency, i.e., the direct-current energy delivered by them is only a small fraction of the high frequency energy supplied to them [Art. 1626]. Too great a loss is involved in the double transformation, first from electrical energy to * The following arrangement (TELEFTJNKEN Co. 302 ) is very interesting. The interrupter U in Fig. 398 is replaced by a rectifying detector allowing current to flow in one direction only. The interrupter is inserted between C' and the telephone T. The unidirectional current flowing through the detector charges condenser C', which can not discharge through S 2 on account of the detector. As the interrupter alter- nately connects and disconnects the telephone T to and from condenser C", the latter discharges through the telephone when they are connected and is recharged when T is disconnected. RECEIVERS 335 heat and then back from heat to electrical energy. In the ticker, on the other hand, there is practically no real energy transformation, for the charge momentarily stored in condenser C" (Fig. 398) is directly dis- charged through the telephone. Occasional irregularities in the inter- rupter, causing the "make" and "break" to occur at the wrong instant, will have but little effect in reducing the efficiency if the condenser circuit CS 2 (Fig. 398) is only slightly damped. With good construction of the interrupter, its operation is said to-be very regular and reliable. On the other hand, it does not seem possible to obtain a pure tone in the telephone and so secure the same freedom from atmospheric disturbances with the ticker as is obtainable with the tone transmitter [Art. 1836]. This constitutes a serious disadvantage, particularly in the tropics. 189. Construction of Interrupter for Ticker Method. The construc- tion of a good practical interrupter for use in the ticker systems, is not so simple as might at first sight appear. a. The arrangement, which is probably in widest use at present, is shown diagrammatically in Fig. 399 : b is an electromagnet facing the small armature c, which is vibrated, just as in an electric bell or buzzer, by a battery connected to the winding of the magnet. The spring d, on which the armature is mounted, is fastened to a plate having a slight spring to it. Resting on this plate is a small piece of metal, c, to which a fine gold wire, /, is at- tached. This gold wire together with the small adjustable wire brush, g, constitutes the ticker con- tact. The sensitiveness of this arrangement can be in- creased by using a telephone diaphragm of low damp- FIG. 399. ing to whose natural oscillation the ticker is tuned [see Art. 185a]. b. L. W. AusTiN 303 has described a rotary interrupter which is said to be particularly well suited for use in conjunction with the ticker. It consists of a highly polished copper or nickel disc which is kept in rotation, while a fine copper wire brushes against the disc under very light pressure. 190. Special Arrangements for Undamped Oscillations. a. The heterodyne, receiver of R. A. FESSENDEN. 304 The principle of this receiver is well illustrated by the following description of one form in which it has been constructed. The telephone, instead of having the usual permanent magnet, has a core of fine iron wires within a winding and instead of an iron diaphragm, has one of mica carrying a coil of fine wire. The oscillations induced in the receiver by the incoming waves are led through the coil in the diaphragm. A high frequency current, whose 336 WIRELESS TELEGRAPHY frequency, N f , differs somewhat from that, N, of the incoming oscillations is sent through the winding of the electromagnet. The force* exerted upon the diaphragm coil by the field of the iron wire core varies periodically. Consequently, the telephone diaphragm oscillates at a frequency which = N N f , as will be evident from a simple consideration of the facts. Hence, if N and N r are so chosen that the frequency equal to their difference lies within the range of audible tones, this tone, N N', will be heard in the telephone. b. R. GoLDSCHMiDT's 305 method. The principle upon which this method is based is easily understood from a consideration of Art. 122. The oscillations, of frequency N, which are induced in the receiver by the waves from the transmitter, are led through a fixed coil (S in Fig. 261). In the revolving field of this coil there is a movable coil, R, which rotates at N' revolutions per second. Then there will be induced in this movable coil a current of frequency N N f , which under proper conditions can be heard in a telephone, even though N and N' individually lie far outside the range of audible tones. In practice, of course, the fixed and movable coils, S and R, are re- placed by the stator and rotor respectively of a high frequency generator. By adjusting the speed of the machine, the rotor currents are made audible in a connected telephone. 191. Practical Achievements. a. The question, to what extent tuning the receiver can prevent disturbance from other stations and secure privacy of messages when working with damped oscillations was discussed in Art. 184. The question now arises whether the use of undamped oscillations will materially alter these conditions. In regard to securing secrecy of messages it is evident from Art. 184a alone, that undamped oscillations have a great advantage over damped oscillations; for the lower the amplitude required to attain a given range, the more difficult it becomes to "pick up" a telegram. For this same reason it would seem that undamped oscillations should also provide a greater protection against intentional disturbance by other stations. Actually, however, this advantage is not very great when compared with well-designed stations operating with damped oscillations. b. In regard to interference between two stations (and the same applies to the use of undamped oscillations in multiplex telegraphing), we would expect that undamped oscillations, in view of the very loose coupling in the receiver and the very low damping of the secondary cir- cuit, would offer very decided advantages and secure particularly sharp tuning. And this would undoubtedly be the case if the same conditions * Or rather, to be more exact, its mean value during one period. Of course, this force varies continuously during each period, but these rapid variations do not come into consideration for the motion of the diaphragm. RECEIVERS 337 obtained in the transmitter as with damped oscillations. But in practice the frequency of the undamped oscillations is never quite constant, whether they are produced by a high frequency alternator or by an arc generator. As to just how much the frequency has been found to fluctuate in the high frequency alternators which have been built to date, nothing has been published so far as the author knows. With the arc method, it seems that a sharpness of tuning fully as good as, but not better than the best obtainable with damped oscillations can be secured;* and there is no need for any still greater sharpness. * P. O. PEDERSEN 186 states that in the method of transmitting used by him [Art. 127 c] a dissonance of ^ per cent, in the transmitted wave sufficed to prevent recep- tion. This would indicate a very great sharpness of tuning. 22 CHAPTER XIII DIRECTIVE TELEGRAPHY 192. Characteristic of the Distance Effect. The object in view in "directive telegraphy" is to so confine the radiation of waves from the transmitter to a narrow or rather an acute angle, that only receivers located within this angle will be in the path of the waves. Actual ac- complishment, so far ; consists in transmitters whose waves radiating in different directions have widely differing amplitudes. a. The following method is convenient for obtaining a picture or "curve" of the power of any given transmitter to direct its waves. The amplitude of the waves is measured from point to point on a circle (of suitable radius) whose center is at the transmitter and the values so FIG. 400. obtained are plotted as vectors in the directions or angles corresponding to each amplitude (Fig. 400). The curve obtained by joining the ends of the vectors is the "characteristic of the distance effect" and gives a simple picture of the usefulness of the particular transmitter for directive signaling. It is self-evident that the characteristic of all symmetrical, vertical transmitters is a circle. If the characteristic is of the form shown in Fig. 401, the obvious conclusion is that the transmitter in question emanates waves in all directions, but its effect in the direction SB is considerably less than in all other directions. The case illustrated in Fig. 402 is much more desirable for directive transmission; no waves are sent out in the direction SB, practically all being concentrated in the direction SA, so that in directions diverging even only very slightly from SA, the effect is very much less. A transmitter having this characteristic would be a practical solution of the problem of directive telegraphy; its effect would be confined to an extremely acute angle. 6. For detectors which react upon the current effect, reception de- pends, not upon the amplitude of the waves, but upon the square of the 338 DIRECTIVE TELEGRAPHY 339 amplitude. Hence, to obtain a picture of the distance effect of a trans- mitter with respect to a receiver of this type the squares of the wave amplitudes in the different directions must be plotted as vectors in the diagram. The characteristic of amplitude squares is distinguished from that of the amplitudes in that it is much less like a circle in form* and, therefore, is much better suited to the purpose in view. It follows that detectors which react upon the current effect, as, e.g., thermal detectors, FIG. 402. FIG. 401. are much better adapted for directive signaling than those whose action depends upon the amplitude (first power) of the oscillations [see Art. 163a]. In what follows, the simple characteristics (first powers of the ampli- tudes) are plotted throughout, as these (but not the squares of the amplitudes) also serve as a direct measure of the range of the transmitter in the various directions (see c). c. The characteristic of a transmitter generally depends upon the distancef at which the amplitudes are measured. Strictly speaking, therefore, we can only refer to the characteristic of a transmitter at a given distance. However, as the distance becomes very great in comparison with the wave-length employed, then, in general, further increases in the distance will have little or no effect upon the shape of the characteristic. Hence, we are justified in speaking of "the" long distance characteristic of a transmitter. This same characteristic can also be obtained by plotting the ranges of the transmitter (for a given receiver) over a highly con- ductive ground J as vectors in the various directions. But for distances which are not large compared to the wave-length, or * Thus, if the ratio of the lengths of two vectors in the amplitude characteristic is 1 : 2, then it will be 1 : 4 for the corresponding vectors in the characteristic of the squares. t In this and what follows the effect of the distribution of land and water and other local influences upon the characteristic is not taken into account; the ground is as- sumed to be homogeneous in all directions. t Otherwise absorption would complicate the conditions. 340 WIRELESS TELEGRAPHY perhaps even shorter, the shape of the characteristic is largely dependent upon the distance. Consequently a characteristic determined by measurements relatively close to the transmitter gives no definite indica- tion of the long distance characteristic and can not serve as a measure of the practical usefulness of the transmitter. A transmitter whose character- istic at a short distance appears very advantageous for directive signal- ing, may nevertheless have a long distance characteristic which is almost a circle. 1. THE FIRST ATTEMPTS 193. Use of Reflectors. To attain the almost ideal case represented by Fig. 402, an adaptation of HERTZ'S parabolic mirror method as employed in his well-known experiments, readily suggests itself. In fact it has often been proposed to use such reflectors of sheet metal or wires to send the waves out in a single direction. MARCONI also conducted some early experiments with reflectors. This was reasonable enough as long as it was customary to work with very short waves. In modern practice, however, the wave-lengths em- ployed range from 300 to 6000 m. or more. A reflector, to have the desired result, as obtained in optics or in HERTZ'S experiments, would have to have dimensions commensurate with the wave-length. This requirement is sufficient to eliminate the practical use of reflectors for the wave-lengths in question. 194. Attempts at Screening, J. ZENNECK. A characteristic of the kind shown in Fig. 401, i.e., with very little radiation in one direction (SB in Fig. 401), was obtained by the author as early as 1900 in the following manner. At a station, A (Kugelbake, near Cuxhaven), two vertical wires, did 2 (Fig. 403), about 30 m. long were suspended about 6 m. apart. A X* X / FIG. 403. The receiving station, B (Altenbruch Lighthouse) was situated about 9 km. from A and nearly, though not quite, in line with did^. With only one aerial wire in use the messages sent out could be well under- stood at the receiving station, but at twice this distance reception was no longer possible, so that we were working with a safety factor of a bit less than 2. The following tests were then made: 1. di used as transmitting aerial, d 2 not grounded; the signals were clearly audible at B; 2. di again transmitting, d 2 grounded; no reception at B; DIRECTIVE TELEGRAPHY 341 3. c? 2 transmitting, di grounded; signals clearly received at B. From 1 and 2 it was concluded that it is possible to greatly reduce the range in a given direction by means of a grounded wire parallel to the transmitting aerial; from 3, that this does not materially affect the range in the opposite direction. The results of the experiment left no doubt that, e.g., a station A (Fig. 404) can send telegrams to another station B, while a third station FIG. 404. C, at the same distance as B, does not receive these messages, if wire d\ at A is used as the transmitter, d 3 is grounded and d 2 is insulated from the ground. Insulating d 3 and grounding d 2 reverses the conditions so that C, and not B, receives. These tests were taken up by the TELEFUNKEN Co. at a later date and the results verified. It was here shown that an essential factor consists in having the screening aerial tuned to the transmitter frequency, which condition was really fulfilled in the tests when the screening aerial was grounded. These tests were not carried out far enough to form definite conclusions of just what can be accomplished in this direction.* 2. METHODS EMPLOYING SEVERAL ANTENNAE 195. The Field of Several Antennae. General Consideration. If two vertical antennae, oscillating at the same wave-length, are a given distance apart, then the amplitude of the resultant wave produced by the two antennae is never uniform in all directions, whether or not the currents in the two antennae are in phase. At any distant point P (Fig. 405), the two waves which are there superimposed have traveled different distances and in view of this difference (AD, Fig. 405) they are not in phase with each other [Art. 206]. This phase difference

f r , between the two antenna currents is so chosen that their resultant effect is zero either in the direction OA or in the direction OB (Fig. 405) in the plane of the antennae (3f = TT -r j , the characteristics obtained have a distinct difference from those obtained in the other two cases. The latter always consisted of two symmetrical halves, so that * The characteristic is determined by the equation E ro = 2E sin = 2E sin cos f This is evident from the fact that for this case we can write : E ro = 2E . ^ ird rox. = 2E . -r~ cos i A J This amplitude is approx. = 2E . cos tf approx A 2E X 1 for d = % X 2E X 0.71 for d = Y \ 2E Q X 0.31 for d = Y^ \ DIRECTIVE TELEGRAPHY 345 the same range is obtained in any two directions 180 apart; the arrange- ment is said to be "bilateral." Transmitters of the third class however are "unilateral," i.e., the ranges in any two directions 180 apart are different. In Figs. 413, 414 and 415 the distance effect characteristics* are given for the following cases: d < >{ 2 * (Fig. 413); d = Y\ (Fig. 414); d = (dotted curve, Fig. 415); d = %\ (full line curve, Fig. 415). *** FIG. 413. FIG. 414. FIG. 415. Just as in case b the characteristic becomes unfavorable for directive signaling in this case also, as soon as the distance between the antenna is made more than one-fourth of the wave-length, whereas very small distances between the antennae are very advantageous. But, again as in case 6, the maximum range is reduced at the same time as the ratio - is decreased. A 197. Double Antennae, One-half Wave-length Apart (S. G. BROWN, A. BLONDEL, J. STONE STONE 308 ). a. The case discussed in Art. 1966, which is particularly advantageous both with respect to directive power and range two similar antennae placed a half wave-length apart, with their currents of equal amplitude but opposite phase has been fre- quently proposed since 1899, by various experimenters. To produce the oscillations the antennae are joined at their bases, A and B, by a conduct- ing circuit which is suitably coupled [Art. 198a] to a condenser circuit. This arrangement is not entirely identical with the case discussed in Art. 1966, for to the effect of the vertical antennae AC and BD is added that of the horizontal portion AB, which under certain conditions [Arts. 2036 and 206] may be quite considerable. It is almost self-evident that a pair of antennae of the kind just de- scribed will serve for directive reception,^ i.e., will respond with varying * If the phase difference is so chosen that the fields of the two antennae neutralize each other in the direction OA or OB (Fig. 405), then we have E ro = 2E sinT 7 ^ (cos & - 1)1 or E ro = 2E sm\~ (cos & + 1)1 LA J LA J For small values of - these equations may be simplified into E ro = 2E . (cos A A *T1). f For this purpose a detector circuit is coupled to the antenna pair at the anti-node of current. 346 WIRELESS TELEGRAPHY intensity as the direction of the approaching waves varies. Thus waves whose direction is perpendicular to the plane of the antennae, induce potentials of equal phase and amplitude in both antennae, so that these would neutralize each other and produce a zero effect in the system shown in Fig. 416. But waves approaching in the plane of the antennae if their wave-length is 2AB (Fig. 416) induce potentials of opposite phase FIG. 416. in the two antennae, so that their effect upon the oscillatory system is additive.* b. A somewhat different form of the double antenna (Fig. 417) has been proposed by A. BLONDEL. 308 When used for transmission, the same condenser circuit is coupled with the coils S and S' in such manner that the currents in the two antennae will be of opposite direction. Then, so B B 1 C f FIG. 417. far as distance effect is concerned, the currents in the vertical parts AB and A'B' entirely neutralize each other, and all that remains is the effect of the currents (of opposite direction) in parts CD and C'D' (whose dis- * The distance effect characteristic, at least for the vertical portions of this double receiving antenna, would be the same as for the antenna pair when used for transmit- ting, as will be readily understood by reversing the conditions in the discussion of Art. 196. DIRECTIVE TELEGRAPHY 347 tance apart is made about equal to a half wave-length) and of the cur- rents in the horizontal portions BC and B'C 1 '. 198. The Methods of E.Bellini and A. Tosi. 309 a. BELLINI and Tosi have also adapted the case described in Art. 1966, using two antennae with currents of equal amplitude but opposite phase, but the distance between their antennae is sometimes only slightly, sometimes much less than half the wave-length. Instead of being vertical, however, the antennae are slightly inclined (Fig. 418). This arrangement has the advantage of being more easily suspended from a single mast. When located over ground of very high conductivity (sea water) the action of such a pair of inclined antennae is not much different from that of a vertical pair of the same height, but somewhat closer together (as represented by the dash-and-dotted lines in Fig. 418). But over ground of relatively low conductivity, the distance effect characteristic is apt to be considerably different from that obtained with two vertical antennae [see Art. 205]. In this last case the horizontal portion AB (Fig. 418) is likely to have a material effect. In order to obtain oscillations of opposite phase in the two inclined antennae, BELLINI and Tosi make use of the second upper harmonic* \ * Corresponding to the second upper harmonic shown in Fig. 34. The fundamental oscillation, in which A' ABB' is equivalent to one-half the wave-length, can also be used for this purpose. 348 WIRELESS TELEGRAPHY third harmonic of the entire system [Art. 22] (Fig. 418). For any given distance between the inclined antennae, the upper harmonic is obtained, say, by inserting self-induction (or perhaps condensers also) of suitable dimensions. The oscillations are then induced by means of a condenser circuit, CS, Fig. 418, tuned to the frequency of the desired harmonic. When this arrangement is to be used for reception, the condenser cir- cuit is replaced by a detector circuit. The system then reacts with the greatest intensity upon waves whose direction lies in the plane of the antennae. b. With the arrangement of Fig. 418, the direction of maximum wave amplitude lies in the plane of the antennae. If this direction is to be FIG. 419. varied at will it is necessary to turn the entire system. This would be impracticable on shipboard and particularly on fixed land stations. In view of this, BELLINI and Tosi have introduced another method for obtaining the desired result.* They combine two pairs of antennae (AB and AiBi, Fig. 419), each being of the form illustrated in Fig. 418, so that their planes are at right angles to each other. Similarly the coupling coils /S'and S" (Fig. 419) are arranged so as to be perpendicular to each other. The coil S, which is part of the condenser circuit CS, used for excitation, can be rotated within the coils S f and S". If, then, the distance d between the antennae is small compared to the wave-length (d ^ gj [Art. 1966] a very simple calculation 309 will bring out the following facts: 1. The direction of maximum range lies in the same plane as coil *A. BLONDEL 310 has also proposed other methods for securing the same results. DIRECTIVE TELEGRAPHY 349 S;* the amplitude of the waves in this direction is always equal to the maximum amplitude of a single pair of the antennae, independently of the position of S. 2. The distance effect diagram has the same form, that of Fig. 410, for all positions of S } and consists of two tangent circles, whose line of centers lies in the plane of coil S. But if the distance between the antennae is greater, d being from - to ~, then condition 1 is retained, i.e., the direction of maximum range u ^ lies in the plane of S* and can be varied at will by rotating S, the 0*270 FIG. 420. maximum amplitude does not remain constant for all positions of S; in fact it has its greatest value for /? = 45 and /? = 135 and its minimum for fi = and /3 = 90 f [see Fig. 420, which gives the maximum ampli- tudes for all the different positions of S, i.e., different values of (Fig. 419)]. The distance effect characteristic is also changed somewhat, in * When these antennae are mounted on shipboard, the metallic masses in the ship and particularly the rigging are apt to affect the distribution, so that the direction of the maximum wave amplitude no longer coincides with that of coil S. Then empirical calibration of the radio-goniometer is necessary (see what follows). f The maximum amplitudes for these two cases differ by 8 per cent, when d = ^ and by 24 per cent, when d = ~ 350 WIRELESS TELEGRAPHY this case, as the position of S is varied. This, however, is not of great practical importance.* BELLINI and Tosi have combined the two coupling coils, S' and S" together with the movable coil S, in a single apparatus called the trans- mitting "radio-goniometer" (Fig. 421). The two coils S' and S" (Fig. 419) are wound on a cylinder inside of which S rotates. FIG. 421. For the reception of waves tuned to the goniometer, the so-called "receiving radio-goniometer" is used, which is the same in principle as the transmitting goniometer, but whose coils are wound with a different number of turns. The movable coil is joined to a detector circuit. A simple consideration of the action of the transmitting goniometer with the conditions reversed for reception makes it evident that the receiving goniometer will respond with the greatest intensity to waves approaching * If the antennae and the exciting condenser circuit are closely coupled, two waves will in general be transmitted; their frequency, however, is not changed as the position of S is varied. 309 DIRECTIVE TELEGRAPHY 351 in the direction of the plane of the movable coil and that it will fail to respond when this direction is perpendicular to the approaching waves. The methods of BELLINI and Tosi have been put to extensive practical tests in France and seem to have given very satisfactory results. A large station has been erected on this principle at Boulogne. The aerials A' F IG . 423. are supported by means of 4 steel towers, are 36 m. high, 80 m. apart at the top and 127 m. apart at their bases. The horizontal portions (AB, Fig. 418) are 8 m. above the ground and the wave-length is 300 m. The Boulogne Station has communicated at night, using only 0.5 kw. energy with Algiers (1500 km.) [See Art. 145/ in this connection.] c. The distance effect characteristic of the double antennae discussed in a and b has the disadvantage of being bilateral, i.e., the effect in any two directions 180 apart is alike. A unilateral characteristic is secured by placing a simple vertical antenna in the center of the pair of antennae shown in Fig. 418, thereby obtaining the arrangement illus- trated in Fig. 422. If the current in this middle antenna is in phase with that in an- tenna BB', the effect in the direction OB is strengthened, while that in direction OA is weakened; under suitable conditions, there- fore, a distance effect characteristic of the form of Fig. 423 is obtained, i.e., the ampli- tude has a decided maximum in direction OB and a decided minimum in direction OA. If it is desired to make the direction of maximum amplitude of this arrangement variable at will, the principle discussed in b can be directly ^ 352 WIRELESS TELEGRAPHY FIG. 425. applied for this purpose; to the radio-goniometer with its two pairs of antennae and their fixed coupling coils S' and S" (Fig. 419) there is added a simple vertical antenna (OD, Fig. 424) whose coupling coil is mechanic- ally joined to the excitation coil S (Fig. 419) and, therefore, turns with S. This offers a simple means of varying the direction of maximum radia- ^ tion at will, the distance effect characteristic being of the form shown in Fig. 423. If this arrangement is used 2 without any modification as a receiver it will not have the same distance effect character- istic as it has when transmit- ting, as the potential induced in the central vertical antenna would not be in phase with one of the inclined antennae. This must therefore be taken into consideration. 199. The Methods of F. Braun. 311 One of the methods with which F. BRAUN experimented in 1906 is illustrated in Fig. 425. The oscilla- tions in antennae $2 and 83 are in phase with each other, while those in antenna Si are displaced 270 from the others. The amplitudes in the three antennae are proportioned as follows: AI :Az :A S = 1 :0.5 :0.5; the distance, A, between them is j. Calculating the values for the char- acteristic in this case (on the assumption of ground of very high con- ductivity), the curve b of Fig. 426 is obtained, i.e., there is maximum radiation in the direction OA and zero radiation in the opposite direc- tion OB. This was borne out in the tests made by the very strong effect obtained in direction OA. In the opposite direction, how- ever, the effect, though very slight, did not entirely disappear.* Theoretically, even more ad- vantageous characteristics for directive signaling are obtained by means of four antennae suitably arranged (curve c, Fig. 426). 200. Production of any Desired Phase Difference with Undamped Oscillations (G. E. PETIT 312 ). In the methods of BRAUN, as well as * In one test, e.g., the deflection of the measuring instrument used in the receiving set was 30 scale divisions in direction OA and only 2 scale divisions in the opposite direction. FIG. 426. DIRECTIVE TELEGRAPHY 353 nSTOra in those discussed in Art. 196c, the chief difficulty consists in exciting oscillations of a certain desired phase difference in the transmitters.* This problem can be solved very easily, at least in principle, in the case of undamped oscillations. An arrangement suitable for this purpose is sketched in Fig. 427. The primary condenser circuit CiSiS'i, in which undamped oscillations are induced by means of a high frequency generator or the arc method, acts inductively (coupling coils S f i and S'z) upon a second condenser circuit CzSzS'z, which is in resonance with Ci/Si/S'i. Consequently undamped oscillations are induced in the secondary circuit CzSzS'z, but these are 90 out of phase with those in circuit CiSiS'i* The planes of the two coils Si and $ 2 are at right angles to each other. As the currents flowing through Si and 82 are 90 out of phase, a rotating magnetic field is produced in the space surrounding these coils; this field is circular in form if the dimensions and coupling of the two condenser circuits are so chosen that the magnetic fields of each of the coils Si and 82 are equal in amplitude. If, now, two other coils, $ 3 and $ 4 , having an angle < between their planes, are inserted in this rotating field, electro- motive forces, having a phase differ- ence <, will be induced in them. Hence, if $ 3 and $ 4 are each connected to one of two similar antennae, the currents in the latter will also have a phase difference $. The amplitudes of the two antenna currents thus obtained can also be given any desired ratio by choosing the number of turns of the two coils Si and S 2 accordingly. 201. Production of any Desired Phase Difference with Damped Oscillations. This far more difficult problem has been solved by L. MANDELSTAM and N. PAPALEXi, 313 whose method will be understood from the following consideration. a. Let the condenser circuit FC'AC L BC"F (Fig. 428) be caused to oscillate. Let V represent the voltage between points B and A, Vi the voltage across the terminals of condenser Ci, Si the e.m.f. induced along AL'iCiU'iB. Then, if the ohmic resistance is very low, V = Vi + 8 t - approximately. Vi leads the current which is marked i in Fig. 429 and the following * The method customary in alternating current practice (light and power) viz., branching off between inductive and non-inductive resistance is not applicable in this case, as the non-inductive resistances would have to be so great as to increase the damping far beyond permissible limits. 23 rooo 01 Si FIG. 427. 354 WIRELESS TELEGRAPHY figures by 90 and 8; lags behind the current by 90. Their curves are, therefore, about as shown in Fig. 429. b. Now let the points A and B be connected through a coil of very great self-induction. The rapid oscillations of the condenser circuit then continue just as if this coil were not there [Art. 416]. But during the time in which the condenser circuit is being charged by the induction coil (or transformer) the coil between A and B acts as a short-circuit across con- To Induction Coil , C ' L' FIG. 428. denser, C. Hence, the potential Vi must have an initial value of zero, and cannot start at its maximum as shown in Fig. 429. Moreover, a constant potential whose amplitude is equal to the maximum amplitude of the variable or alternating potential V\ of Fig. 429 is added to the latter, so that curves V and V\ are raised, appearing as in Fig. 430 if Vi >& io) and otherwise as in Fig. 431. In the first case, Vi >&i , which is equivalent to stating that ^r>uL i} i.e., when the FIG. 429. FIG. 430. condensance of circuit ACiB is greater than its inductance, it is essential for what follows that the maximum of potential V occurs after half a period of the condenser circuit FC'ACiBC"F. In the second case, which is of no interest in regard to what follows, the maximum of potential V occurs immediately after the beginning of the oscillations. c. Let another condenser circuit, 77, be added to the arrangement of DIRECTIVE TELEGRAPHY 355 Fig. 428, as shown in Fig. 432. Spark gap FI is so adjusted in length that sparks are just able to jump across it whenever a spark passes across F. Condenser circuits /* and II are tuned to be in resonance with each other. Then if a' spark jumps across F, condenser circuit II and condenser circuit FC'ACiBC"F will oscillate simultaneously. But the spark at FI and, therefore, the natural oscilla- tions of condenser circuit I do not begin until the potential Vi at FI has reached its maximum, i.e., until half a period of condenser circuit FC'ACiBC"F has elapsed. As the natural period of this condenser circuit can be adjusted within cer- tain limits by varying the coils L'L", we have in these a means of controlling the time (within those limits) which will elapse before the oscillations of condenser circuit / commence after those of circuit II have started, i.e., the means of giving the oscillations of circuit I any desired phase displacement (within certain limits) from those of circuit II. d. For carrying this method out in practice, the following points should be noted: 1. Above all the condition that n >o>Li must be secured. For this FIG. 431. is equivalent to making the frequency of condenser circuit FC'AC\BC"F less than that of condenser circuit / [Art. 5a\. C' To Induction Coil [I L Ii To Induction Coil n" FlG. 432. 13 2. It is advantageous to have the resultant capacity of condensers C' and C" equal to that of Ci and of 2, as this makes the efficiency of the entire system a maximum. 3. The three parts into which the system divides itself must have * That is 356 WIRELESS TELEGRAPHY no appreciable inductive effect upon one another. Otherwise the various reactions which occur would be far more complicated than as stated above. 4. To insure prompt sparking at FI as soon as the potential there is at its maximum, it is advisable to let the ultra-violet rays from spark gap F fall upon gap FI or use some other means of ionizing gap FI [Art. 426]. 3. AERIALS HAVING HORIZONTAL OR INCLINED PORTIONS 202. Marconi's Bent Antenna. MARCONI 314 approached the prob- lem of directive signaling in a way quite different from any of the methods described in 2. His method is to use an aerial consisting of a short vertical and a long jp G horizontal portion, which in its simplest form appears as shown 433.* The mere fact that MARCONI has shown that, at a distance of about one wave-length, this transmitter has a characteristic of the form of Fig. 434 f proves nothing (according to Art. 92c) in regard to the effect at great distances. However, MARCONI has demonstrated by means of long distance tests, that this form of transmitting aerial has a much greater effect in direction AC than in the opposite direction and has a particularly small effect in the direction perpendicular to the plane of the aerial. Hence the characteristic at great distances must also have a greater length (vector) in the direction AC than in the opposite direction. Fig. 43 5 J is a sketch of the actual construction of an antenna of the type of Fig. 433, as used by MARCONI for his transatlantic stations. The fact .350136 180 170' FIG. 434. The direction marked 360 corre- * When MARCONI'S experiments were made, it was found that the best results were obtained when the horizontal portion of the aerial was one-fifth of the wave-length. Fig. 434 is the characteristic under this con- dition. f From Proc. Royal Soc., A77, p. 415, 1906. sponds to the direction AC in Fig. 433. $ From the Jahrbuch fur drahtl. Tel., 1, 608, 1908. The Clifden station is reported 315 as having 30 masts each 60 m. high, between which 200 parallel wires are stretched over a length of 2000 m. and a width of 330 m. The fundamental wave-length of this antenna is said to be 4000 m. Later reports state that MARCONI now employs separate transmitting and receiving antennae in his transatlantic stations. The transmitting aerial is said to be 600 m. long, the re- DIRECTIVE TELEGRAPHY 357 that MARCONI has adopted this form for his transatlantic stations is perhaps the best evidence of its merits. 203. The Action of the Bent Marconi Antenna when Transmitting. a. The action of the MARCONI antenna can not be explained as long as we retain the assumption of perfect conductivity for the earth. For under this assumption we would be justified in replacing the transmitter of Fig. 433 and the effect of the earth by the double trans- FIG. 435. mitter of Fig. 436 without any ground [Art. 138a] and in calculating the field of this transmitter from the effect of the individual current elements of the antenna [Art. 256], With a flat earth's surface the field in the equatorial plane is the important factor. But in the equatorial plane the fields due to the horizontal portions of the antenna (Fig. 436) tend to neutralize each other as the distance from the transmitter increases. At very great distances, which of course are always in question in wire- FIG. 436. less telegraphy, practically nothing remains except the effect of the vertical portion of the antenna, and this is the same in all directions in view of the symmetry of the vertical portion. Under these conditions, therefore, this transmitting antenna could not be used for directive signaling. From this it follows, on one hand, that the bent MARCONI antenna can have little or no directive power when located over sea water, i.e., on shipboard,* and would radiate uniformly in all directions. On the other hand, the directive power which this antenna actually has when used on land, can be explained only by taking the action in ceiving aerial 1800 m. long and only 2-4 wires are used. [Translator's Note. The MARCONI Co. has adopted separate transmitting and receiving stations for all its new transatlantic stations, as e.g., New Brunswick and Belmar.] * Or rather, to be more exact, on a wooden raft; for the metal rigging of a modern ship affects the radiation and destroys its uniformity. 358 WIRELESS TELEGRAPHY the ground and the latter's conductivity and dielectric constants into consideration. b. The first real explanation of the bent MARCONI antenna was given comparatively recently by H. VON HoERSCHELMANN, 316 a pupil of A. SOMMERFELD. His theory, based on the assumption of homogeneous ground in the vicinity of the transmitter in both horizontal and vertical directions may be developed as follows : The action of a horizontal antenna stretched out over ground of mod- erate conductivity, consists in its inducing powerful earth currents in its immediate vicinity in the upper strata of the earth. The amplitude of the vertical components of these currents has a sharply defined maxi- mum at a certain distance to either side of the middle of the antenna (in the plane of the antenna) and the phases of the vertical component currents to the right and to the left of the middle point are opposite. In accordance with the theory, we may now consider all the vertical components of the earth currents as being concentrated at the two maximum points mentioned above and the entire action then pro- ceeds as if two simple wave series were being radiated from two vertical antennae erected at the two points of maximum and whose currents J"' 1 I r u ! 1 I II !, \o U A FIG. 437. were opposite in phase. This imaginary vertical double antenna in short is, so to say, automatically produced in the ground by the horizontal trans- mitting antenna. The field of the bent MARCONI antenna as can be shown from the theory, is easily calculated by superimposing the field of the vertical portion A B (Fig. 437) upon that of the two imaginary antennae XX' and YY' pro- duced by the horizontal portion BC, both being calculated according to the rules of Art. 25, just as if the conducting earth were not present. This system of antennae therefore resembles the arrangement dis- cussed in Art. 198c, the combination of a simple vertical antenna with a pair of antennoB oscillating in opposite phases. But in the case before us the distance d = XY between the pair of antennae, is not optional, being in fact equal to the height h ( = AB Fig. 437) of the MARCONI antenna. Moreover, the phase of the oscillations in the double antenna is not the same as (nor opposite to) that of the oscillations in antenna A B, but the oscillations in XX' lag 45 behind those in AB. Finally, the amplitudes of the waves radiated by each of the imaginary pair DIRECTIVE TELEGRAPHY 359 of antennae XX' and YY' though equal to each other, are not equal to the amplitude of the wave radiated by antenna AB. Denoting the former amplitude by E fo and the latter by E ho , their relation is given by* (1) The two imaginary antennae according to Art. 1966 produce a field whose amplitude at a very distant point P* is E' Q = 2E fo ~ cos X = 2E fo ~ cos tff (2) A A If we superimpose the wave radiated by this imaginary double antenna upon that radiated by the vertical antenna, keeping the 45 difference in phase in mind we obtain the amplitude of the resultant wave, (3) = ti ho \1 +0*008* #+ V^.jS cos #J 7 1 in which 2crXc This relation determines the distance effect characteristic of the bent transmitting antenna. Its form depends upon the value of /3, i.e., aside from the wave-length, it depends mainly upon the ratio of the length of the horizontal portion of the antenna to the vertical portion and upon the conductivity of the ground. In Fig. 438 the distance effect characteristics are shown for @ = 4{ (heavy full line curve 6) and for /? = 1.4 (lighter curve c); they correspond to ground of poor conductivity. The former, 6, is very similar to that observed experi- mentally by MARCONI (Fig. 434) ; the theory therefore gives results which agree well with the actual facts. The maximum directive power is obtained when /3 = 1 (characteristic very similar to curve c) ; with $ = 0.2 the characteristic (dot-and-dash curve d, Fig. 438) has already lost its directive form to a very large extent. If the conductivity of the ground is very great, making (3 very small, then, in equation (3) the first term under the radical sign becomes * Under the following assumptions : 1. Height, h, and length, I, of the antenna <^X. 2. The expression j ^1.0 [where a- = specific conductivity of the ground, VL = velocity of light and k = dielectric constant of the ground, all in c.g.s. units]. This assumption is always correct for the conditions encountered in practice. t $POA = & [see Art. 196]. J Corresponding, e.g., to : a = 1.2 X 10~ 16 c.g.s. units; X = 2000 m.; l/h = 5. Corresponding, e.g., to : er = 10~ 15 c.g.s. unitsj X = 2000 m. ; l/h = 5. 360 WIRELESS TELEGRAPHY the determining factor and, as was to be expected from a, E rc Z 120/ 4194304 In these equations the correction factor AL S = 4arrn (C + D) (E. B. RosA 353 ). The values of C and D are given in the table following below. 5 Flat spiral, in which the product ng < 0.5r (r is here the radius of the middle turn, i.e., the mean radius) (A. EsAU 353 ). L s = lirr \ n flog.- + 0.333) + n(n - 1) flog. - 2 } - A + |^ I \ P I \ 9 I 8r 2 [ ( loge 7 + 3 ) ("' (g l8~ X) ) - 1] 1 c g ' s - units 6. Rectangle whose sides are a and 6, of wire whose radius is p : 2ab 2ab L s = 4 {a log e - --=== + 6 l oge r(a + Va 2 + 6 2 ) r(6 + Va 2 + 6 2 ) + 2(\/aM :r & 2 - a - 6)} c.g.s. units * L = effective coefficient of self-induction, calculated under the assumption that the current flows only through a very thin surface sheath or "skin." 394 WIRELESS TELEGRAPHY Table for A and B n A B n A B 1 16 354 4 35,694 2 17 415.8 46,757 3 1.386 8.315 18 482.8 60,427 4 4.970 43.296 19 555.5 76,662 5 11.33 140.82 20 634.2 96,910 6 20.90 366.95 21 718.9 119,330 7 34.06 794.73 22 809.7 146,517 8 51.11 1,499.55 23 906.6 178,140 9 72.32 2,590 . 62 24 1,009.8 217,338 10 97.92 4,187.55 25 1,119.4 259,868 11 128.17 6,572.94 26 1,235.4 305,044 12 163.14 9,769.47 27 1,357.9 359,767 13 202.1 14,042.1 28 1,487.1 421,783 14 248.2 19,532.2 29 1,618.1 491,819 15 298.6 26,740.1 30 1,765.4 570,515 Table for C 2p 9 C 2p 9 C 2p 9 C 1.00 0.5568 0.79 0.3211 0.59 . 0292 0.99 0.5468 0.78 0.3084 0.58 0.0121 0.98 0.5367 0.77 0.2955 0.57 -0.0053 0.97 0.5264 0.76 0.2824 0.56 -0.0230 0.96 0.5160 0.75 0.2691 0.55 -0.0410 0.95 0.5055 0.74 0.2557 0.54 -0.0594 0.94 0.4949 0.73 0.2421 0.53 -0.0781 0.93 0.4842 0.72 0.2283 0.52 -0.0971 0.92 0.4734 0.71 0.2143 0.51 -0.1165 0.91 0.4625 0.70 0.2001 0.50 -0.1363 0.90 0.4515 0.69 0.1857 0.45 -0.2416 0.89 0.4403 0.68 0.1711 0.40 -0.3594 0.88 0.4290 0.67 0.1563 0.35 -0.4928 0.87 0.4176 0.66 0.1413 0.30 -0.6471 0.86 0.4060 0.65 0.1261 0.85 0.3943 0.25 -0.8294 0.64 0.1106 0.20 -1.0526 0.84 0.3825 0.63 0.0949 0.15 -1.3403 0.83 0.3705 0.62 0.0789 0.10 -1.7457 0.82 0.3584 0.61 0.0626 0.81 0.3461 0.60 0.0460 0.80 0.3337 TABLES Table for D 395 n D n D n D 1 . 0000 35 0.3119 300 0.3343 2 0.1137 40 0.3148 400 0.3351 3 0.1663 45 0.3169 500 0.3356 4 0.1973 50 0.3186 600 0.3359 5 0.2180 60 0.3216 700 0.3361 6 0.2329 70 0.3239 800 0.3363 7 0.2443 80 0.3257 900 0.3364 8 0.2532 90 0.3270 1000 0.3365 9 0.2604 100 0.3280 10 0.2664 125 0.3298 15 0.2857 150 0.3311 20 0.2974 175 0.3321 25 0.3042 200 0.3328 30 0.3083 396 WIRELESS TELEGRAPHY Table VII. Effective The figures give the resistance of 1 m. in ohms, under the assumption are correct within Diam. of wire in mm. "Station- ary" current N = 5 X 10< eye. /sec. X = 6000 m. N = 1 X 105 cyc./sec. X = 3000 m. ./V = 1.5X105 cyc./sec. X = 2000 m. N = 2XW 5 cyc./sec. X = 1500 m. ./V = 2.5X105 cyc./sec. X = 1200 m. N = 3X105 cyc./sec. X = 1000 m. 0.2 0.554 0.55 0.56 0.56 0.56 0.56 0.56 0.4 0.138 0.139 0.141 0.143 0.148 0.152 0.157 0.6 0.0615 0.063 0.067 0.072 0.078 0.086 0.093 0.8 0.0346 0.0370 . 0422 0.0498 0.056 0.062 0.067 1 0.0221 0.0254 0.0323 0.0382 0.0434 0.0480 0.052 1.2 0.0154 0.0196 0.0262 0.0314 0.0354 0.0393 . 0427 1.4 0.0113 0.0164 0.0221 0.0263 0.0298 0.0331 0.0359 1.6 0.00865 0.0140 0.0189 0.0226 0.0258 0.0285 0.0311 1.8 0.00683 0.0123 0.0169 0.0199 0.0226 0.0251 0.0273 2 0.00554 0.0110 0.0148 0.0178 0.0202 0.0225 0.0245 2.2 0.00457 0.0098 0.0133 0.0159 0.0182 0.0203 0.0221 2.4 0.00384 0.0089 0.0121 0.0146 0.0166 0.0185 0.0202 2.6 0.00328 0.0081 0.0111 0.0134 0.0153 0.0171 0.0186 2.8 0.00282 . 0075 0.0102 0.0123 0.0141 0.0158 0.0172 3 0.00246 0.0069 0.0095 0.0115 0.0132 0.0147 0.0160 3.2 0.00216 0.0065 0.0089 0.0107 0.0123 0.0137 0.0149 3.4 0.00192 0.0061 0.0083 0.0101 0.0116 0.0129 0.0141 3.6 0.00171 0.0057 0.0079 0.0096 0.0110 0.0122 0.0133 3.8 0.00153 0.0053 5 0.0074 0.0090 0.0103 0.0114 0.0125 4 0.00138 0.0051 0.0070 0.0085 0.0097 0.0108 0.0118 4.2 0.00125 0.00479 0.0066 0.0080 0.0092 0.0103 0.0112 4.4 0.00114 0.00456 0.0063 0.0077 0.0088 . 0098 0.0107 4.6 0.00105 0.00438 0.0061 0.0074 0.0085 0.0094 0.0103 4.8 0.000961 0.00417 0.0058 0.0070 0.0081 0.0090 0.0096 5 0.000886 0.00400 0.0055 5 0.0067 0.0077 0.0086 0.0094 5.2 0.000819 0.00383 0.0053 0.0065 0.0074 0.0083 0.0090s 5.4 0.000759 0.00368 0.0051 0.0062 0.0071s . 0080 0.0086 5.6 0.000706 0.00354 0.00493 0.0060 0.0069 . 0076 0.0083 5.8 0.000658 . 00341 0.0047s 0.0058 0.00665 0.0074 0.0081 6 0.000615 0.00330 0.00458 0.0056 0.0064 0.0071 0.0078 6.2 0.000576 0.00319 0.00443 0.0054 0.0062 0.0069 0.0075s 6.4 . 000541 0.00309 . 00429 0.0052 0.0060 0.0067 0.0073 6.6 0.000508 0.00299 0.00415 0.0050s 0.0058 0.0064s 0.0071 6.8 . 000479 0.00290 0.00403 0.00489 0.0056 . 0063 0.0068s 7 . 000452 0.00281 0.00391 0.00475 0.0055 . 0061 0.0067 7.2 0.000427 0.00272 0.00379 0.00461 0.0053 0.0059 0.0064s 7.4 0.000404 0.00265 0.00369 0.00448 0.0051 . 0058 . 0063 7.6 . 000383 0.00257 0.00359 0.00433 0.0050 0.0056 . 0061 7.8 0.000364 0.00251 0.00350 . 00426 . 00488 0.0055 0.0059 8 0.000346 0.00244 0.00341 0.00415 0.00477 0.0053 0.0058 TABLES 397 Resistance of Copper Wires 354 that the specific conductivity a 1 to 2 per cent. 57.5 X 1C- 5 c.g.s. units. The figures cyc./sec. X = 857m. N = 4X105 cyc./sec. X = 750 m. -$" = 4.5X105 cyc./sec. X = 667 m. cyc./sec. X = 600 m. AT = 10 _ 1 234 567 Maximum Gap Length in cm. FIG. 469. In these figures (468 and 469), r is the radius of the spherical electrodes; the dotted curve in Fig. 469 refers to very shallow bowl-shaped electrodes. The values plotted are the normal' discharge or ignition voltages, i.e., the voltages which are just sufficient for the discharge to take place in air having no appreciable ionization. The values of Fig. 468 are due to A. HEYDWEiLLER, 64 those of Fig. 469 to C. MtiL- LER 56 (for the short gap lengths) and E. HuPKA 64 and those for the dotted curve in Fig. 469 are due to W. WEiCKEB; 64 barometric pressure 745 mm., temperature about 18 C. The figures for Fig. 468 were determined in dry air at 18 C. temperature and 745 mm. pressure; an increase of 8 mm. pressure and a decrease of 3 temperature cause an increase of 1 per cent, in the voltages. TABLES 401 Table X. Determination of Percentage Coupling According to Art. 87, the degree of coupling is N 1 In the following table, the degree of coupling is given in percentage; thus for K f = 0.02 the figure given is 2 (per cent.). I II III ^0^ x N i Percentage coupling X 11 N *-V' Percentage coupling ALor^ 7 X 11 N 1 Percentage coupling 0.999 0.20 1.001 0.20 1.001 0.100 0.998 0.40 1,002 0.40 1.002 0.200 0.997 0.60 1.003 0.60 1.003 0.299 0.996 0.80 1.004 0.80 1.004 0.398 0.995 1.00 1.005 1.00 1.005 0.498 0.994 1.20 1.006 1.20 1.006 0.596 0.993 1.40 1.007 1.40 1.007 0.695 0.992 1.59 1.008 1.61 1.008 0.799 0.991 1.79 1.009 1.81 1.009 0.897 0.99 1.99 1.01 2.01 1.01 0.99 0.98 3.96 1.02 2.04 1.02 1.98 0.97 4.91 1.03 6.09 1.03 2.97 0.96 7.84 1.04 8.16 1.04 3.92 0.95 9.75 1.05 10.2 1.05 4.87 0.94 11.6 1.06 12.4 1.06 5.82 0.93 13.5 1.07 14.5 1.07 6.76 0.92 15.4 1.08 16.6 1.08 7.68 0.91 17.2 1.09 18.8 1.09 8.60 0.90 19.0 1.10 21.0 1.10 9.50 0.89 20.8 1.11 23.2 1.11 10.4 0.88 22.6 1.12 25.4 1.12 11.3 0.87 24.3 1.13 27.7 1.13 12.2 0.86 26.0 1.14 30.0 1.14 13.0 0.85 27.8 1.15 32.2 1.15 13:9 2G 402 WIRELESS TELEGRAPHY Table X. (Continued) 1 II ill X J N ^ r N I ~ Percentage coupling X 11 N ^ r ^ Percentage coupling X J N 11 x Ij V Percentage coupling . 0.84 29.4 1.16 34.6 1.16 14.7 0.83 31.1 1.17 36.9 1.17 15.6 0.82 32.8 1.18 39.2 1.18 16.4 0.81 34.4 1.19 41.6 1.19 17.2 0.80 36.0 1.20 44.0 1.20 18.0 0.79 37.6 1.21 46.4 1.21 18.8 0.78 39.2 1.22 48.8 1.22 19.6 0.77 40.7 1.23 51.3 1.23 20.4 0.76 42.2 1.24 53.8 1.24 21.2 0.75 43.8 1.25 56.2 1.25 22.0 0.74 45.2 1.26 58.8 1.26 22.7 0.73 46.7 1.27 61.3 1.27 23.5 0.72 48.2 1.28 63.8 1.28 24.2 0.71 49.6 1.29 66.4 1.29 24.9 0.70 51.0 1.30 69.0 1.30 25.6 0.69 52.4 .31 26.4 0.68 53.8 .32 27.1 0.67 55.1 .33 27.8 0.66 56.4 .34 28.5 0.65 57.8 .35 29.1 0.64 59.0 .36 29.8 0.63 60.3 .37 30.5 0.62 61.6 .38 31.1 0.61 62.8 .39 31.8 0.60 64.0 .40 32.4 .41 33.0 .42 33.7 .43 34.3 .... .44 34.9 .45 35.5 .46 36.1 .... .47 36.7 1.48 37.3 1.49 37.9 1.50 38.5 1.55 41.2 1.60 43.8 1.65 46.3 . 1.70 48.6 1.75 50.7 1.80 52.8 .... 1.85 54.8 1.90 56.6 1.95 58.4 2.00 60.0 TABLES 403 Table XLResonance Curve of the Current Effect [Art. 740] Let di and d z represent the decrements of the primary and secondary circuits, respectively, 7 2 e // the current effect in the secondary circuit and I\ // the same at resonance between the two circuits. The resonance curve is obtained by plotting the values of the ratio / 2 e// : I\ eff as ordinates, y, and the values of the dissonance between the two circuits as abscissae. Let x = - r the meaning of x\ and Xz being obvious from Fig. 470. Then : di + d z = x X 27r = xA The assumptions are : 1. z< 1.0. 2. di + d 2 <27r and 3. Very loose coupling between primary and secondary circuits. In the following table the value of A and log A is given for different values of y. y log A A y log A A 0.998 2.1472 140 0.958 1.4773 30.0 0.996 1.9963 99.2 0.956 1.4667 29.3 0.994 1.9078 80.9 0.954 1.4565 28.6 0.992 1 . 8449 70.0 0.952 1.4469 28.0 0.990 1.7960 62.5 0.950 1.4376 27.4 0.988 1.7560 57.0 0.945 1.4157 26.0 0.986 1.7221 52.7 . 940 1.3956 24.9 0.984 1.6926 49.3 0.935 1.3771 23.8 0.982 1.6666 46.4 0.930 1 . 3599 22.9 0.980 1.6433 44.0 0.925 1.3437 22.1 0.978 .6221 41.9 0.920 1.3285 21.3 0.976 .6028 40.1 0.915 1.3142 20.6 0.974 .5850 38.5 0.910 .3006 20.0 0.972 .5684 37.0 0.905 .2876 19.4 0.970 .5530 35.7 0.900 .2753 18. 8 5 0.968 .5386 34.5 0.89 .2522 17.9 0.966 .5249 33.5 0.88 .2308 17.0 0.964 .5121 32.5 0.87 .2110 16.3 0.962 1.4994 31.6 0.86 .1924 15.6 0.960 1.4883 30.8 0.85 .1748 15.0 404 WIRELESS TELEGRAPHY Table XI. (Continued) y log A i A y log A A 0.84 .1583 14.4 0.39 0.7011 5.02 0.83 .1425 13.9 0.38 0.6919 4.92 0.82 .1274 13.4 0.37 0.6827 4.82 0.81 .1130 13.0 0.36 0.6734 4.71 0.80 .0993 12.6 0.35 0.6638 4.61 0.79 .0859 12.2 0.34 0.6542 4.51 0.78 .0730 11.8 0.33 0.6444 4.41 0.77 .0606 11.5 0.32 0.6345 4.31 0.76 1.0485 11.2 0.31 0.6245 4.21 0.75 1.0367 10.9 0.30 0.6142 4.11 0.74 1.0253 10.6 0.29 0.6033 4.01 0.73 1.0141 10.3 0.28 0.5932 3.92 0.72 1.0032 10.1 0.27 0.5823 3.82 0.71 0.9931 9.84 0.26 0.5711 3.72 0.70 0.9822 9.60 0.25 0.5597 3.63 0.69 0.9719 9.37 0.24 0.5479 3.53 0.68 0.9619 9.16 0.23 0.5358 3.43 0.67 0.9518 8.95 0.22 . 5234 3.34 0.66 0.9422 8.75 0.21 0.5105 3.24 0.65 0.9326 8.56 0.20 0.4971 3.14 0.64 . 9230 8.38 0.19 0.4834 3.04 0.63 0.9137 8.20 0.18 0.4690 2.94 0.62 0.9045 8.03 0.17 0.4539 2.84 0.61 0.8953 7.86 0.16 0.4381 2.74 0.60 0.8862 7.69 0.15 0.4216 2.64 0.59 0.8772 7.54 0.14 0.4040 2.54 0.58 0.8683 7.38 0.13 0.3854 2.43 0.57 0.8594 7.23 0.12 0.3656 2.32 0.56 . 8505 7.09 0.11 0.3442 2.21 0.55 0.8418 6.95 0.10 0.3211 2.09 0.54 0.8330 6.81 0.09 0.2958 1.98 0.53 0.8243 6.67 0.08 0.2679 1.85 0.52 0.8156 6.54 0.07 0.2365 1.72 0.51 0.8069 6.41 0.06 0.2008 1.59 0.50 0.7982 6.28 0.05 0.1588 1.44 0.49 0.7895 6.16 0.04 0.1081 1.28 0.48 0.7808 6.04 0.03 0.0434 1.10 0.47 0.7721 5.92 0.02 0.9531-1 0.90 0.46 0.7634 5.80 0.01 0.8004-1 0.63 0.45 0.7546 5.68 0.44 0.7459 5.57 0.43 0.7370 5.46 0.42 0.7281 5.35 0.41 0.7192 5.24 0.40 0.7102 5.13 TABLES Table XII. Resonance Sharpness p = 405 ^-y- [Art. 700] ~r #2 di + d z p di + d* p di +d 2 P di + d 2 P 0.010 628 0.033 190 0.056 112 0.079 79.4 0.011 571 0.034 185 0.057 110 0.080 78.5 0.012 524 0.035 179.5 0.058 108 0.013 483 0.059 106.5 0.081 77.6 0.014 449 0.036 174.5 0.060 105 0.082 76.6 0.015 419 0.037 170 0.083 75.7 0.038 165 0.061 103 0.084 74.8 0.016 393 0.039 161 0.062 101 0.085 73.9 0.017 370 0.040 157 0.063 99.7 0.018 349 0.064 98.2 0.086 73.1 0.019 331 0.041 153 0.065 96.7 0.087 72.2 0.020 314 0.042 150 0.088 71.4 0.043 146 0.066 95.2 0.089 70.6 0.021 299 0.044 143 0.067 93.8 0.090 69.8 0.022 286 0.045 140 0.068 92.4 0.023 273 0.069 91.1 0.091 69.0 0.024 262 0.046 137 0.070 89.8 0.092 68.3 0.025 251 0.047 134 0.093 67.6 0.048 131 0.071 88.5 0.094 66.8 0.026 242 0.049 128 0.072 87.3 0.095 66.1 0.027 233 0.050 126 0.073 86.1 0.028 224 0.074 85.0 0.096 65.5 0.029 217 0.051 123 0.075 83.8 0.097 64.8 0.030 209 0.052 121 0.098 64.1 0.053 118.5 0.076 82.7 0.099 63.5 0.031 203 0.054 116 0.077 81.6 0.100 62.8 0.032 196 0.055 114 0.078 80.5 Table XIII. The Radiation Resistance of Antennae According to Art. lOOc, the radiation resistance, Rz, of an antenna whose height is h and form factor is a and which is erected on ground of high conductivity, is given by: = 2 X 1607T 2 ohms. In the following table the different values of the expression 160ir ( J are given. Hence the radiation resistance in ohms is found by multiplying the figure given in the table by the square of the form factor of the antenna. 406 WIRELESS TELEGRAPHY Wave length X in meters 300 400 500 600 700 800 900 1000 1500 2000 10 1.75s 0.987 0.632 0.439 0.332 0.247 0.195 0.158 0.0702 0.0395 15 3.95 2.22 1.42 0.987 0.725 0.555 0.439 0.355 0.158 0.088 20 7.02 3.95 2.53 1.75s 1.29 0.987 0.780 0.632 0.281 0.158 25 11.0 6.17 3.95 2.74 2.01 1.54 1.22 0.987 0.439 0.247 30 15.8 8.88 5.68s 3.95 2.90 2.22 1.75 1.42 0.634 0.355 35 21.5 12.1 7.74 5.37s 3.95 3.02 2.39 1.93 0.860 0.484 40 28.1 15.8 10.1 7.02 5.16 3.95 3.12 2.53 1.12 0.632 45 35.5 20.0 12.8 8.88 6.53 5.00 3.95 3.20 1.42 0.800 50 43.9 24.7 15.8 11.0 8.06 6.17 4.87 3.95 1.75 0.987 3 55 53.1 29.8 19.1 13.3 9.79 7.46 5.90 4.78 2.12 1.19 1 60 63.2 35.5 22.7 15.8 11.6 8.88 7.02 5.68s 2.53 1.42 d 65 74.1 41.7 26.7 18.5 13.6 10.4 8.24 6.67 2.96s 1.67 s 70 86.0 48.4 30.9s 21.5 15.8 12.1 9.55 7.74 3.44 1.93s 75 98.7 55.4 35.5 24.7 18.1 13.9 11.0 8.88 3.95 2.22 g 80 63 2 40 4 28 1 20.6 15 8 12.5 10.1 4 49 2 53 85 71.3 45.6 31.7 23.3 17.8 14.1 11.4 5.07 2.85 '3 90 .... 80.0 51.2 35.5 26.1 20.0 15.8 12.8 5.68 6 3.20 5 95 89.1 57.0 39.6 29.1 22.3 17.6 14.2s 6.33s 3.56 ;> 100 98.7 63.2 43.9 32.2 24.7 19.5 15.8 7.02 3.95 '3 '^ 110 76.4 53.1 39 29.8s 23.6 19.1 8.49 4.78 120 90. 9 6 63.2 46.4 35.5 28.1 22.7 10.1 5.68s 130 74.1 54.5 41.7 32.9 26.7 11.9 6.67 140 86.0 63.2 48.4 38.2 30.9s 13.8 7.74 150 98.7 72.5 55.4 43.9 35.5 15.8 8.88 160 82.5 63.2 49.9 40.4 18.0 10.1 170 93.1 71.3 56.3 45.6 20.3 11.4 180 80.0 63.2 51.2 22.7 12.8 190 89.1 70.4 57.0 25.3 14.2 200 98.7 78.0 63.2 28.1 15.8 TABLES 407 Wave length X in meters 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 10 0.0253 0.0175s 0.0129 0.00987 0.00780 0.00632 0.0052 0.00439 0.0037 0.00332 15 0.0568 0.0395 0.0290 0.0222 0.0176 0.0142 0.0117 0.00987 0.0084 0.00725 20 0.101 0.0702 0.0516 0.0395 0.0312 0.0253 0.0209 0.0175s 0.0149s 0.0129 25 0.158 0.110 0.0806 0.0617 0.0487 0.0395 0.0326 0.0274 0.0234 0.0210 30 0.227 0.158 0.116 . 0888 . 0702 0.0568s 0.0470 0.0395 0.0336 0.0290 35 0.309s 0.215 0.158 0.121 0.0955 0.0774 . 0639s 0.0537s 0.0458 0.0395 40 0.404 0.281 0.206 0.158 0.125 0.101 0.0835 0.0702 0.0598 0.0516 45 0.512 0.355 0.261 0.200 0.158 0.128 0.106 . 0888 0.0757 0.0653 50 0.632 0.439 0.322 0.247 0.195 0.158 0.130s 0.110 0.0934 0.0806 1 55 0.764 0.531 0.390 0.298s 0.236 0.191 0.158 0.133 0.113 0.0979 S 60 0.910 0.632 0.464 0.355 0.281 0.227 0.188 0.158 0.134 0.116 ; 65 1.07 0.741 0.544 0.417 0.329 0.267 0.221 0.185 0.158 0.136 .a 70 1.2.4 0.860 0.631 0.484 0.382 0.309s 0.256 0.215 0.183 0.158 g 75 1.42 0.987 0.725 0.554 0.439 0.355 0.294 0.247 0.210 0.181 a 80 1.62 1.12 0.825 0.632 0.499 0.404 0.334 0.281 0.239 0.206 85 1.83 1.27 0.931 0.713 0.563 0.456 0.377 0.317 0.270 0.233 S 90 2.05 1.42 1.04 0.800 0.632 0.512 0.423 0.355 0.303 0.261 Z 95 2.28 1.58 1.16 0.891 0.704 0.570 0.471 0.396 0.337 0.291 100 2.53 1.75s 1.29 0.987 0.780 0.632 0.522 0.439 0.374 0.322 '3 W 110 3.06 2.12 1.56 1.19 0.943s 0.764 0.632 0.531 0.452 0.390 120 3.64 2.53 1.86 1.42 1.12 0.910 0.752 0.632 0.538 0.464 130 4.27 2.96s 2.18 1.67 1.32 1.07 0.882 0.741 0.631 0.545 140 4.95 3.44 2.53 1.93 1.53 1.24 1.02 0.860 0.732 0.632 150 5.68s 3.95 2.90 2.22 1.76 1.42 1.17 0.987 0.840 0.725 160 6.47 4.49 3.30 2.53 2.00 1.62 1.34 1.12 0.957 0.825 170 7.30 5.07 3.72s 2.85 2.25 1.83 1.51 1.27 1.08 0.931 180 8.19 5.68s 4.18 3.20 2.53 2.05 1.69 1.42 1.21 1.04 190 9.12 6.33s 4.65 3.56 2.81s 2.28 1.88s 1.58 1.35 1.16 200 10.1 7.02 5.16 3.95 3.12 2.53 2.09 1.75 1.49s 1.29 BIBLIOGRAPHY AND NOTES ON THEORY 1 Works covering the general subject of radio-telegraphy: a. F. ANDERLE, Lehrbuch der drahtlosen Telegraphic und Telephonic, Leipzig and Vienna, 1912. fe. J. ERSKINE-MURRAY, A handbook of wireless telegraphy, its theory and practice. 3d Edit., London, 1911. c. J. A. FLEMING, The principles of electric wave telegraphy. 2d Edit., London, Longmans, Green & Co., 1910. d. G. W. PIERCE, Principles of wireless telegraphy. New York, McGraw-Hill Book Co., 1910. e. H. REIN, Radiotelegraphisches Praktekum. 2d Edit., Berlin, Springer, 1912. /. C. TISSOT, Manuel elementaire de telegraphic san fil. Paris, 1912. g. A. ZAMMARCHI, La telegraphia senza fili di Guglielmo Marconi. Bergamo, 1904. (Of historical interest only.) h. J. ZENNECK, Elektromagnetische Schwingungen und drahtlose Telegraphic. Stuttgart, 1905." i. Theoretical: C. TISSOT, Les oscillations electriques. Paris, 1910. 2 Special arrangements for the use of the Braun tube with rapid oscillations: L. MANDELSTAM, Jahrb., 1, 124, 1908. (The same method employed by D. ROSCHANSKY, Ann. Phys., 36, 281, 1911.) H. HAUSRATH, Phys. Zeitschr., 12, 1044, 1911; also Jahrb., 6, 185, 1912. K. ORT, Jahrb., 6, 119, 1912. E. L. CHAFFEE, Proc. Amer. Acad. Arts and Sciences, 47, 311 et seq., 1911. 3 W. FEDDERSEN, Pogg. Ann., 113, 437, 1861; also 116, 132, 1862. Also see Beriichte der sachs. Ges. der Wissenschaften, 61, 151, 1909. For frequency determina- tions, the method of HEMSALECH (C. R., 132, 912, 1901, illumination of a slit by the spark) gives particularly suitable pictures. 4 E. GEHRKE, Verhandl. Physik. Ges., 6, 176, 1904; Zeitschr. f. Instrumentenkunde, 15, 33, 278, 1905. Reproductions by means of the incandescent lamp oscillo- graph: H. DIESSELHORST, Ber. phys. Ges., 5, 320, 1907; 6, 306, 1908; ETZ, 29, 703, 1908. 6 W. THOMSON, Phil. Mag. (4), 6, 593, 1855. 6 J. A. FLEMING, Elecn., 63, 459, 1909. H. ANDERSON, Phys. Rev., 34, 34, 1912. 7 M. WIEN, Phys. Zeitschr., 11, 282 et seq., 1910. H. RIEGGER, Diss. Strassburg, 1911; Jahrb., 5, 35, 1911. For explanation, see D. ROSCHANSKY, Phys. Zeitschr., 11, 1177, 1910. 8 In regard to more recent work, see H. DIESSELHORST, Jahrb., 1, 263, 1908. 9 To be more accurate, this should be ^ L/o 2 , the energy transferred in a half cycle (see E. COHN, Das elektromagnetische Feld, p. 360. Leipzig, 1900). 10 F. RICHARZ and W. ZIEGLER, Ann. Phys., 1, 468, 1900. J. ZENNECK, Ann. Phys., 13, 822, 1904. 11 This refers to gaps in air. According to E. L. CHAFFEE, 2 a straight line amplitude curve is also obtained with aluminium electrodes in hydrogen and with carbon electrodes in air. 12 A. HEYDWEILLER, Ann. Phys., 19, 649, 1906; 25, 48, 1908. W. STUFF, Diss. Munster, 1907. H. BARKHAUSEN, Phys. Zeitschr., 8, 624, 1907. 408 BIBLIOGRAPHY AND NOTES ON THEORY 409 13 That is, R g is defined by = the energy consumed during one spark discharge. 14 This arrangement was proposed by MARESCA (Phys. Zeitschr., 4, 9, 1902), and the method, in the form described in the text, by K. SIMONS (Ann. Phys., 13, 1044, 1904). 15 Determinations of the gap resistance or decrement: G. REMPP, Diss. Strassburg and Ann. Phys., 17, 627, 1905. (His values, particularly for gaps over 6 mm. long, are too high as the effect of brush discharge was not understood at that time.) H. RAUSCH VON TRAUBENBERG and W. HAHNEMANN, Phys. Zeitschr., 8, 498, 1907. K. E. F. SCHMIDT, Phys. Zeitschr., 8, 617, 1907. C. RICHTER, Phys. Zeitschr., 10, 703, 1909. M. WIEN, Ber. physik. Ges., 12, 736, 1910; Ann. Phys., 29, 679 et seq., 1909. W. F. ZORN, Jahrb., 4, 269 et seq., 382 et seq., 1911. also whence T Fo *o = F 17 M. WIEN, Ann. Phys., 29, 679 et seq., 1909. 18 Measurements by W. EICKHOFF at the physikalisches Institut Braunschweig (see Phys. Zeitschr., 8, 497, 1907). 19 D. ROSCHANSKY, Jahrb., 3, 81, 1909. 20 W. EICKHOFF, Phys. Zeitschr., 8, 494, 1907. In regard to the voltage conditions in series spark gaps see P. NORDMEYER, Jahrb., 3, 334 et seq., 1910. 21 B. MONASCH, Ann. Phys., 22, 905, 1907. W. HAHNEMANN and L. ADELMANN, ETZ, 1907, 988, 1010. M. WiEN. 17 J. A. FLEMING and G. B. DYKE, EL, 66, 658 et seq., 1911. L. W. AUSTIN, Jahr. 5, 420, 1912. According to AUSTIN the glass furnished by the Wireless Specialty Apparatus Co. is particularly good. 22 This phenomenon is identical with the "corona" of high-tension transmission cir- cuits (see, e.g., W. PETERSEN, Hochspannungstechnik, p. 308 et seq., Stuttgart, 1911). 23 A. MEISSNER, Jahrb., 3, 57 et seq., 1909. 24 Detailed treatment in EMS, p. 498 et seq., 743 et seq. (Note 1 h) 26 According to F. HARMS (Ann. Phys., 23, 60, 1907) the velocity of propagation and hence also the frequency are less for wires with an insulating sheath. 26 M. ABRAHAM, Wied. Ann., 66, 435 et seq. F. HACK, Ann. PhyS., 14, 539, 1904. The field of an oscillator whose current amplitude is the same at all points (a = 1) has been calculated by H. HERTZ, Wied. Ann., 36, 1, 1888; Ges. Werke II, 45. 27 F. HACK, Ann. Phys., 18, 634, 1905. 28 Detailed treatment of oscillations in coils, P. DRUDE, Ann. Phys., 9, 593, 1902. J. A. FLEMING. 1 29 M. WIEN, Jahrb., 1, 474, 1908. 30 G. SBIBT, ETZ, 1902, 411. Also experiments in the physikal. Inst. Braunschweig. 410 WIRELESS TELEGRAPHY 31 This follows from the well-known "telegraph equation" of KIRCHHOFF. See, e.g., B. C. TissoT. 1 32 Experimental method for determining the current anti-node in an open oscillator, A. ESAU, Phys. Zeitschr., 13, 495, 1912. 33 If the current, /, is of the form 7 = /o sin wt, <;hen at a distance r, E = Eo cos I cot - J and M = M cos I tat ~ (The algebraic sign before E and M being in accordance with Fig. 37, p. 35, / is taken positive in direction from A to B.) 34 M. ABRAHAM, Theorie der Elektrizitat II, p. 286. Leipzig, 1905. Application to various forms of oscillators by A. MONTEL, Lum. el., 6, 199, 207, 1909. 35 I = I I dx (I = length of oscillator). 35 In this case, we have for each half of the oscillator, /o = |/o| A - ~ 1 I 2j o 1 I I dx l/ol 2 where x = distance from middle of oscillator. Hence 37 Here , . , . TTX -/O = |/0| COS -- 370 This follows directly from the fact that the radiation S = ^ [2Af ] in which [EM] is the product of the vectors E and M . 38 This is easily arrived at from M. ABRAHAM, 34 p. 301 et seq. 39 R. RUDENBERG, Ann. Phys., 25, 446, 1908. Also see H. BARKHAUSEN, Jahrb., 2, 40, 1908. P. BARRECA, Jahrb., 4, 31 et seq., 1910. 40 All difficulties which otherwise are apt to be encountered in coupled circuits can be avoided by proceeding as follows: At any point x on the oscillator, the current Furthermore let the energy consumed per second as heat be expressed by | RW x I*dx = I/ 1 ,* | and the following inch d, so far as the oscillati j LMl*dx = H\I\ 2 I (the integral in this and the following includes the entire oscillator), the energy of the magnetic field, so far as the oscillations are concerned, by BIBLIOGRAPHY AND NOTES ON THEORY 411 and the energy of the electrical field by Moreover I CWV*dx = KIF1 2 I C^v The differential equation of the oscillation is then : / 2 C CD 2 d 11/2 f L (l) 2 1 I J *" 2 J f 5i 2 or ../ r r J R^f(xydx + ^ J = This can be reduced to the same form as pertains to the natural oscillations of a condenser circuit, viz.: 1/1 of + by substituting the values: r R = I RMf(x)*dx, -/ I ^'(l),-/'^^2^/^ si The preceding applies to oscillators without condensers in series but is easily modified so as to apply to oscillators having series condensers. 41 A. BLONDEL, Assoc. franc, pour Pavancement des sciences. Congres d'Angers, 1903. 42 Elementary treatment of the action of capacities and inductive coils in antennae, see, e.g., A. GUYAU, Lum. 61., 16, 13, 1911. 43 Detailed treatment in EMS, p. 400 et seq. 44 Discussions of coefficients of self-induction and mutual induction : G. GLAGE, Jahrb., 2, 361 et seq., 501 et seq., 593 et seq., 1909, and particularly E. B. ROSA and F. W. GROVER, Bullet. Bur. of Standards, 8, 1 et seq., 1911. 46 Articles on the resistance, self-induction and capacity of coils of solid wire and wire braid: (1) Theoretical: A. SOMMERPELD, Ann. Phys., 24, 609, 1907. J. W. NICHOLSON, Jahrb., 4, 26 et seq., 1910. L. COHEN, Bull. Bur. of Stand- ards, 4, No. 76, 1907-1908. W. LENZ, Ann. Phys., 37, 923, 1912. H. G. MOLLER, Ann. Phys., 36, 738 et seq., 1911 (regarding braided wires). Experi- mental: TH. P. BLACK, Ann. Phys., 19, 157, 1906. A. MEissNER. 23 A. ESAU, reference list summarizing his articles: Jahrb., 4, 490 et seq., 1911. R. LINDE- 412 WIRELESS TELEGRAPHY MANN, reference list of articles: Jahrb., 4, 561 et seq., 1911. K. HERRMANN, Verb, physik. Ges., 13, 978, 1911. 46 Concerning the effective resistance of wires subjected to two simultaneous undamped, sinusoidal and damped, non-sinusoidal oscillations: BRYLINSKI, Bull, de la soc. intern, des electriciens (2), 6, 255, 1906. 47 Rheostat, for rapid oscillations, of wires having small cross-section and low conduc- tivity: C. TISSOT, Bull, de la Soc. intern, des electr. (2), 6, 340, 1906. W. HAHNEMANN, Jahrb., 2, 314, 1909. 48 P. BRENOT, Lum. el., 15, 259 et seq., 1911. 49 Resistance and current distribution in rectangular wire (bands) : W. EDWARDS, El., 68, 18, 1912 (theoretical). J. BETHENOD, Jahrb., 2, 379 et seq., 1909 (ex- perimental). 50 N. TESLA'S researches in polyphase currents, etc., by TH. C. MARTIN, pp. 222, 314. (Halle, 1895.) 51 Construction of C. LORENZ Co. to whose courtesy the illustration is due. 510 This is not the only possibility. For instance, the same image would be produced if the discharge frequency and the revolutions per second were in the ratio 3:4 or 5:4. 52 E. NESPER, Jahrb., 2, 92 et seq., 319 et seq., 3, 376 et seq., 1910. 53 The Rendahl variometer was apparently proposed independently by PERI (see ETZ, 32, 247, 1911). 54 Thanks are due to the DR. E. F. HUTH, G. m. b. H., Berlin, SO, Erdmannshof, for this illustration. 55 Thanks are due to DR. L. COHEN (Nat. Elec. Sign. Co.), for this illustration. 56 See J. MOSCICKI, ETZ, 25,' 527, 1904. C. MULLER (Ann. Phys., 28, 585 et seq., 1909) also proposed a good form of jar. 57 Compressed air (or gas) condensers, proposed by T. JERVIS-SMITH (Nature, 48, 64, 1893, quoted in El., 55, 912, 1905). R. FESSENDEN, ETZ, 1905, 950. M. WlEN. 17 In regard to the dielectric strength of compressed gases see M. WOLF, Wied. Ann., 37, 306, 1889, and E. A. WATSON, Journ. Inst. Elec. Engs., 40, 6, 1908. 570 According to G. W. PIERCE, l p. 114, the variable condenser was proposed by KORDA as early as 1893. 58 From a pamphlet of the physikalisch-technischen Laboratorium : DR. G. SEIBT, Berlin-Schoneberg. 59 From Jahrb., 4, 439, 1911. 60 Construction of H. BOAS Co. (Berlin). 61 From Jahrb., 4, 229, 1911. 62 See P. BRENOT, Lum. el. (2), 11, 427, 1910. 8V 63 This, of course, also follows directly from / = C -rr- 64 Gap length and breakdown potential: A. HEYDWEILLER, Wied. Ann., 48, 235, 1893. S. M. KINTNER, Proc. Amer. Inst. El. Engs., 24, 523, 1905. J. A. FLEMING. 1 More recent works are: J. ALGERMissEN. 65 E. VOIGT, Ann. Phys., 12, 403, 1903. C. MuLLER 56 (in conjunction with MULLER, see M. TOEPLER, Ann. Phys., 29, 153, 1909). E. HUPKA, Ann. Phys., 36, 440 et seq., 1911. W. WEICKER, ETZ, 32, 436 et seq., 460 et seq., 1911. In regard to prin- cipal points in measurement of gap length see M. TOEPLER, Ann. Phys., 19, 191, 1906; ETZ, 28, 998 et seq., 1907. 65 See J. ALGERMISSEN, Diss. Strassburg, 1906; Ann. Phys., 19, 1016, 1906. 66 E. WARBURG, Wied. Ann., 59, 1, 1896; 62, 385, 1897. 67 If V is the potential across a poor insulator of resistance R, then the quantity of electricity which is lost by leakage through the insulator in a given time, t, BIBLIOGRAPHY AND NOTES ON THEORY 413 is equal to ^ I I V\ dt (in which | F| is the absolute value of F). The quantity lost, therefore, increases as the duration of the potential increases. Concerning insulating materials for rapid oscillations see S. H. HILLS, EL, 65, 303, 1910. Veff for 7 = 7 oe T . sin wt and d"^. 2-n- [Art. 8c]. - tf -*( r" i ,\ 70 For one discharge we have I I z dt = ,,. 7 2 , if 7 = 7o ( 1 Tfi t } sin wt and a ^ 1 [Art. 9a]. 71 See A. WASMUS, Diss. Braunschweig, 1909. ETZ, 31, 199, 1910. 72 Jahrb., 5, 517, 1911; Jahrb., 6, 28, 1913. W. STEINHAUS Phys. Zeitschr., 12, 657, 1911. 73 A. ESPINOSA DE LOS MONTEROS, Jahrb., 1, 323, 1908. 74 Articles on bolometers: C. TISSOT, Ann. Chim. Phys. (8), 7, 1906 or separately: Etude de la resonance des system es d'antennes, p. 20 et seq., Paris, 1906. K. E. F. SCHMIDT, Phys. Zeitschr., 8, 601, 1907. BELA GATI, EL, 58, 983, 1907; Jahrb., 2, 109, 1908; Phys. Zeitschr., 10, 322, 897, 1909. J. RAUTENKRANZ, Phys. Zeitschr., 10, 93, 1909. H. ZOLLICH, Phys. Zeitschr., 10, 899, 1909. W. KEMPE, Phys. Zeitschr., 11, 331, 1910. B. S. COHEN, Journ. Inst. El. Engs., 39, 503, 1907. 75 Articles on thermocouples: H. BRANDES, Phys. Zeitschr., 6, 503, 1905. W. VOLGE, ETZ, 1906, 467. L. W. AUSTIN, Phys. Zeitschr., 12, 1133, 1226, 1911. C. M. DOWSE, EL, 65, 765, 1910. 76 W. DUDDELL, Phil. Mag. (6), 8, 91, 1904; Electrician, 55, 260, 1905. 77 W. GERLACH, Phys. Zeitschr., 13, 589, 1912. 78 A. ESPINOSA DE LOS MONTEROS, Jahrb., 1, 327, 1908. 79 L. W. AUSTIN, Bull. Bur. Stand, 7, 315, 1911; Phys. Zeitschr., 12, 1133, 1911. According to the latter article, the "perikon" detector produced a deflection of 3 scale divisions in a 2000 ohm galvanometer (1 scale div. = 1.28 X 10~ 9 amp.), for a MORSE dash when the tone was just audible in the most sensitive telephones. In regard to a magnetic detector for measuring purposes see R. ARNO, Lum. el. (2), 6, 344, 1909. 80 That is, &n = -Ls l2 ,-jf; 8 i2 = -Ls 2l -^ The differential equations for two circuits carrying quasi-stationary current and which are magnetically coupled are ^ + Ri ^df +Li d^ +Lsi2 ^ J = ll dl, d*I 2 d*I\ (1) If the circuits have pure conductive coupling, (2) 414 WIRELESS TELEGRAPHY in which Ri and Rz are the total resistances of th(x)] and therefore the values of R, C and L are different for the two oscillations (A. SLABY). So far as the author knows no theory (mathematical) which takes this fact into account has been worked out to date ; however, it is not probable that the results are much different than those obtained with the present theory. 81 J. v. GEITLER (Wien. Ber., 104, II, 169 et seq., 1895; Wied. Ann., 55, 513, 1895) and J. ZENNECK, Phys. Zeitschr., 4, 656, 1903. 82 Thanks are due to the TELEFUNKEN Co. of Berlin for this illustration. 83 See EMS, 634 et seq. 84 V. BJERKNES, Wied Ann., 44, 74, 1891; 55, 121, 1895. 85 These relations hold for primary circuits containing a spark gap only if the ampli- tude curve is an exponential curve. Nor do they hold in the case di = dz = d. In this case, the oscillation in the secondary circuit is of the form 7 = Iie~if t sin ut. 86 M. WIEN, Jahrb., 1, 462, 1908; Ann. Phys., 25, 625, 1908. 87 This is true only if there is no quenching action [Art. 62 et seq.] and even if this is BIBLIOGRAPHY AND NOTES ON THEORY 415 not the case, the amplitude of one of the oscillations can be zero. Whether or not this occurs depends upon the initial conditions (see 91 ). 88 H. DIESSELHORST, Ber. deutsch. physik. Ges., 5, 320, 1907; 6, 306, 1908; ETZ, 1908, 703. H. RAU, Jahrb., 4, 52, 1910. (RAU, in order to obtain spark photographs inserted a small gap in the secondary circuit also.) 89 P. DRUDE, Ann. Phys., 13, 512 et seq., 1904. For the theory of coupled circuits also see: B. MACKU, Jahrb., 3, 104 et seq., 329 et seq., 1910. A. KALAHNE, Jahrb., 4, 357 et seq., 1911. The disadvantage of the approximation methods of L. COHEN (Jahrb., 2, 448 et seq., 1909) and J. S. STONE (Lum. 61., 12, 435, 1910; ETZ, 33, 111, 1911) is that the degree of accuracy of their results cannot be predetermined. 90 C. FISCHER, Ann. Phys., 22, 265, 1907. M. WIEN, Phys. Zeitschr., 7, 871, 1906, 8, 10 et seq., 1907; ETZ, 1906, 839. Also see J. KAISER, Phys. Zeitschr., 10, 886, 1909. C. FISCHER, Phys. Zeitschr., 11, 420, 1910. W. BIERLEIN, Jahrb., 6, 29, 1912. E. TALSCH, Jahrb., 6, 35, 1912. 91 What follows holds true only under the following initial conditions: when t = 7i = Vx , /i = 0; 7 2 = 0; h = Under other conditions in fact it may happen that only one oscillation occurs. (A. SLABY, ETZ, 1904, 1086, M. WIEN, ETZ, 1906, 837.) The relations given in a and b are easily deduced from the work of P. DRUDE. 89 The vector dia- gram holds for the beginning (initial conditions) of the oscillations. After- ward it applies only to currents of constant frequency. 910 J. ZENNECK, Phys. Zeitschr., 6, 198, 1905. 92 M. WIEN, Jahrb., 1, 469, 1908; 4, 135, 1911; Ann. Phys., 25, 625, 1908; Phys. Zeitschr., 11, 76, 311, 1910. 93 H. BOAS, Jahrb., 5, 563, 1912. 94 A. ESPINOSAS DE LOS MoNTEROS, Jahrb., 1, 480, 1908. The hydrogen spark gaps have been very carefully studied by B. GLATZEL. Summary of his articles in Jahrb., 4, 400, 1911. For special methods of connection employing two or more spark gaps in series see Jahrb., 5, 437, 1912; EL, 68, 428 et seq., 1911. 95 R. RENDAHL, Phys. Zeitschr., 9, 203, 1908. B. GLATZEL, Ber. der deutschen phy- sik. Ges., 6, 54, 1908; Jahrb., 2, 65, 1908. A. ESPINOSA DE LOS MoNTERos. 94 950 The coupling, however, must not be so loose that the duration of half of a pulsa- tion occupies considerable time during which the oscillations have an appreci- able amplitude. For, as the two coupling waves exist during the first half pulsation, the object of the quenched gap would not be completely secured. 96 Regarding the relation of the spark frequency to the effectiveness of the quenching action, see H. ROHMANN, Phys. Zeitschr., 12, 649, 1911. 97 B. MACKU, Ann. Phys., 34, 941, 1911. 97a S. SUBKIS, Jahrb., 5, 507, 545, 1912; Diss. Braunschweig, 1911. Also see C. FISCHER. 115 98 G. GLAGE, Experimental investigations with the resonance inductor. Diss. Strassburg, 1907. H. BOAS, Jahrb., 3, 432, 607, 1910. K. ROTTGARDT, Phys. Zeitschr., 12, 652, 1911. 8. KIMURA, Jahrb., 5, 222, 1911, 6, 459, 1912. For the theory, see G. SEIBT, ETZ, 1904, 276. G. BENISCHKE, ETZ, 28, 2d issue, 1907. J. BETHENOD, Jahrb., 1, 534, 1908. Historical: P. BRENOT, Lum. 61. (2), 11, 167, 1910. 99 Integral of the equation for the discharge of condenser circuits in the case of aperiodic discharge. 416 WIRELESS TELEGRAPHY 100 Integral of the differential equation * + *-' - 101 More detailed treatment in EMS, Chap. XIII and XIV. 102 The wave-length, X, obtained in this way, is really too small by an amount AX, which is given by AX di(di+d 2 ) X 87T 2 (B. MACKLT, Jahrb., 2, 251, 1909). In regard to a zero method for determining the frequency, see G. SEIBT, Jahrb., 5, 407, 1912. 103 E. DORN, Ann. Phys., 20, 127, 1906. 104 See, e.g., the corresponding paragraphs in F. KOHLRAUSCH, Lehrbuch der praktis- chen Physik. 105 Comparative tests by different methods: H. DIESSELHORST.* Also see A. CAMP- BELL, EL 64, 612 et seq., 1912. 106 See, e.g., EMS, p. 711. 107 More accurate discussion of the resonance method and the necessary corrections: B. MACKu. 102 Also see M. K. GROBER, Phys. Zeitschr., 12, 121, 1911. 108 H. BRANDES, Ann. Phys., 22, 645, 1907. Graphic method by F. EGER, Diss. Greifswald, 1908. 109 L. KANN, Jahrb., 4, 297, 1911; Phys. Zeitschr., 11, 503, 1910. The BRANDES method is also the basis of an arrangement of P. LUDEWIG (Phys. Zeitschr., 12, 763, 1911; Jahrb., 5, 390, 1912), which gives a direct indication of the decrement. 110 See G. JONAS, Diss. Strassburg, 1907. 111 H. RlEGGER. 7 112 B. MACKtr, Ann. Phys., 34, 941, 1911. 113 M. WiEN 86 and Phys. Zeitschr., 9, 537, 1908. B. MACKU 102 and Phys. Zeitschr., 9, 437, 646, 1908. 114 S. LOEWE, Jahrb., 6, 325, 1912. 115 Concerning the POULSEN arc for measuring : RAUSCH VON TRAUBENBERG and B. MONASCH, Phys. Zeitschr., 8, 925, 1907; 9, 251, 1908. C. FISCHER, Ann. Phys., 28, 57, 1909; 32, 979, 1910. F. KIEBITZ, Ber. physik. Ges., 12, 99, 1910; Jahrb. ,2, 357 et seq., 1909. PHYS. TECHN. REICHSANSTALT: Zeitschr. f. Instru- mentenkunde, 28, 148, 1908. R. LINDEMANN, Ber. physik. Ges., 11, 28, 1909. K. VOLLMER, Jahrb., 3, 123, 1909. According to G. SZIVESSY (Jahrb., 3, 250 et seq., 1910) an arc in bisulphide of carbon vapor gives very steady oscillations. 116 In regard to precautions for the use of these spark gaps for measuring purposes, see S. LOEWE. 114 117 H. TH. SIMON, Phys. Zeitschr., 4, 737, 1903. G. W. PIERCE, Phys. Zeitschr., 6, 426, 1904. 118 W. EICKHOFF, Phys. Zeitschr. 8, 923, 1907. According to W. F. ZORN IS the point on copper electrodes causes an increase in the spark damping. J. A. FLEMING and H. W. RICHARDSON (EL, 63, 175, 1909) recommend air blowers to make the discharges more regular. This result, however, is not always accomplished (i.e., with all types of gaps) by a blower. 119 Another procedure is to make the coupling looser gradually until the value, which is obtained from the resonance curve in accordance with Art. 74 remains con- stant. The theoretical requirement for this condition is: ir 2 K 2 <^cW 2 . 86 Also see R. LINDEMANN'S 115 method. 120 M. WIEN, Phys. Zeitschr., 8, 764, 1907: with di = 0.11, d* = 0.015 and K = 0.014 the error becomes 30 per cent. BIBLIOGRAPHY AND NOTES ON THEORY 417 121 Complete treatment in the book, Die Frequenzmesser und Dampfungsmesser der Strahlentelegraphie, by E. NESPER, Leipzig, 1907. 122 E.g., EL, 68, 249 et seq., 1911. 123 THOR. G. THORNBLAD, Jahrb., 4, 97 et seq., 109 et seq., 217 et seq., 1911. 124 E. NESPER, Jahrb., 1, 112, 1907. 125 J. A. FLEMING, EL, 58, 495 et seq., 536 et seq., 1907. 126 Lum. el., 9, 391, 1910. 127 Ann. Phys., 8, 211, 1902. 128 R. HIRSCH, Jahrb., 4, 250, 1911. 129 L. MANDLESTAM and N. PAPALEXI, Jahrb., 4, 605, 1911. The high degree of ac- curacy of the method for determining frequency is well shown in the article by H. ROHMANN, Diss. Strassburg, 1911; Ann. Phys., 34, 979, 1912. 130 The dynamometer effect is = y I IJ 2 dt 131 Another method for determining the dynamometer effect by means of a differen- tial air thermometer by L. KANN 109 and L. ISAKOW, Phys. Zeitschr., 12, 1224, 1911. 132 The e.m.f. induced in the ring is displaced 90 with respect to 7' 2 , and 7 3 is in turn displaced 90 from this e.m.f. 133 On the assumption that the current curve is the same in both cases [see Art. lie, 2]. 134 W. EICKHOFP, Phys. Zeitschr., 8, 564, 1907. A. JOLLOS (Diss. Strassburg, 1907) was probably the first to show that an unsymmetrical resonance curve was the result of condenser brush discharge. Concerning the brush discharge of con- densers also see M. WiEN 17 and L. W. AusTiN. 21 135 From C. FiscHER. 90 136 This method was developed at the suggestion of the author by C. FISCHER, Ann. Phys., 19, 182, 1906. 137 The use of damping meter of P. LUDEWIG for determining the degree of coupling (Phys. Zeitschr., 13, 450, 1912) is also based upon this relation. 138 B. MACKU, Jahrb., 3, 580 et seq., 1910. 139 TELEFUNKEN Co., EL, 68, 171, 1911. 140 Experiments at the physik. Inst. Danzig-Langfuhr. 141 Spark photographs can also be used instead of the resonance curves. H. RAU. SS 142 R. A. FESSENDEN, ETZ, 1906, 690. 143 Antennae with increased end capacity were one of the first forms of antennae used by MARCONI and LODGE. Their fundamental advantages are given by A. BLONDEL. 41 144 Additional details regarding the TELEFUNKEN Go's antennae: SIEWERT, ETZ, 1906, 965. R SOLFF, ETZ, 1906, p. 875 et seq. COUNT ARCO, A. E. G. lec- tures, lecture of Dec. 9, 1911. H. BREDOW, Jahrb. der Schiffbau-technischen Ges., 1912, 105 et seq. Various articles in the TELEFUNKENZEITUNG. 145 O. LODGE and A. MUIRHEAD, EL, 51, 1036, 1903. See EL, 62, 170, 1908. 146 COUNT ARCO. 144 147 L. W. AUSTIN, Bullet. Bur. Stands, 7, 315 et seq., 1911. 148 Various constructions for masts: Jahrb., 3, 203, 521, 1910; 4, 309, 652, 1911. EL, 68, 213, 1911. 149 O. LODGE and A. MUIRHEAD at times erected their counterpoise several meters above the ground (see Jahrb., 3, 1, 1909). 150 W. BURSTYN, ETZ, 1906, 1117. F. KIEBITZ, Ann. Phys., 32, 961, 1910. M. REICH, Phys. Zeitschr., 13, 228 et seq., 1912; Jahrb., 6, 176 et seq., 253 et seq., 1911. H. TRUE, Jahrb., 5, 125 et seq., 1911. P. BARRECA, Jahrb., 6, 285 et seq., 1912. 27 418 WIRELESS TELEGRAPHY 151 Regarding radio-apparatus for airships and tests therewith see Jahrb., 3, 315, 434, 1910, 4, 227, 1911, 6, 70, 1912. FERRIE, Lum. el. 12, 99 et seq., 1910. TELE- FUNKENZEITUNG, 1, 66, 1911. K. SOLFF, Jahrb., 3, 392 et seq., 1910. K. LUBOWSKY, ETZ, 32, 1265, 1911. M. DIECKMANN, Jahrb., 6, 51, 1912. "Luftfahrt und Wissenschaf t " No. 2, 1912. P. LUDEWIG, Jahrb., 6, 10, 1912. H. MOSLER, Jahrb., 6, 44, 1912. 152 Regarding oscillations induced in all metal parts near quenched gap circuits, see 5. LOEWE. 114 153 According to DR. MEISSNER, however, the dangerous effect upon the ropes and the bag of the balloon is greatly increased when large current effect is used. 154 Tests of the accuracy of the methods given in Art. 97 : A. ESAU, Phys. Zeitschr. 13, 658, 1912. 155 C. FISCHER, Ann. Phys., 32, 979 et seq., 1910. 156 Concerning the increase in antenna capacity due to ice, rain, snow, etc., and the relation between antenna damping and weather conditions see A. ESAU, Phys. Zeitschr., 13, 721, 1912. 157 COUNT ARCO, ETZ, 1910, 508. 158 See, e.g., B. C. TissoT, 74 p. 139, 148 et seq. 169 Regarding antenna insulators used by the TELEFUNKEN Co., see H. BREDOW. 144 For methods of reducing brush discharge see Jahrb., 4, 441, 1911. H. LANGE, Jahrb., 4, 442, 1911. 160 COUNT ARCO, Jahrb., 2, 551 et seq., 1909. 161 Regarding total antenna resistance and its determination, see C. FISCHER, Phys. Zeitschr., 12, 295, 1911. L. W. AUSTIN, Phys. Zeitschr., 12, 924, 1911; Jahrb., 6, 574 et seq., 1912. There seems to be a relation between the total antenna resistance and the wave-length of the oscillation, such that the total resistance first decreases and then again increases in a straight line (uniformly) as the wave-length is increased. The decrease at first is probably due to the decrease in R% accompanying the increase in wave-length, the subsequent increase in total resistance must be due to an increase in the resistance of the earth, at any rate it varies with the amount of moisture in the ground. 162 J. ERSKINE-MURRAY, Jahrb., 6, 499, 1912. M. REICH, Phys. Zeitschr., 13, 228 et seq., 1912. 163 8 = I E x dx, where dx is an element of the antenna, E x the component of the electric field strength along this element and h the length of the antenna. 164 Bullet. Soc. d'encouragement p. Pindustrie rationale, 3, 1632, 1898. 165 F. BRAUN, D.R.P. No. 109378 (1899), Electrician, 62, 19, 1904; Phys. Zeitschr., 6, 193, 1904. 166 L'Electricien, 42, 107, 1911. 167 Literature of compressed air spark gaps, F. JERVIS-SMITH, EL, 63, 720, 1909. 168 G. EICHHORN, D.R.P., 157056 (1903). The modification (Fig. 218) of his orig- inal arrangement is due to P. PICHON (TELEFUNKEN Co.). 169 B. GLATZEL, Phys. Zeitschr., 11, 893, 1910. 170 S. EISENSTEIN, EL, 65, 848, 1910. 1700 R. C. GALLETTI (EL, 66, 570, 1911; ETZ, 32, 597, 1911, D.R.P., 245358). 171 Other methods of connection by G. SEIBT, D.R.P., 241114 (1909). B. MACKU, EL, 68, 429, 1911. 172 Jahrb., 2, 229, 1909. 173 COUNT ARCO, 160 B. GLATZEL, Jahrb., 2, 90, 1908. 174 A two plate spark gap was probably first proposed by T. B. KINRAIDE, U. S. Patent 623316 (1898), D.R.P., 108924 (1899). BIBLIOGRAPHY AND NOTES ON THEORY 419 175 EL, 63, 174 et seq., 374 et seq., 1909; 64, 153 et seq., 1909. Tests made by W. H. ECCLES and A. J. MACKAWER, EL, 64, 386, 1909; Jahrb., 4, 294, 1911. 176 COUNT ARCo, 160 Jahrb., 4, 79 et seq., 1910, ETZ, 31, 506 et seq., 1910. 177 O. SCHELLER, Jahrb., 5, 243, 1911. 178 W. PEUCKERT, Jahrb., 3, 199, 1909. A. WASMUS. L. H. WALTER, EL, 64, 550 1910. 179 Regarding other stations of the TELEFUNKEN Co., see COUNT ARCO, 176 H. BREDow. 144 180 F. G. LORING, EL, 67, 27, 1911. 181 H. RAU. 88 182 See C. MtiLLER 56 and then also M. ToEPLER. 64 183 Such meters were built, though for quite other purposes, by Siemens and Halske, Berlin, at the suggestion of the author. 184 See, e.g., G. BRION, Leitfaden zum elektrotechnischen Praktikum, p. 102 (B. G. TEUBNER, 1910). The Dolezaleck electrometer can also be used for power measurements; e.g., see M. REICH. 15 185 W. BURSTYN, Jahrb., 6, 217, 1912, proposes methods for making and breaking the circuit of the field excitation current. 186 See P. O. PEDERSEN, Jahrb., 4, 524, 1911. Regarding the high-speed telegraph apparatus used by POULSEN, see ETZ, 32, 1164, 1911. 187 EL, 60, 546, 883, 1908. Other spark gaps having smooth rotating electrodes: S. EISENSTEIN, Jahrb., 5, 245, 1911. W. BURSTYN, Jahrb., 6, 212, 1912. 188 L'Electricien, 42, 107, 1911. 189 EL, 65, 847, 1910. A similar arrangement by G. FERRIE, EL, 65, 135, 1910. 190 EL, 64, 512 et seq., 1910. This gap is used by the Soc. franc, radioelectrique. 191 G. MARCONI, EL, 67, 532, 1911; EL World, 59, 887, 1912; Jahrb., 6, 438, 1913; also see E. NESPER, Helios, 18, 429, 1912. As MARCONI employs relatively loose coupling (5 per cent.), the duration of half a pulsation and hence of the time during which there are two coupling waves present in the antenna, is rather long. Consequently they are evident in the resonance curve. See 95a and Art. 90a. 192 See, e.g., C. C. F. MONCKTON, EL, 56, 514, 1906. 193 W. H. ECCLES and A. J. MACKOWER (Jahrb., 4, 253, 1911; EL, 65, 1014, 1910) find a considerably lower efficiency from their measurements, which latter, however, are open to criticism. 194 W. DUDDELL, Phil. Mag. (6), 9, 299, 1905; Proc. Royal Inst., May 17, 1912. Also see Jahrb., 4, 202, 1911. 195 E. F. W. ALEXANDERSON, Trans. Amer. Inst. EL Eng., 28, I, 399 et seq., 1910. 196 R. A. FESSENDEN, D. R. P., 228,365, 1908. 197 R. GOLDSCHMIDT, ETZ, 32, 54, 1911; Jahrb., 4, 341 et seq., 1911. 198 w e ma y either conceive the alternating field of frequency JV' as made up of two rotating fields of opposite direction or we may proceed on the basis that the magnetic flux passing through R must be of the form A sin (2irN . t + ) cos (2irN f . t + ') sin [2w(N + N')t + (a + ')] + sin [27r(AT - N')t + (a - ')] For the theory of the GOLDSCHMIDT machine see E. RUSCH, Jahrb., 4, 348 et seq., 1911. B. MACKU, Jahrb., 5, 5, 1911. See Note. 344 1980 In the commercial form, condensers d and C 3 are omitted. 199 Very little has so far been published regarding the new high frequency generator of COUNT ARCO, which was exhibited at the international convention in London in 1912. See Jahrb., 5, 529, 1912. TELEFUNKENZEITUNG, Vol. 2, No. 7, p. 18. In regard to the generation of undamped oscillations by means of a series ma- 420 WIRELESS TELEGRAPHY chine with a condenser connected in parallel see F. FITZGERALD (Eclair. e"l., 18, 386, 1892). O. M. CORBINO (Phys. Zeitschr., 8, 924, 1907; 9, 195, 704, 1908; Electrician, 61, 56, 1908). R. RUDENBERG (Phys. Zeitschr., 8, 668, 1907; 9, 556, 1908). H. BARKHAUSEN, Das Problem der Schwingungser- zeugung. Diss. Gottingen, 1907, p. 37. It is impossible to conclude from the results obtained to date whether it will ever be possible to produce undamped oscillations for wireless telegraphy by this method in a practical and useful way. 200 EL. THOMSON, U. S. Pat., July 18, 1892 (quoted in EL Review, 60, 328, 1907) ; U. S. Pat. No. 500630, July 4, 1893. N. TESLA in MARTIN'S "Nichola Teslas Untersuchungen iiber Mehrphasenstrome." Halle, 1895. FESSENDEN claims to have made his first attempts in this direction in 1899. 201 W. DUDDELL, Electrician, 46, 269, 310, 1900. 202 J. WERTHEIM-SALOMONSON, Electrician, 52, 126, 1904. Eclairage electr., 38, 144, 1904. N = 400,000 eye. per sec. 203 V. POULSEN, Danish Pat. 5590 (Sept. 9, 1902), D.R.P. 162945 (July 12, 1903). 204 p or further details see the TELEFUNKEN Co.'s pamphlet describing their standard radio-telephone station. C. SCHAPIRA on the efficiency of the high frequency arc lamp with subdivided arc. Diss. Charlottenburg, 1908 and Jahrb., 2, 54 et seq., 1908. 205 P. BRENOT states in Lum. 41. (2), 11, 170, 1910 (also see Lum. el. (2), 11, 197, 1910) that A. BLONDEL employs two plates in petroleum as electrodes and impresses about 2000 volts across the arc. It is claimed that this gives greater regularity than the POULSEN method but does not secure high frequencies so easily. The author does not know whether the BLONDEL method, which is very similar to the PEUCKERT method, has ever been used in practice. According to Jahrb., 4, 522, 1911, F. JACOVIELLO employs metallic electrodes, potentials of 40,000 to 80,000 volts and impinges a stream of gas upon the arc, approximately in the direction of the length of the arc. Whether a quenched gap or undamped oscillations were used is not clear from what has been published. 206 From W. DUDDELL, Proc. Royal Inst., May 17, 1912. 207 Jahrb., 1, 307, 1908. 208 Data on POULSEN stations: a. LYNGBY and CULLERCOATS, Jahrb., 1, 154 et seq., 1907; Electrician, 60, 355 et seq., 1907. 6. KNOCKROE, Jahrb., 1, 430, 1908; ETZ, 1908, 15. 209 According to the C. LORENZ Co., this method of connection originated with W. HAHNEMANN and O. SCHELLER. 210 P. O. PEDERSEN, EL, 60, 547, 1908. C. LORENZ, Jahrb., 4, 333, 1911. 211 H. REIN, Jahrb., 4, 196, 1911 and "Der radiotelegraphische Gleichstromtonsender," Langensalza, 1912. 212 A large number of investigations of these phenomena have been made during recent years, the principal ones being the following : a. O. M CORBINO, Atti. Assoc. Elettrotecnica Ital., Oct., 1903 (mentioned in Phys. Zeitschr., 9, 197, 1908). 6. A. BLONDEL, Eel. El., 44, 41 et seq., 81 et seq., 1905. BLONDEL was the first to distinguish the different kinds of oscillations. c. H. BARKHAUSEN, Jahrb., 1, 234 et seq., 1907. d. H. TH. SIMON: Various articles, in part jointly with M. REICH. The articles are quoted in the general discussion by H. TH. SIMON in Jahrb., 1, 16, 1907. e. W. DUDDELL, Electrician, 46, 268, 310, 1900. /. G. GRANQVIST, Nov. Act. Reg. Soc. Scient. Upsaliensis (4), 1, No. 5. g. E. RIECKE, Gottinger Nachr. Math.-phys. Kl., 1907, 253. BIBLIOGRAPHY AND NOTES ON THEORY 421 h. K. H. WAGNER, "Der Lichtbogen als Wechselstromerzeuger." Leipzig, 1910. 1. For a comprehensive survey of this subject see H. BARKHAUSEN, "Das Problem der Schwingungserzeugung." Diss. Gottingen, 1907. In that article one question, namely, under what conditions the various kinds of oscillations are stable, which has not been considered in this book, is dis- cussed in detail. In other respects the treatment of this subject in what follows, is very similar to that given by BARKHAUSEN. 213 This is really a portion of a sine curve. See, e.g., EMS, p. 547. 214 The charging curve is determined by the equation V = 7 |_1 -e S^J in which VQ = dynamo voltage, C = capacity of condenser. 215 G. W. NASMYTH (Jahrb., 5, 269 et seq., 367 et seq., 1912) has given formulae for the frequency of oscillations generated by the arc method. These formulae how- ever have been discussed by K. VOLLMER IIS and P. O. PEDERSEN (Jahrb., 6, 496, 1912). 216 Regarding the function of the magnetic blowout see H. RAUSCH VON TRAUBENBERG, ETZ, 28, 559, 1907. H. TH. SIMON, 208 p. 65. H. BARKHAUSEN, 208 p. 256. K. .BIRKELAND (Jahrb., 2, 137, 1908) suggests a radial magnetic field for producing a rotating arc. 217 See H. BARKHAUSEN, Jahrb., 2, 40, 1909. 218 In Arts. 138, 139, 140 and 142 it is assumed that (1) the oscillations are undamped, (2) the atmosphere is an absolute non-conductor; furthermore in Arts. 138 to 140 it is taken for granted that the assumed conductivity holds for the entire portion of the earth which comes into consideration. As to the first assump- tion, L. W. AUSTIN (Jahrb., 6, 524, 1912) was unable to find any difference between undamped waves and waves having a decrement of 0.15 at a distance of 30 miles. 219 As early as 1898, A. BLONDEL (Compt. rend. Assoc. franc.. Avancement des sciences. Congres de Nantes, 1898, p. 212 et seq.) pointed out in conjunction with a remark of POINCARE that the action of the earth in a grounded transmit- ter could be replaced by that of an image of the transmitter, i.e., that a grounded transmitter can properly be conceived as one-half of a HERTZ lineal oscillator. 220 A comparison of the two limiting cases, the lineal transmitter (Figs. 27-30) on one hand with transmitter having uniform current amplitude throughout, as shown for example in EMS, Figs. 613-621. 221 J. ZENNECK (Ann. Phys., 23, 846, 1907). Previous to this, K. ULLER (Diss. Rostock, 1903) had already investigated the action of the waves under the assumption that they are entirely surface waves and that the earth's surface possesses a high degree of conductivity. 222 A. SOMMERFELD, Ann. Phys., 28, 665, 1909; Jahrb., 4, 158, 1910. 223 Regarding the conductivity of sea water, earths and rocks, see H. SCHMIDT, Jahrb., 4, 636 et seq., 1911. K. ULLER, Jahrb., 4, 638, 1911. H. LOEWY, Ann. Phys., 36, 125 et seq., 1911 and discussion thereof by J. A. FLEMING, Jahrb., 5, 515, 1912. 224 This conception that the waves of radio-telegraphy are of the nature of surface waves was probably first presented by A. BLONDEL 219 and by E. LECHER (Phys. Zeitschr., 3, 273, 1901-1902). Also see K. ULLER, "Die Mitwirkung der'Erde und die Bedeutung der Erdung in der drahtlosen Telegraphic, Jahrb., 2, 8, 1908. 225 P. EPSTEIN, Jahrb., 4, 176 et seq., 1910. 422 WIRELESS TELEGRAPHY 226 H. POINCARE, Jahrb., 3, 445, 1910. J. W. NICHOLSON, Review of his articles, Jahrb., 4, 20, 1910; Phil. Mag., 21, 281, 1911. H. W. MARCH (Ann. Phys., 37, 29, 1912. Note corrections in subsequent issue of Ann. Phys.). H. MAC- DONALD, Phys. Zeitschr., 10, 771, 1909. 227 H. B. JACKSON, Proc. Royal Soc., 70, 254 et seq., 1902. 228 W. DUDDELL and J. E. TAYLOR, Electrician, 55, 260, 1905. C. TISSOT, Electri- cian, 56, 848, 1906. 229 F. HACK, Ann. Phys., 27, 43, 1908. The assumptions are the same as in. 221 230 Electrician, 55, 409, 1905. 231 F. KIEBITZ, Verh. physik. Ges., 13, 876 et seq., 1911. 232 A reflection of this kind has been demonstrated in laboratory experiments with very short waves: F. ERB, Diss. Braunschweig, 1912. Data on observations in practice: P. SCHWARZHAUPT, ETZ, 31, 113, 1911. 233 See L. ZEHNDER, ETZ, 32, 1101, 1911. 234 In regard to the influence of the weather upon the antenna oscillations see A. EsAU, 156 O. GULDENPFENNIG, Jahrb., 5, 73, 1911. WILDMANN (see ERSKINE- MuRRAY 1 ) made systematic observations for over a year on the effect of the weather upon the communication between two stations. 235 H. EBERT (Jahrb., 4, 160, 1911) found the conductivity of the air at a height of 2500 m., in bright sunlight and in a downward current of air, to be twenty- three times as great as the conductivity just over the earth's surface. Re- garding the effect of meteorological conditions upon the ionization of the atmosphere see K. FISCHER, ETZ, 32, 339, 1911. 236 Jahrb., 5, 532, 1911. 236a A. BLONDEL 41 was probably the first to indicate that these upper strata might play an important part in determining wave propagation (see, e.g., B. J. ERKSINE- MuRRAY 1 ). This view is based on the assumption of a very good conductivity for the upper layers of the atmosphere. There is no justification for supporting this assumption by reference to conditions in the GEISSLER tube or in J. J. THOMSON'S current loop without electrodes, for in both these cases the gas is ionized by a very strong electric field, which does not exist in the upper atmos- pheric strata in wireless telegraphy. 237 J. A. FLEMING, The Marconigraph, 2, 179, 1912. G. W. PIERCE,* p. 139 states that A. E. KENNELLY has shown the effect of day and night to be due to a change in the wave front. 238 According to DR. A. MEISSNER, heavy winds of long duration, which tend to eradi- cate existing heterogeneity in the atmosphere, increase the distance effect. LEE DE FOREST (Jahrb., 6, 167, 1912) reports a very remarkable observation of interference due to heterogeneity. 239 G. MARCONI, EL, 49, 521, 1902; 54, 824, 1905. 240 See Jahrb., 5, 621, 1912; 6, 151, 154, 1912. A. TURPAIN, C. R., 154, 1457, 1912. W. H. ECCLES, EL, 69, 109, 1912. J. A. FLEMING, EL, 69, 190, 1912; TELE- FUNKENZEITUNG, 1, 89, 1912. So many observers have failed to find any effect due to solar eclipses and others have found so slight an effect, that it may be concluded that this effect is hardly greater than the extent of the errors involved in these measurements. 241 G. MARCONI, EL, 64, 379, 1909. H. J. ROUND had already (EL, 56, 714, 1906) stated that the difference between day and night range is much greater with short- than with long waves. 842 Tests made at the Braunschweig physik. Inst., 1907. 243 For tests on damping of antennse in daylight and at night see note 234 ; also H. MOSLER, ETZ, 30, 301, 1909 and P. SCHWARZHAUPT, ETZ, 32, 1313, 1912. 244 L. W. AUSTIN, Jahrb., 5, 75, 1911; Bull. Bur. Stands., 7, 315 et seq., 1911. BIBLIOGRAPHY AND NOTES ON THEORY 423 245 L. W. AUSTIN, Jahrb., 6, 417, 1912. 2 K. SOLFF, ETZ., 1906, 896. 247 J. A. FLEMING/ J. ERSKiNE-MuRBAY 1 and G. W. PiEBCE 1 give general and in some cases more complete presentations of this subject. Also see S. SACHS, Jahrb., 1, 130, 279, 434, 584, 1908. E. NESPER, Jahrb., 4, 312, 423, 534, 1911. 248 C. TISSOT, Electrician, 56, 848, 1906; Industrie electrique, 14, 161, 1906; Journal de Physique, 6, 279, 1907. 249 G. MARCONI, Proc. Royal Soc., 77, 413, 1906; Electrician, 57, 100, 1906. 250 Regarding thermal detectors: W. H. ECCLES, El., 60, 587, 1908. C. TISSOT, Jahrb., 2, 115 et seq., 1908. E. NESPER 247 , references in Jahrb., 3, 370, 430 (1910); 4, 232 et seq. (1911). On thermal detectors with a rotating electrode, see El., 62, 211, 1908 (L. W. AUSTIN); Jahrb., 2, 144, 1908 (TELEFUNKEN). 251 Review of various magnetic wave indicators not covered by note 247 : L. H. WALTER, Electrician, 55, 83, 1905. For explanation of their action see L. H. WALTER and E. MADELUNG, Ann. Phys., 17, 861, 1905; W. H. ECCLES, Elec- trician, 57, 742, 1906; J. RUSSEL, Proc. Royal Soc., Edinburgh, Nov. 20, .1905. E. WILSON, Electrician,^ 51, 330, 1897, was probably the first to describe a magnetic detector. 252 As the magnetic detector is most sensitive in a definite portion of the magnetiza- tion cycle, the magnetic detector of the so-called BALSILLIE system is comprised of three detectors, of the type shown in Fig. 317, each of which is automatically connected into circuit at the moment when it is in the most sensitive part of the magnetization curve (Jahrb., 4, 292, 1911; El., 64, 512 et seq., 1910). 253 G. MARCONI, Electrician, 54, 825, 1905. 254 R. ARNO, Electrician, 55, 469, 1905; ETZ, 1904, 480. J. A. EWING and L. H. WALTER, Proc. Roy. Soc., 73, 120, 1904. L. H. WALTER, Proc. Roy. Soc., 77, 538 et seq., 1906. W. PEUCKERT, ETZ, 1904, 992. A. G. Rossi, Phys. Zeitschr., 10, 549, 1909. R. A. FESSENDEN, D.R.P., 227102 (1909). 255 Review of a large number of articles dealing with the action of the coherer in note 247 and also by P. WEISS, Journal de Phys. (4), 5, 462, 1906. A. BLANC, Journal de Phys. (4), 4, 743, 1905. 256 German patent application by A. KOEPSEL in 1902. O. LODGE and A. MUIRHEAD, Electrician, 50, 930, 1903. 267 L. H. WALTER, Jahrb., 2, 120, 1908; Electrician, 61, 683, 1908. 258 J. E. IVES, Jahrb., 4, 112, 1910. 259 Regarding the liquid barretter see S. M. KINTNER, Proc. Amer. Inst. El. Engrs., 26, 65 et seq., 1907. J. E. IVES, Phys. Zeitschr., 11, 1181, 1910. 26 Jahrb., 5, 432, 1912. 261 Also see C. TISSOT, Electrician, 60, 25, 1907; C. R., 145, 226, 1907. J. S. SACHS, in Jahrb., 1, 584 et seq., 1908, quotes additional articles on the action of the electrolytic detector. 262 R. FESSENDEN, ETZ, 1905, 950. 263 F. BRAUN, ETZ, 1906, 1199; Electrician, 58, 569, 1907. PSILOMELAN detector, Jahrb., 4, 432, 1911. DUNWOODY detector, El. World, 48, 370, 1906. G. W. PIERCE detector, Lum. el., 1, 92, 1908; Jahrb., 3, 370, 1910. G. J. PICKARD detectors, Jahrb., 3, 430, 1910; Lum. el., 11, 172, 1910 (article by P. BRENOT). W. H. ECCLES, EL, 60, 588, 1908. 264 Regarding the action of crystal detectors see G. W. PIERCE, l H. SUTTON, EL, 69, 66, 1912. C. TISSOT, 1'Electricien, 39, 331, 1910. R. H. GODDARD, Phys. Rev., 34, 423, 1912. 265 J. A. FLEMING, Proc. Roy. Soc., 74, 476, 1905; El. 55, 303, 1905. Data relative to incandescent lamp type of detectors and their connections in Electrician, 61, 804, 843, 1006, 1908; 62, 211, 1908; 63, 504, 1909; 64, 68, 1909. Review of 424 WIRELESS TELEGRAPHY various detectors with rarified gases: C. TISSOT, EL, 58, 729, 1907; ETZ, 1908, 172. 266 A. WEHNELT, Ann. Phys., 19, 153, 1906. 267 H. BRANDES, ETZ, 1906, p. 1015. 268 Jahrb., 3, 429, 1910 and Q. MAJORANA, Jahrb., 2, 347 et seq., 1909. 269 P. LUDEWIG, Jahrb., 3, 411, 1911 (electrolytic cell); G. W. PIERCE, Jahrb., 3, 498, 1910; El. Review, 28, 56 et seq., 1909; El., 64, 183 et seq., 1909 (electrolytic cell); El., 64, 425, 1909 (crystal detectors). K. BANGERT, Phys. Zeitschr., 11, 123 et seq., 1910 (galena detector). L. W. AUSTIN, Bull. Bur. Stands., 6, No. 1, 1908. W. H. ECCLES, El., 65, 735, 1910; EL, 66, 166 et seq., 1910. 270 Compare C. TISSOT, Electrician, 58, 730, 1907; 60, 25, 1907. J. A. FLEMING.! 271 Apparatus or methods for testing detectors described in the following: J. A. FLEMING and G. B. DYKE, EL, 63, 216, 1909. P. JEGOU, ETZ, 720, 1908. The commercial form of detector testing apparatus of the TELEFUNKEN Co. is described in Jahrb., 6, 391, 1913. 272 Jahrb., 4,212 et seq., 1910. 273 COUNT ARCO. 160 G. EICHHORN, Jahrb., 6, 301 et seq., 1911. For other proposed method for sound intensification, see P. JEGOU, 1'Electricien, 37, 129, 1910. HENRY, 1'Electricien, 38, 11, 1910. 274 Phys. Zeitschr., 13, 38, 1912. 275 For further details see H. SIMON, Jahrb., 2, 409 et seq., 1909. 276 Bulletin No. 12 of the TELEFUNKEN Co. Capillary relay of ARMSTRONG-ORLING, ETZ, 1906, p. 385. M. CANTOR had already constructed a similar relay as early as 1900. 277 COUNT ARCO. 160 G. EICHHORN, Jahrb., 4, 405 et seq., 1911. Also see ETZ, 32, 776, 1911. Regarding proposal of C. LORENZ to use a selenium cell see Jahrb., 3, 622, 1910. 278 J. TAYLOR, EL, 41, 278 et seq., 1911. GRUNICKE, ETZ, 32, 64, 1911. TH.^BAKER, ETZ, 32, 696, 1911 and EL, 67, 363, 1911. 279 See EL, 54, 825, 1905; 63, 908, 1909; 1'Electricien, 39, 93, 1910; Jahrb., 4, 524, 1911. ETZ, 32, 1164, 1911. 280 MARCONI, in some of his stations, makes use of the so-called "earth arrester," which consists in the main of two metal plates close together, inserted in the ground connection of the antenna and having the receiving circuit in parallel thereto. When transmitting the spark resulting between these plates short-circuits the receiving circuit. As soon as transmission is over the receiving system is back in circuit. See E. NESPER. ISI 281 G. O. SQUIER, Electrician, 54, SZQetseq., 1905; 56, 453, 1905. 282 See R. RiiDENBERG, Ann. Phys., 25, 446, 1908. Also see H. BARKHAUSEN, Jahrb., 2, 40, 1908; 5, 261, 1912. 283 J. ERSKiNE-MuRRAY. 1 This also shows illustrations of LODGE-MUIRHEAD apparatus. 284 F. BRAUN, D.R.P. 136641 (1901). 285 F. KIEBITZ, ETZ, 33, 132, 1912. K. BANGERT, Phys. Zeitschr., 11, 123 et seq., 1909. 2850 A great number of methods for making use of the two coupling waves have been proposed. J. A. FLEMING (EL, 63, 333, 1909) e.g., proposes to use an ar- rangement like Fig. 391; but one portion shall be tuned to one wave, the other portion to the other wave. The currents of the two detectors act upon the same telephone, which has two windings for this purpose. Probably no such arrangement has ever been used in practice. ^H. RIEGGER, Jahrb., 5, 35, 1911. For the theory of three very loosely coupled circuits see B. MACKU, Jahrb., 4, 188, 1911. P. O. PEDERSEN, Jahrb., 3, 283, 1910; 4, 449, 1911. F. MULLER, Jahrb., 6, 13, 1912. BIBLIOGRAPHY AND NOTES ON THEORY 425 ETZ, 33, 376, 1912. 288 L. W. AUSTIN, Bull. Bur. Stands., 7, 301, 1911. 289 M. WIEN, Ann. Phys., 8, 696, 1902. Experiments of L. MANDELSTAM and H. BR ANDES at the Strassburg Forts in the summer of 1902. 290 R. FESSENDEN, El. Rev., 59, 77 et seq., 1906; Electrician, 62, 172, 1908; 65, 314 et seq., 1910. The use of a number of condenser circuits in the receiver for the purpose of increasing the sharpness of resonance has been proposed by both J. S. STONE and the MARCONI Co. (Electrician, 62, 171, 1908). 291 F. BRAUN, address at Strassburg, 1905. 292 COUNT VON ARCO, ETZ, 31, 506 et seq., 1910; Jahrb., 4, 79 et seq., 1910. 293 Regarding atmospheric disturbances see J. ERSKINE- MURRAY, Jahrb., 5, 108, 1911. P. SCHWARZHAUPT, EL, 65, 820, 1910. J. E. TAYLOR, EL, 66, 1022, 1911. W. H. ECCLES and H. M. AIREY, Proc. Roy. Soc., 85, 145, 1911. A. ESAU, Phys. Zeitschr., 12, 798, 1912. F. G. LORING, EL, 67, 27, 1911. M. DlECKMANN. 151 294 See J. ERSKiNE-MuRRAY, 1 EL, 68, 465, 1911. 295 See Jahrb., 4, 404, 1911. 2950 In the "reducteur d'interference" of the BALSILLIE System (Lum. el, 9, 404, 1910) a highly damped condenser circuit is coupled with the antenna. It is tuned to the frequency of the interfering station and is intended to absorb the oscillations caused by it in the receiver. L'Electricien, 41, 278, 1911. 2960 R. FESSENDEN, El. Rev., 59, 38, 1906. 297 A. BLONDEL, Compt. rend., 130, 1383, 1900. 2970 M. WIEN, Phys. Zeitschr., 13, 1034, 1912. 298 ANDERS BULL, Electrician, 54, 142, 1904. Regarding the HOVLAND apparatus see Jahrb., 5, 394, 1912. 299 E. NESPER, Jahrb., 4, 534 et seq., 1911. 300 Jahrb., 1, 430, 1908; 2, 419, 1909. 301 Regarding "ticker" and ticker connections see Jahrb., 1, 144, 1907. E. NESPER, Jahrb., 4, 317, 547, 1911. H. MOSLER, ETZ, 32, 1027, 1911. F. KIEBITZ, ETZ, 33, 132, 1912. 302 Jahrb., 5, 113, 1911. 303 L. W. AUSTIN, Phys. Zeitschr., 12, 867, 1912. 304 Electrician, 59, 985, 1907; El. Rev., 60, 251 et seq., 329, 368 et seq., 1907. Report of DE FOREST on tests with this detector in Electrician, 60, 135, 1907. 305 R. GOLDSCHMIDT, EL, 68, 464, 1911; Jahrb., 5, 341, 1911. 306 See E. BELLINI, Jahrb., 2, 381 et seq., 1909 and L. H. WALTER, EL, 64, 790 et seq., 1910. In these articles a large number of distance effect characteristics are calculated and plotted. 307 If r represents the distance between the points P and (Fig. 405), then, under the assumptions of Fig. 406, the field of antenna B at the point P is of the form v f 27rr Eo sin that of antenna A is of the form whence the resultant field is of the form 426 WIRELESS TELEGRAPHY It follows from this that the amplitude of the resultant field is