MDDC 725 UNITED STATES ATOMIC ENERGY COMMISSION OAK RIDGE TENNESSEE THE USE OF HEARING-AID TYPE TUBES IN PORTABLE COUNTING- RATE METERS AND AMPLIFIERS by L. Nierman Published for use within the Atomic Energy Commission. Inquir- ies for additional copies and any questions regarding reproduction by recipients of this docximent may be referred to the Documents Distribution Subsection, Publication Section, Technical Information Branch, Atomic Energy Commission, P. O. Box E, Oak Ridge, Tennessee. Inasmuch as a declassified docxmient may differ materially from the original classified dociunent by reason of deletions necessary to accomplish declassification, this copy does not constitute au- thority for declassification of classified copies of a similar docu- ment which may bear the same title and authors. Document Declassified: 3/6/47 This document consists of 10 pages. OEPC Digitized by tine Internet Arcinive in 2011 with funding from University of Florida, George A. Smathers Libraries with support from LYRASIS and the Sloan Foundation http://www.archive.org/details/useofhearingaidtOOusat MDDC 725 -1- THE USE OF HEARING- AID TYPE TUBES IN PORTABLE COUNTING-RATE METERS AND AMPLIFIERS (Work done November 1944 through April 1945) L. Nierman The purpose of this report is to make available the general results of development work done on portable counting- rate meters used with proportional counters; such portable instruments have hitherto been com- paratively impractical because of the excessive weight of the electronic equipment. By adaptation of standard circuits to the low-power tubes used in hearing-aids and incorporation of components now available, a great reduction in size and weight has been achieved. This general re- port will be followed by detailed reports on specific instruments using the circuits and materials described here. A SIX- TUBE PROPORTIONAL AMPLIFIER AND COUNTING-RATE METER. Figure 1 is the schematic diagram of a six-tube model, consisting of a four-stage amplifier and a pulse-height selector. The amplifier consists of a pair of degenerative two- stage amplifiers. The overall gain of the amplifier, from grid of the input stage to grid of the pulse-height selector, is a little less than 200, In frequency response, the amplifier is flat to well beyond 200 kc, thus providing sufficient pulse discrimina- tion for the separation of an alpha pulse-height distribution from an in- tense beta pulse-height distribution. The negative feedback network con- sists of a voltage divider between the plate of the second tube and ground, impressing on the cathode of the first tube of each two-stage amplifier a portion of the output of that two-stage amplifier. The tubes used, Raytheon CK505 AX, are designed for operation with the filaments of two tubes in series across a l| volt cell. It will be noted that the circvut used requires three A batteries for the four-stage amplifier, the total A drain being 90 milliamperes at 1 .4 volts: one-third of the filament power delivered to the amplifier is wasted in resistors placed in series with filaments. A two-stage amplifier using 1L4 tubes gives substantially the same per- formance and requires only slightly more filament power, but | the B bat- tery current drain is considerably in excess of that required for the cir- cuit of Figure 1. In attempting to reduce the nvunber of filament batteries and to eliminate the power loss due to voltage-dropping resistors in series with the filaments, a number of other feedback circuits (wherein the fila- ments could be operated at the same A. C. potential) were tried, includ- ing common cathode resistors in the odd stages, plate-grid feedback over MDDC 725 three stages (instead of plate- cathode feedback over two stages as in Figure 1) and feedback to screen grids. None of these could be made to give the desired performance and at the same time be reasonably stable. It was likewise found impractical to replace the cathode resistors with chokes or to achieve the desired broadband amplification by video-type compensating coupling networks. The pulse height selector is an adaptation of the basic circuit de- veloped by Wendell Bradley for standard vacuum tubes and used in numer- ous applications as a pulse-former, pulse-height selector and counting- rate meter. The last tube is normally cut off. The circuit is so dcoigiied that any signal at the input to the pair which is less than a predeterminea level will produce no appreciable plate current in the last tube. However, a negative pulse at the input which is greater than the predetermined le 'el will produce in the last tube a rectangular current pulse whose amplitude and duration is unaffected by the shape or amplitude of the input pulse. Thus the D. C. component of the plate current in the second tube is pro- portional to the number of pulses above the desired height appearing at the input, and the number of these pulses is measured by placing a meter in the plate circuit, with an appropriate series resistor and shunting con- denser to give the desired speed of response to statistical fluctuations. The chief requirements in adapting this pulse-height selector cir- cuit to use with extremely low-power tubes are first, a sensitivity suffici- ent to make it operate properly with the limited amplifier gain which is available, and second, the production of current pulses of sufficient ampli- tude to drive available D'Arsonval D. C. meter movements. The portion of the random input pulses which will not be counted because of the resolving time of the circuit may be calculated. Let these losses, equal to unity minus the ratio of the counts registered to the actual number of input pulses, be represented by L. Let T represent the resolv- ing time of the circuit, i.e., the period during vhich the circuit has not sufficiently recovered from the effects of a previous pulse to respond to another pulse, and let t represent the average time between input pulses, t is equal to 1/R, where R is the average rate of occurrence of input pulses random in time, R being expressed in input pulses per unit time. Then L may be approximated closely at T/t, or RT. (See Strong, Procedures in Experimental Physics. 1938. p. 95). In the circuits discussed, i.e., pulse-height selecting counting-rate Meters, for any given size of input pulse, the resolving time is proportional to the length of the pulse formed, or T is equal to kd, where d is the duration of each output pulse and k is a constant depending on the tubes and parameters used in the circuit. Thus L = Rkd. But i = Rid, where i is the average current as read on a meter and I is the current which is drawn during the rectangular pulse. Thus MDDC 725 •3- L = ki/I, or the portion of counts lost is equal to the product of the ratio of resolving time to pulse length and the ratio of average (D.C) current to peak current. It may be seen that with standard radio type tubes, a fairly insensitive meter can be used without greatly impairing linearity of calibration. Let us apply the above analysis to the design of the present pulse- height selecting counting-rate meter. It must first be noted that k is not a constant independent of pulse height, but becomes smaller as the input pulse size is increased; in other words, when the circuit has been adjust- ed to respond to given amplitude of input pulse, and it is fed with pulses sli^tly above that amplitude, it will have a resolving time of kj times the duration of the output pulse: if the size of the pulses is increased the re- solving time will be k2 times the duration of the output pulse (which will be unaffected), k2 being smaller than kj. In the present case, the pvilse- height selector will respond to a pulse of about 0.3 volt amplitude. When pulses of this size are impressed, k is about 6. With pulses of ten times this amplitude, k is down to as low as 2, the circuit being "forced" to respond to new pulses although it has not completely recovered from the previous pulse. This effect may be observed with an oscilloscope at the grid of the last tube. Measurement of the resolving time for any input pulse amplitude is easily made by using a pulse generator input with phones on the output. As the frequency of the input is increased, the frequency of the output will be the same as the input frequency up to a point where it suddenly drops to one-half the pulse generator frequency; at a further point, the frequency of the output will drop to one-third the pulse generator frequency. The resolving rime for any given input pvilse amplitude is the period of the lowest input frequency at which the output frequency falls. With 1 elatively equal pulse amplitudes at the input, as with GM tubes, it is possible, by use of the above equation, to predict the linearity of this type of counting-rate meter with considerable accuracy. It is to be noted that if, as in most cases, k, the ratio of resolving time to pulse length, is a constant, the non-linearity will be the same for all ranges, where a range switch is provided. As an example, suppose a i»rticular circuit has a resolving time three times the pulse length for the pulse size to be count- ed and further suppose the peak current delivered during the output pulses is 1 milliampere. Then it appears that if a 50 microampere meter is used, 15% of the counts will be missed when the meter reads full scale. (It must, however , be noted that the above analysis cannot be used with ac- curacy for losses over 20% or so because of the approximation involved in the original derivation.) In the design of the present instrument, involving pulses varying in amplitude by a factor of 50 or more, the computation can not be made with MDDC 725^ -4- any exactness. Nevertheless, it is clear from the above discussion that in order to keep the non-inearity negligible with a 20 microampere move- ment, the most sensitive obtainable for such uses, the current during the pulses must be of the order of 1 milliampere. This consideration, together with the sensitivity requirement, led to the present design. To achieve high sensitivity without danger of oscillation, it is necessary to bias the last tube somewhat beyond cut-off and have high gain in the first tube of the pair. It will be observed that the latter is an ordinary degenerative amplifier till the input signal reaches such amplitude as to effect the "triggering" action. The latter occurs when the incremental current in the cathode resistor caused by current flow in the last tube is sufficient to overcome the decrement in current in the input tube of the pair caused by the negative signal pulse, thus making the total effect of the cathode resistor regenerative, rather than degenerative. High peak current is obtained by using a bias battery to cut off the last tube in the absence of signal, instead of relying wholly on the voltage drop which appears across the cathode resistor as a result of normal plate current in the input tube of the pair. This device allows^ the use of a much smaller cathode resistor than would otherwise be required, thus increasing the gain of the first tube and decreasing the load on the last tube. The counting-rate meter has no perceptible non-linearity over the full scale. A model of this instrument, after calibration with a pulse generator, was used with a boron- walled leutron counter at various distances from a radium-beryllium source. The counter was surrounded by a paraffin moderator. When readings (averaging by eye) of the counting-rate meter were compared with runs with a standard proportional amplifier and scaler, the results agreed within 3%. The model built is incorporated in a 6 inch cubical chassis, exclud- ing high voltage batteries for the counter and B batteries, both of which are intended to be carried in a knapsack. It weighs 62 pounds, including Ihe A batteries, which are in the chassis. A FOUR TUBE PROPORTIONAL AMPLIFIER AND CQTTNTING-RATE METER . It was found that the standards of accuracy required in the survey work for which these instruments are primarily designed are such as to make possible the use of a much lighter and simpler unit, requiring only one adjustment in addition to the range switch and the on-off switch. It is remarkably free of microphonic effects. A schematic diagram of this circuit IS shown in Figure 2. The amplifier is a two-stage resistance coupled nondegenerative MDDC 725^ -5- design. To discriminate against pickup and microphonics the grid and screen circuit time constants are such as to produce an amplifier with no flat region in its frequency response curve, which reaches a maximum in the vicinity of 12,000 cycles, where the gain is about 400, and is down to 0.707 of its peak value at 4500 cycles and 35,000 cycles. The pulse-height selector is essentially the same as that used in the instrument previously described except that the bias batteries are eliminated. The non-linearity thus introduced is, however, negligible in comparison with the error limits within which the instrument is designed to make measurements. This circuit, unlike the one previously described, is not intended to count all the pulses produced by the proportional counter. The plan for use of this instrument is to provide a convenient calibration sample to be slipped onto the counter at intervals of time (not yet determined) in the field, thus eliminating much of the error due to drifts in the batter- ies or circuit. In initial calibration of the instrument, a considerable percentage of counts will be deliberately lost. The calibration will be held constant by adjusting the single control until the standard sample gives the same reading as upon initial calibration. When this can no long- er be done without bringing in spurious counts which appear in the absence of radiation, the batteries must be replaced. It is expected that the bat- teries incorporated in the present design will last about two weeks with the instrument in use six to eight hours daily. The model currently under laboratory test is incorporated in an aluminum chassis 5" x 4^ x 3|", weighing less than three pounds, includ- ing all batteries except the high voltage for the counter, but excluding the meter, which is in a separate case 2 3/4" x 3|", connected to the circuit chassis by microphone cable. It is expected that this model will shortly be in the field, together with appropriate probes and light-weight high voltage batteries, used as a portable proportional alpha rate-meter. Later a poi^abie slow neutron rate- meter will be constructed. The present general report will be followed by subsequent reports on these instruments when all features of their design are complete. TUBES Before designing the circuits described above a survey was made of low filament- power tubes available, since most of the bulk of the port- able vacuum tube power supply is in the A batteries. A number of brands and types of hearing-aid tubes were considered and tested. Of these, those made by Raytheon are the most satisfactory. Zenith and Western Electric tubes, comparable in performance, are no longer manufactured, MDDC 725 -6- although samples were made available for testing. Hjrtron tubes require far more filament power than the others mentioned. It appears that a number of the hearing aids formerly made with other tubes have been redesigned for use with the Raytheon tubes. The electrometer tubes made by Victoreen also may be adapted advantageously to certain A. C. applica- tions. In the following summary of the tubes found most suitable, the fila- ment voltage is designated on the basis of 1.5 volt cell operation, but in all cases there is relatively little change in emission as the cell rims down to 1.2 volts. All tubes are filament type. Raytheon CK505AX Pentode. 30 mils, at 0.75 volts. Gm with 45 v. on plate and screen 175 micromhos. Maximima cur- rent as diode (suppressor internally connected to filament) approximately 1.5 mils. Raytheon CK509AX Triode. 30 mils, at 0.75 volts. Amplification factor 20. Plate resistance 150,000 ohms. Raytheon CK510AX Double tetrode 50 mils, at 0.75' volts. Designed for use as double triode with virtual cathode formed at first grid, which is common to both sections. Amplification factor of each second grid 20. High plate resistance, of the order of 1/2 megohm. Low plate currents. Raytheon CK512AX Pentode 20 mils, at 0.75 volt Gm with plate and screen at 45 volts. 120. Raytheon CK502AX - 503AX 5Q6AX 507 AX Power pentodes, 30 and 50 mils at 1.5 volts. Gm up to 500. A CK-507AX has been incorporated in a BG survey instrument model built by the Health Division to drive a small speaker in place of the usual headphones. Victoreen VE32 Electrometer triode. 10 mils, at 1.5 volts. Amplifi- cation factor 1.75. Maximum plate current as diode approximately 2 mils. Victoreen VE124 Electrometer tetrode. 10 mils, at 1.5 volts. When connected as high-mu triode, amplification factor ij 17, with sharp cut-off characteristic. Maximum plate current as diode approximately 2 mils. Be- cause of the relatively high peak current, a pair of triode- connected VE124 tubes make an excellent counting rate meter where high sensitivity is not MDDC 725^ -7- required, as in the Mark 1, Model 31A GM tube portable meter. All of the above tubes are made without bases, tinned leads being brought directly out of the envelope and soldered into the circuit. Some of the Raytheon tubes are available with bases under the same numbers without the designation AX. Tube- checking is done here by applying standard voltages to the electrodes and observing the plate current. A simple tube- checker in- corporating a 1 mil. meter has been constructed; a switch inserts I-2- volts bias on the control grid for a rough measurement of trans conductance. The tubes are inserted on a screw- type terminal board. COMPONENTS In mounting circuits in small spaces, it is necessary to utilize smaller components than those appearing in most general electronics stockrooms. In fixed resistors, 1/2 watt sizes such as those made by Allen-Bradley and Erie will ordinarily fulfill the requirements. (See Landsman's Reports CP2746 and CP2861 for voltage and temperature coefficients, life tests and other data on various brands of resistors). In variable resistors, the midget rheostats and potentiometers made by most manufacturers may be found sufficiently compact. But it is well to note that there are available smaller units made especially for hearing-aid and instrument use, such as the Centralab NS series and the Stackpole Type PSM. In choosing components, the greatest saving of space is to be ac- complished by selection of appropriate condensers. In capacities greater than 0.001 microfarad, the tubular paper condensers ordinarily used, rated at 400 volts, are extremely bulky. The midget paper condensers manu- factured by D\imont Electric Company as their number PIN, rated at 150 volts, occupy a small portion of the volume required for the usual paper condensers. They have proven particularly useful in 0.1 and 1 microfarad sizes. Small 150 volt bathtub condensers are made by Mallory. Although the usual small mica condensers will in most cases be adequate for capaci- ties less than 0.001 microfarad, the button- type silver- micas made by Centralab may be considered. Concerning batteries, it has heretofore been the custom to use flash- light cells for filament power, although hearing-aid type B batteries h/.ve commonly been employed. Flashlight cells, designed for high-drain service over short periods, will be found to give much less life per unit weight on low drains than the hearing-aid batteries which are designed to supply MDDC 725 -8- filament power. The plug-in feature oi many of the hearing-aid batteries also simplifies battery replacement. It should be pointed out that many of the coaxial cables in general use, such as RG26/U, are poorly suited for portable applications as probe cables because of their lack of flexibility. For low- current high voltage applications, it is possible to procure cable of smaller diameter, and thus much more flexible, without sacrificing low capacity. Thus RG6/U cable has been used at 2500 volts with a proportional counter without corona effect or other spurious counts. At lower voltages smaller di- ameter cables may be used. These include RG 58/U and RG62/U. ■9- MDDC 725 n ft^ V5 —AA/V^ % H ^^1 <^y^ r. .5^? > i^ or II '0 -10- MDDC 725 i iii<^ "f L„ crr-^ -i-jC I « 5^ ^ ^J > X Of % So. V -* J» i" '^ e S S > S ,o t ~ ^ '% ^ % ^ vj u 0*0 u Q V- "^ «5h ^>^ * a — ^ h < a V ^ 1 c 2 1 ; 1- v^> p^ S ^- > -V 5» ^ K • oi S! 5 3t ^5 5" t:' I- UNIVERSITY OF FLORIDA liiilllillliliiliiii 3 1262 08910 5422