HIN .■HI IM HBHB HHT iUfUoUm LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 5IO. 84- UQ>r no.703 -?06 Cop. 2L Xi-7&5 uiucDCS-R-75-705 /7ia^C4 March, 1975 CAECOTRON : A TUBE FOR THE BLIND by Bernard Ka Pang Tse DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN URBANA, ILLINOIS Digitized by the Internet Archive in 2013 http://archive.org/details/caecotrontubefor705tseb UIUCDCS-R-75-705 CAECOTRON : A TUBE FOR THE BLIND Bernard Ka Pang Tse March, 1975 Department of Computer Science University of Illinois Urbana, Illinois 6l801 This work was supported in part "by the Regional Health Resource Center and was submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering at the Univer- sity of Illinois, Urbana, Illinois, March, 1975. CAECOTRON: A TUBE FOR THE BLIND Bernard Ka Pang Tse, Ph.D. Department of Electrical Engineering University of Illinois at Urbana-Champaign, 1975 ABSTRACT Caecotron is a project in visual prostheses that derives visual information from two television cameras and outputs the information to the user, namely a visually deprived person, in the form of a modulated tone sequence. The aural picture consists of a 16 X l6 matrix mapped onto the field of the television cameras and the audio system scans across this matrix picture in a way similar to that in which a television camera scans its field. Each of the l6 lines is represented in the aural picture by a distinct frequency and each of the l6 cells of each line is displayed serially: Frequency shifts correspond to brightness variations, while an amplitude modulation of the tone corresponds to the distance of a given element. Ill Ac kn owl edgements The author is deeply grateful to Professor ¥. J. Poppelbaum, Director of the Information Engineering Laboratory at the University of Illinois, who suggested the thesis topic and provided invaluable support and encouragement. Beginning from that February morning in 1970 when the author went to him with his many problems, Professor Poppelbaum' s generosity with his valuable time and his kind friendship have been very much appreciated. The author also wants to thank Professor W. J. Kubitz for his valuable suggestions and frank criticisms during the implementation of Caecotron. Messrs. Frank Serio, Bill Marlatt and Sam McDowell of the printed circuit laboratory were most helpful during the construction phase of the project. Ms Evelyn Huxhold typed portions of this thesis. The author likes to thank them for their very fine efforts. The author also likes to express his thanks to Grace who helped to make life a lot livelier. Finally, the author would like to thank his parents for their support, encouragement and, above all, their many sacrifices over the years, IV Table of Contents Page 1. Introduction 1 2. An Overview of Caecotron 3 3. The Caecotron Telemetry System 12 3.1 Caecotron Telemetry 12 3.2 The Left and the Right Matrix Pictures ik h. System Description of Caecotron 21 5. A Description of the Sub-system of Caecotron 2U 5.1 The Cameras and the Camera Controller 2h 5.2 High Frequency Clock Generator 26 5.3 The Input Video Signal Conditioning Circuit 29 ^.k The Line and Cell Integrators 32 5.5 The Controller 39 5.6 The Processor k^ 5-7 Output Variable Gain Amplifier 5^ 6. Alternate Approaches to the Implementation of Caecotron 56 7. Conclusion 6l Appendix 63 References 76 Vita 77 1. Introduction Caecotron — a tube for the blind — as proposed by Professor Poppelbaum is a project in visual prostheses that derives visual information from two television cameras and outputs this information to the user, namely a visually deprived person, in the form of a modulated tone sequence. Both brightness (black/white) and depth information are derived from the video signal of the two side-by-side television cameras and presented to the user. However, due to the limited bandwidth of the human ear as compared to that of the human eye, the amount of information that can be extracted from the video signals has to be condensed and drastically simplified. In the actual system, the 'picture' presented to the user consists of a l6 X 16 matrix mapped onto the field seen by the television cameras. The audio system scans across this matrix picture in a way similar to that in which a television camera scans across its field (although there is no interlace. ) Each of the 16 lines is represented in the aural picture by a distinct frequency and each of the 16 'cells' of a particular line is displayed serially: Frequency shifts correspond to brightness variations, while an amplitude modulation of the tone corresponds to the distance of a given element. In order to determine depth, an electronic telemetry system using two television cameras is used. These two cameras are mounted side-by-side and are situated at about eye level. By measuring the relative 'shift' of the signal of the left-hand camera as compared to the signal of the right-hand camera, depth information can be extracted. This depth information is used to amplitude modulate the output as described in the previous paragraph. Since all of the information available from the two television cameras is not utilized in the output to the user of Caecotron, various interesting 2 options can be added to the basic system: Instead of operating on an entire TV frame, attention can be focused on a smaller section of the picture. Hence it is possible to achieve a close-in view of the object of interest. This idea can be further extended to simulate the effects of variable focal length lenses on the cameras by restricting the system to different-sized sections of the television picture. Such a system has been implemented at the Information Engineering Laboratory of the University of Illinois to study the feasibility and useful- ness of such a machine. This thesis represents the system design of Caecotron; and, in the pages following, the functions and design of Caecotron as well as its sub-systems are described. 2. An Overview of Caecotron Caecotron is basically a fairly low resolution TV camera with an aural output. It has an electronic telemetry system that uses a simple algorithm to determine depth information about the objects that it is looking at. However, before going into the description of the design of the system, it is worthwhile to look, in some detail, into some of the things that Caecotron does. To begin with, Caecotron converts a television picture into a l6 X l6 sound picture. The latter is displayed to the user as a l6-step tone sequence , beginning at the top of the picture, represented by the highest note in the sequence, and ending at the bottom of the picture, represented by the lowest note of the sequence. Each note is sub-divided into l6 time slots, with each slot representing a 'cell' of the l6-cell matrix lines. In other words, the l6-step tone sequence displays the tele- vision picture serially from the top left to the bottom right just like a TV monitor displays the video picture, but, in this case, the display is aural instead of visual. The choice of tones in the 16 step tone sequence is important, however, since these tones should be easily recognizeable and, what is more, they should fall within a fairly narrow range of the audio spectrum so as to accommodate people with limited hearing capabilities. During the early phases of the project, experiments were done to determine the ability of the ear to distinguish tones and tone sequences. It was found that for a given percentage change in the frequency, the ear can more readily detect a change in frequency for a square wave than for a sine wave. This is not unreasonable since a change in the repetition rate of a square wave creates a change in each of its overtones. For this reason, square wave tone sequences were chosen for the aural pictures. An upper frequency of about 2112 Hz, a C, was chosen for the top line of the aural picture. 15 other tones, each one lower in frequency than the one preceeding and within a few percent of a corresponding note in the well-tempered scale, were chosen as the fundamental frequencies of the other 15 other aural lines. In this case, the lowest note has a frequency of U95 Hz a B. Such a tone sequence easily falls within the hearing range of most people; and if these tones are played in sequence, they can easily "be recognized by non-musical operators of the system. A table of these tones and their frequencies is shown in Figure 1. It was also mentioned in the Introduction that each of the 16 X 16 cells is presented to the user either as black or white in shade. A cell is considered to be white if the location in space corresponding to the part- icular cell is brighter than a pre-established average value of brightness as determined by the previous line. Otherwise the cell is considered to be black. If, for example, a black cell is located on a white line, then this balck cell is represented to the user as a slight elevation in the tone frequency in the time slot corresponding to that particular black cell. In other words, the 16 step tone sequence is frequency modulated to display the sensation of brightness ( i.e. the blackness or whiteness of particular cells. ) It was also discovered by the early experimentation that a change of 1/6U of the fundamental frequency ( less than a half tone ) can be noticed without much difficulty. Consequently, when the system encounters a black cell as it scans across a matrix line, the frequency of the aural line is Line number Tone Frequency 1 C" 2112 2 B' 1980 3 A' 1760 k G' 1581+ 5 F» 1U08 6 E' 1320 7 D« 1188 8 C 1056 9 B 990 10 A 880 11 G 792 12 F 10h 13 E 669 Ik D 59^ 15 C 528 16 B ^95 Figure 1. Caecotron Tone Frequencies increased by 1/6U of that of its fundamental frequency. An example of a display pattern to demonstrate the scheme described above is shown in Figure 2. The loudness with which the tone picture is displayed tells the user approximately how far the majority of the objects in the picture is away from the user. If the intensity of the aural picture is high, it means that there is the possibility of having some object very close to the user, i.e. the intensity of the aural picture alerts the user to a possible danger! On the other hand, if the aural picture is displayed with a relatively low intensity, it means that most of the objects in the field of view are probably fairly far away. In the Caecotron system, 5 loudness levels of the tone sequence are used to display the distance information. These loudness levels corres- pond approximately to objects situated at infinity, 13 feet, 7 feet, U.5 feet and for the highest intensity level, a distance of about 3.5 feet. ( It will be shown in a later section how these distances are arrived at and why they are used. ) It may be observed that the resolution of this system is biased in the 3 feet to 15 feet region; however, this cannot be considered to be a liability since this region is generally the area that we are most interested in, while in most circumstances, whether an object is 500 feet away from the observer or whether it is 600 feet away is of little importance. The Caecotron system is also capable of 'magnifying' certain areas of the displayed picture so that particular arears of interest can be given a more detailed 'look'. Caecotron is able to examine a field of view corresponding to the center l/k and l/l6 of the standard TV format in addition to the full television field. This gives the system extra versatility when the ordinary l6 X l6 resolution of the displayed picture does not give enough t t LU -p o •H P-i 03 CJ •H -P W O a < CM bO •H En CM CM to X K X O ~-i CM LU z LU Z z CM ro UJ 8 detail of specific areas to the user. A portion of the picture, namely the areas around the center of the field of view of the cameras, can be selectively- magnified to extract finer details. If a still closer look of the picture is required, then the center l/l6 can be magnified for scrutiny. Figures 3, h and 5 show the areas of a normal TV picture that is included in the display when magnifications of XI, X2 and X3 are used. The number of vertical lines indicated in the figures corresponds to the appropriate number of vertical lines that make up an individual matrix line; and, the amount of time desig- nated in the figures corresponds approximately to the portion of a horizontal TV line occupied by the appropriate cells. ~18 MS ~12 MS :Hi * m \ ~3.6m» ~5.' i XI 75/is — 1— Ims I ! i i [X2 i \ t""" i:::::: ya Aj > 15 VIDEO LI Figure 3. The Format of the Matrix Picture X2 10 ■*-*) 1.2 usees 9.2 s 1 ^ § § £ s ^ ^ s s & s ^ $ ^ yS S § Ks S & ^ V "v & £ § s \\ $ & ^J § ^ K ^ s^ S ^ O- N ^ > O ^ 5 £ ^ ? is 3 $ •H p4 u I bO •H id -P CD 1-1 CD H £> •H CO CO o OS CD 19 of the lenses used and the distance that they are placed apart. As an example, when two 25 mm ( or 1 inch ) lenses are used on two 1 inch vidicon cameras that are placed 8 inches apart, a shift of 1 cell corresponds to a distance of about 13 feet. Shifts of 2 and 3 cells correspond to distances of 7 and k.3 feet respectively, while a shift of k cells indicates that the object is only about 3.5 feet from the observer. Figure 10 shows the relationship between the amounts of shift of the left picture matrix and the object distance when the above cameras are used on the Caecotron telemetry system. 20 Object Distance Separation of the Two Cameras 7" 8" 9" 10" 11" 12" Amount of Shift I N F I N I T Y 12' 13' 15' IT' 18' 7' 8' 8.5' 9' U.5 1 5.5' 3.5' k> *+' fc.5 1 20' 10' Figure 10. Object Distance as a Function of Camera Separation and Amount of Picture Shift 21 h. System Description of Caecotron As described earlier, a two camera telemetry system is employed to determine the depth information. Two Cohu cameras with associated Camera Controllers are presently being used in the Caecotron system. The camera controller, as shown in the block diagram of Caecotron in Figure 11, supplies all the synchronization signals that the two cameras require, as well as the basic 31.5 kHz clock of the system. From the latter, clock signal of 13 X 31.5 kHz, 26 X 31.5 kHz and 53 X 31.5 kHz are derived using the Frequency Multiplier . These clock frequencies correspond respectively to the case when the system is in the XI magnification state, X2 magnification state and the X3 magnification state; and, depending on the magnification state that the machine is operating under, one of these clock signals is used as the system clock that control all the functions of Caecotron. The signals form the television cameras are amplified and conditioned by the Clipper and Gain Control which strips the standard video of all the synchronization signals, and, what is more, establishes a black level reference for the integrators inthe following section. The Line Integrators integrate the video signal of each of the horizontal lines; and, this signal serves as the standard brightness, or pre-established average brightness level for the Cell Integrators during the following line. This standard brightness reference derived from the previous line is used in Caecotron mainly because it tends to enhance contrast and, under certain circumstances, also tends to bring out extra features which may be lost when a fixed reference is used. This is very well exemplified by the case when there is a sharp difference in brightness level between the top 22 8 2 * 1 i-n 0> Q s « at o u o ■■ S) J I 8 "5 c e u T- i ii i I i o - t i. e 2L ° .9- c E 8 2 5 « o i 1 j bD cd •H Q .« o O H PQ £ O -P o o 0) a5 o H bO •H o u XT IT 23 half and the "bottom half of a picture. If a fixed reference level is used in this case, the matrix picture will probably show only the sharp difference in brightness level of the two halves, while the finer details within these regions will probably be lost. Using the average brightness level and cell integrals supplied by the line and cell integrators, the Digitizer and Vertical Cell Integrators reduce the video signals into the matrix pictures. The digitizer compares the cell integral with the reference brightness level and determines whether the particular cell is white or black. With this information, the vertical cell integrators sum up the white cells in each of the components of the l6 line matrices, and, if the number of white cells in an element exceed a certain level, then the particular element is considered to be white. With the left and the right matrix pictures provided by the system described so far, the Processor determines how far the majority of the objects in the field of view of the cameras is away from the user. The idea behind its operation is described in Chapter 3 and essentially involves the comparison of the left matrix picture with the right pictur matrix as well as shifted versions of the latter. All these operations are controlled by the block labelled ' Controller ' which contains the majority of the circuits needed for the control of the active RC integrators and vertical integrators, as well as for the generation of the timing and control signals necessary for the rangefinding process. The Output circuits contain primarily the tone generators that produce the l6-step tone sequence for the output of the aural picture, as well as the circuits that produce the slight elevations of these l6 frequencies for the indication of black cells within the aural lines. 2k 5. A Description of the Sub-system of Caecotron 3.1 The Cameras and the Camera Controller Caecotron uses two Cohu 3000 series environmental resistant television cameras that are mounted on top of a tripod dolly. These cameras are enclosed within two very rugged cylindrical enclosures, and, except for the optics, are sealed against moisture and dust. As mentioned in earlier chapters, the optics is provided by two 25mm fl.U lenses. These lenses are focused at about ten feet with the aperatures stopped down to provide two pictures that are fairly well focused from a few feet in front of the cameras to a point at infinity. The control signals and power required to drive the cameras are provided by a Cohu 3900 series camera controller which sits on a platform on top of the dolly. The controller synchronizes the two tele- vision cameras to the same clock signal and also provides the rest of the system with the 31.5 kHz basic clock as well as the horizontal and vertical synchronization signals. Higher frequency control signals necessary for the processing of the video signals are generated and synchronized to these signals from the camera controllers. A block diagram of this subsystem is shown in Figure 12. 25 O LU Q > o UJ O > et it O «/> V H O *H -P O O o3 CD I o O O z o u u z >- 2 < ec tu Z UJ o u ho •H Q *i o O H PQ < ec m < H 0) •H 7ZT ZT 26 5.2 High Frequency Clock Generator The reason why the 31.5 kHz clock from the camera controller is used as the basic system clock in Caecotron is that the operation of the system has to be in synchronization with the video signals coming from the two cameras. However, a clock of this frequency is not adequate because some of the operations of Caecotron have to be performed at a much higher rate. A good example of such operations is the integration of the video signals in the generation of the matrix pictures. A higher frequency clock signal is therefore necessary. A step variable high frequency clock is chosen for Caecotron. In the actual system, the period of the high frequency clock is dertermined by the duration of the matrix cell of the matrix picture; and, since the duration of the cell is dependent on the particular magnification that the system is operating under, this high frequency clock is necessarily depen- dent on the magnification too. In the XI case, for example, the clock period is chosen to be 1/26 th of that of the video line. 20 of the 26 equal time slots of the video line are therefore situated so that they mark off the cell periods, while the remaining 6 time slots are taken up by the horizontal synchronization signals. In the X2 case, however, the cells are only half as wide as those in the XI case and therefore, the clock frequency has to be corresponding- ly higher. The horizontal line is subdivided into 52 equal periods and again, 20 of them are used in the cell matrix. The same rule applies for the X3 case, where 20 of the 10^ time slots of the video line are used. This means that in the X3 magnification case, a high frequency clock signal 27 of 1.8U MHz is required. To generate these clocking signals, a phase-lock loop is utilized in the frequency multiplier mode ( i.e., with dividers in the feedback loop ) to obtain clock signals of ^09.5 kHz, 8l9 kHz and 1.81+ MHz respec- tively for the XI, X2 and X3 modes of magnification. A block diagram of this subsystem is shown in Figure 13. The small table to the bottom left of the figure shows the appropriate periods ( T ) of the clock signals. 28 u v C4 I in n < u O at Z o u M W 4) O o o wt *• CH >o 3 e« — o ^-* ^ c* CO u u u o -p a3 *H EJ •H w H O m i o z o CM CO •H En o — o o CN hi to the other inputs of the EXCLUSIVE-OR gates for comparison with the time shifted copies of the right matrix line. Each time a mismatch exists between two cells, the output of the corresponding EXCLUSIVE-OR gate goes to '1' state and such mismatches can be recorded by using counters at the outputs of the EXCLUSIVE-OR correlators. Care has to be taken, however, that only matrix cells that are meaningful are compared. In Figure 22, it can be seen that as the first cells of the two matrix lines exit from the shift registers, the n-1, n-2, n-3 and n-k outputs of the 5 bit shift registers do not contain meaningful information because data has not been shifted into them at that moment. The bottom EXCLUSIVE-OR gate in the column, however, has at its inputs the first cells of the left and right matrix lines and hence its output is valid. During the next clock period, the bottom two gates are valid; and, this continues until all the outputs from the EXCLUSIVE-OR comparators become meaningful. A mask therefore has to be designed so that only valid information is compared by the correlators, while others are masked from the comparison counters. Figure 23 shows the block diagram of such a mask together with its masking signals at the bottom of the figure. Each one of the control signals labelled 'a' through 'e* is used to gate the five EXCLUSIVE-OR outputs of Figure 22; and, since the system is designed so that the masks are 'off when the control signals are 'l f , the nth comparator of Figure 22, controlled by signal 'a' in Figure 23, has its correlation process initiated first, and it is also the comparator that has its operation terminated first at the end of the first l6 eel s of each line. A partial block diagram of the correlator system is shown in U8 21 pulses O -O e -O d -O c -O b -O a COMPARE ON "1" 16 20 1 1 1 1 1 1 1 1 MM 1 1 1 1 1 1 input 1 i 1 a 1 K 1 «" i 1 fi i 1 Figure 23. 5-bit Shift Registers used as a Mask U9 Figure 2k. It consists mainly of the circuits of Figure 22 and 23 combined and it also contains some of the timing signals required for the latching and comparison of the information signals. As mentioned earlier, the outputs of the EXCLUSIVE-OR gates are connected to 5 counters that count up the total number of mismatches for each of the segments of the left matrix picture. Again, as described in Chapter 3, the amount of shift of the right picture that gives a minimum amount of mismatched cells at the end of the correlation process gives us the depth information that we are seeking. The EXCLUSIVE-OR correlators of Figure 2k are connected to the five sets of 8 bit counters in Figure 25. Since the maximum number of dissimilarities between two l6 X l6 matrix pictures does not exceed 256, the 8 bit counters are sufficient under the worst case condition. After the l6 X l6 matrices are compared, the contents of these counters are counted down to zero simultaneouly by the clock; and, the counter that reaches a zero count first ( indicating a best fit for that particular picture shift ) sets a latch, and the count down process is stopped. The latches are then sampled by the output circuits to determine the amplitude of the output tone sequence. One word should be mentioned about how the right matrix picture ( l6 X l6 ) is stored so that it can be displayed to the user of the system. Since the elements of the l6 X 16 matrix are available serially from the integrators and they are also displayed serially to the user, a 256 bit shift register is used for this purpose. The right matrix picture is stored, cell by cell, into the shift register as it is available. During the playback, the data stored is shifted out serially by a much 50 9 9 9 9 Q u o -p ai H H £— t>- £—■ t»~ ck O latch Figure 25. Difference Counters 52 slower clock signal. A block diagram of this matrix picture memory is shown in Figure 26. 53 STORE/SHIFT O J, SYSTEM CLOCK O TO STORE SLOW CLOCK O TO PLAY 256 bit SHIFT REGISTERS ck Figure 26. Matrix Picture Memory 5U 5.7 Output Variable Gain Amplifier Since the output tone sequence displayed to the user can be in one of five loudness levels to indicate depth information, a variable gain amplifier, controlled by the output of the processor, is needed to modulate the l6-step tone sequence. A digital-to-analog converter is designed to serve as such a variable gain amplifier. The inputs of the circuit in Figure 27 is fed by the control signals from the circuit in Figure 25, together with the tone sequence produced by the tone generators. If, for example, ah object is at distance DO away from the user ( i.e. fairly far away ), DO is at a high level, while the other control signals are at a low level, and the digital-to-analog converter puts out a fairly low level tone sequence. On the other hand, if Dl is high, then the 2 input, as well as the 2 U input to the D/A converter will be switched low, and the tone sequence at the output of the system will be twice the ampli- tude of that in the DO case. Note moreover that D^ has priority over D3, which also has priority over D2 and so on. This means that if the system has both DU and Dl at a high level, then the louder tone signal corresponsing to DU is displayed to indicate that an object is at close range. 55 QA^D u •H M cd to u Figure 28. Block Diagram of a Possible Microprocessor Controller 59 microprocessor systems because of the relatively simple demands of the correlation process. Most of the outlay is in the microprocess and read- only-memory, which, at the present, cost about $200 in single quantities. It is especially attractive if the microprocessor is fast enough for the other control functions of Caecotron, since such a system would be essentially the same as the configuration as the block diagram of Figure 28, though a somewhat more elaborate control system is necessary to control the integrators. Recently, a few semiconductor manufacturers have announced bipolar versions of microprocessors which are 10 to 15 times faster than the corresponding MOS versions. The 3000 series of computing elements announced by Intel, for example, have an add time of 125 microseconds, and what is more, can be designed to have a l6 bit word length. Such a product is doubtlessly fast enough for application in the control of the cell integration process, and, because of the l6 bit word length, is ideal for the processing of the l6 cell columns of the matrix pictures. Such a system costs quite a bit more: At the time of this writing, a l6 bit word processor costs in the order of $700 in single quantities. Still this is a very attractive approach to the problem because of the virtues ( low power, compactness and programmability ) of the microprocessor. Yet another approach to the implementation is this: Instead of using two television cameras to acquire the matrix pictures, two photodiode arrays can be used. This dispenses of the use of cell and line integrators described previously to generate the matrix pictures, since integration is done directly on the photodiode arrays. Two analog delay lines in the form of charge-coupled devices ( CCD ) can be used to shift out the contents 6o of the phtodiode arrays and the analog ( ! ! ) contents of the photodiodes can be compared using analog comparators. These operations can, of course, be controlled by a microprocessor of the sort similar to the Intel 8008 since speed is not the problem. Such a system is almost ideal since the photodiode cameras are much more compact than the vidicon cameras, and, what is more, significantly simplify the operation of Caecotron. The disadvantage of such a system however, is that it is significantly more expensive: A 50 X 50 photodiode array together with the CCD, for example, costs in excess of $1500 each! It is, however, not inconceivable that they will come down in price as they become more widely used, and if that is the case, their use in a Caecotron-like system should be contemplated. 6i 7. Conclusion The system described in the preceeding pages has been imple- mented at the Information Engineering Laboratory of the University of Illinois to study the feasibility and usefulness of such a machine. Since the time that Caecotron has become operational, it has been demon- strated to both sighted and unsighted individuals and the results obtained has been very encouraging. The first introduction that a user obtains on the system is a short explanation of then encoding scheme used in the aural picture: Frequency shifts correspond to brightness variations, while the amplitude modulation of the tone corresponds to the distance of a particular element in the matrix picture. The user is then allowed to use Caecotron to identify tone patterns at a playback rate that is comfortable. After a few minutes' practice, the user can usually identify objects like tall fences, short fences, black vertical stripes on a light background, black boxes in mid-air and diagonal stripes. Again, after some practice, the playback rate can be increased significantly: Most users find the tone sequences intelligible at a playback rate of between 5 to 10 seconds for the l6-step tone pattern. The next logical step to improve the performance of Caecotron is obviously the miniaturization of the system by using smaller components since the machine, as it now stands, is not really portable. Some of the alternative components suggested in the preceeding chapter will help tremendously in this direction if they can be obtained for use in Caecotron: Miniature cameras using solid-state imaging devices will decrease the bulk of Caecotron by an order of magnitude, and will also decrease its weight by 62 much more than an order of magnitude since , as described in Chapter 6, the processing of the information will be much simplified. 63 Appendix The circuit diagrams included herein contain the working schematics of the main systems that make up Caecotron. The following circuits are included: 1. High Frequency Clock Generator 2. Integrator Controller 3. Cell-line Controller ^4. Cell Integrators 5. Line Integrators 6. Cell Integrators and Multiplexers 7. Video Amplifiers, Line Integrators and Sample-and-Hold Circuits 8. Processor #1 9. Processor #2 10. Tone Generator 11. Output Amplifier and Level Controller 12 Output Control Circuits 6U oo * 00 iH ¥ X X o o o Q > -i Hi' HH 6 > in 6 > + 6 Q g* — ^ O, CJ. X -WNr- O CJ X o CO 3-|h 10 ■I- <\J o 3 ^ fs. 00 J 1 O (VI N - nO o + I vw- ||l co — . IO o m uj * -J t«- uj CO ~ o sAsAsA ro o l-t x CM 6 X 6 in X CO O ^ CM co o -P aJ U CD a (iil K feS) 9 9 9 sg- ?i ^fiX&l ill ( s © *!. j3 Jj£a| -»— (si * O ®(" o -p PI o o o -p a3 *h bO 0) -P a CM 66 <3> £ CE> — ' r ^F^^g^ O < gpW»og ) 21 CCLL-LINC t»UM»C» «SET OCCuM »T END Of L*$T CELL LIME -o ClTst_linF> 1 * ^ END OF L»ST LINE Of CELL LINE BEGINNING OF LAST LINE OF CELL. LINE (t|) (5) (5) (tD *" U " E •"""" 10 IS 14 16 Figure A3. Cell-line Controller 67 DISCHARGE \ INTEGRATE VIDEO IN CD4016E — T 50pf 10K 10 K ► WV • VW- 005 CIRCUIT DIAGRAM PF INTEGRATER AND SWITCH SWITCH O VIDEO IN» ■o SUM SYMBOLIC DIAGRAM Figure kk. Cell Integrators 68 DISCHARGE / 15 o INTEGRATE VIOEO INO H5 CIRCUIT DIAGRAM OF INTEGRATOR (LINE) AND SWITCH SWITCH O- VIDEO INO OSUM SYMBOLIC DIAGRAM Figure A5. Line Integrators 69 +5V » t— H*% 06 » /u— "P^ 7 , LEFT VIDEO - IN 10 K 4 L + 5V A » /.— H£* +5V i 06 - /«o— h>£ 67 RIGHT VIDEO , IN 13 vi »CR 1 vi *CR /„ CD 4016AE 12 4 11 CD 4016AE LEFT -O INTEGRAL 60 OUT 15K CD 4016AE 4 11 CD 4016AE 10 RIGHT -O INTEGRAL 56 OUT 15K Figure A6. Cell Integrators and Multiplexers TO W -P •H O »-( •H O T) H o a i a) I 0) rH CO xJ G aJ en U o -p ^c hO 0) -P •H W M •rH (in 71 ® $x"\ HI— H- > xiil 06666 © @ I © 8 a m M in a k k *- f 9 ? X X ® ® ® © ® fHiAM 1$ A !r*6caeaiftca •E ^ S - ■*>■ -SJI fjj 8+ 9 «s Q. * 5 -»-W>. * o o • < H;s rr* 0- 'A ZE g m < v*. K^S ( I *P^9* I I Httf* I (3)0)0)© a g 72 CM O CO 10 0) Ih