L I B R.AFLY OF THE U N IVLRSITY Of ILLINOIS 510.84 IZQt no. 265-270 cop. 2 The person charging this material is re- sponsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for disciplinary action and may result in dismissal from the University. UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN MAY 1 1974 m 2 6 1974 WAY 311913 L161 — O-1096 Digitized by the Internet Archive in 2013 http://archive.org/details/optoelectronicss270koot Hbw Report No, 270 frv. Z70 jTULtft JUL 2 4 1968 OPTOELECTRONICS SWITCHING MATRIX by TUH KAI K00 May 28, 1968 Report No. 27O OPTOELECTRONICS SWITCHING MATRIX by TUH KAI K00 May 28, 1968 Department of Computer Science University of Illinois Urbana, Illinois 6l801 Ill ACKNOWLEDGMENT The author wishes to express his gratitude to his advisor, Dr. W. J. Poppelbaum, for his invaluable advice, illuminating suggestions and encouragement throughout the period of research for this dissertation, Thanks are also due to Dr. L. Van Biljon and Mr. D. P. Casasent for their comments and improving the manuscript. The author also appreciates to his wife, Ping, for her help on programming. Thanks are also extended to Mrs. Carpenter who typed the manuscript in its final form. IV TABLE OF CONTENTS Page ACKNOWLEDGMENT ...................... iii 1. INTRODUCTION. .................... 1 2 . GENERAL DESCRIPTION OF SYSTEM AND OPERATION ..... 3 2.1. Organization of The Opto-Electronic Switching Matrix System. ........... 3 2.2, Structure of The Photoconductive Cell Matrix Panel ................. 5 2.3« Principle of Operation ............ 5 3. TRANSIENT CHARACTERISTICS OF PHOTOCONDUCTOR ..... 13 3.1. Photoconductivity Phenomenon ......... 13 3.2. Solution to Single SRH Center Level. ..... 20 3«3' Transient Problem With Multi-level SRH Centers. ................. 26 k. ANALYSIS OF PHOTOCONDUCTIVE CELL MATRIX ....... 3^ 4.1. Equivalent Circuit of Pulsed Photoconductor ................ 3^ 4.2 o Transmission Efficiency, Injection Noise and Cross -talk 4 9 4.3. Important Factors in Matrix Design ...... 57 5. OPTICAL SOURCE AND SCANNING . 62 5.1. Light Sources. .... ....... 62 5.2. Light Deflection Methods ..... 70 5.3. Resolution of Light Deflectors ........ 80 5.4. Comparison of Light Deflecting Methods .... 83 5.5. Optical. Scanner. ............... 86 6. CONTROL SIGNAL PROCESSOR. .............. 95 6.1. Scanning Pattern and Driving Signals ..... 95 6.2. Driving Signal Generation. .... 97 6.3. Control Signal Memory and Regeneration .... 99 7- EXPERIMENTAL RESULTS. ..... ...... 104 8. SUMMARY ....................... Ill LIST OF REFERENCES .................... 112 VITA .................... 113 OPTO -ELECTRONIC SWITCHING MATRIX Tuh Kai Koo, Ph.D. Department of Electrical Engineering University of Illinois, 1968 A cross -"bar type switching matrix system with photo- conductive cells as basic switching elements and. an optical signal as the switching driving source has been studied and developed . The matrix itself has the form of a two-dimensional panel with a basic switching unit at each cross point in the matrix. The basic, switching unit contains three photo-conductive cells in T-connection, in which the center branch is grounded and the two arm branches are connected to input and output channels, respectively . Switching action is accomplished by scanning the matrix panel with a programmed helium-neon laser beam. The state of a switching unit depends upon whether the center branch or the arm branches are being excited by the light beam. The basic characteristics of the photo-conductive cells required for this application are a large ratio of off (dark) -resistance to on-resistance and large ratio of decay time to rise time. A better transmission efficiency and cross-talk factor will be obtained with a larger ratio of off to on resistance and the injection noise level as well as the transmission efficiency will improve with a large ratio of decay time to rise time. The equivalent circuit of a photo- conductive cell under the specific application has been developed . The complete equivalent circuit contains harmonic terms which are the consequence of optical beam scanning. These harmonic terms interact mutually with each other. First degree approximation has to be used to reduce the equivalent circuit to a practically usable form- All the theoretical evaluation of the system performance is based upon the approximated equivalent circuit . A small scale system which consists of a k by k matrix, an eiectro-mechanical light deflector with 20 kHz frequency response and a k mv helium-neon continuous laser has been constructed and tested. The performance of this system matches the theoretical expectation closely. Feasibility study shows that larger scale operation^ say 100 x 100, is possible when the photo -conductive cell matrix is constructed with the technique of integration yielding a matrix panel of several inches square and when an optical scanning system of higher resolution is used. 1 . INTRODUCTION An opto-electronic system which transfers information between given input and output channels is discussed in this thesis . The system basically contains a switching matrix panel, optical scanner, light source and control unit. The switching matrix panel is a cross-bar type matrix with photoconductive cells as the switching elements . Switching is initiated by a scanning light beam from the optical scanner which is controlled by the control unit,, The basic principle of operation is the asymmetrical relaxation process of longer decay time than rise time, which is observed in some photoconductive materials . This system was proposed by W, J. Poppelbaum some years ago. The transient characteristics of photoconductors are analyzed using the simplified physical model of SRH centers and it is shown that under certain conditions the deep traps and heavy trapping activity give rise to the longer decay time than rise time. The equivalent circuit of a single photoconductor with periodic excitation is then developed using the assumption that the relaxation processes of the photoconductors are an exponential function of time but with different time constants for rise and decay transients. This equivalent circuit can be represented by a constant conductance and either induced voltage sources or induced current sources. These induced sources have frequency spectrums similar to the Fourier spectrum of the fluctuating photo- conductance and amplitudes which are functions of the relaxation time constants of the photoconductor and the condition of excitation. A network analysis of the switching matrix of photoconductors is then presented. The characteristics of the system are described by defining transmission, efficiency, cross-talk factor and noise injecting 2 factor. Satisfactory performance is possible only with a high load impedance at the output channels and when the ratio of the decay time constant to the rise time constant of the phot oconduc tor is larger I the ratio of off -time to on-time of the stimulating light pulses, A prototype system with four input channels and four output channels has "been constructed. The optical scanner consists of two electro-mechanical deflecting mirrors with frequency response to 20 kHz. A novel alternate interlaced scanning method is used to eliminate the requirement of blanking the light beam during fly-fcack as in conventional raster scanning systems A dynamic memory unit, called Phastorj, can be used to store the information of input-output channel connection „ In the Phastor, the information is preserved in the form of time delay with respect tc standard timing, A direct excess to control the light beam can be obtained = It, therefore $ offers a more efficient means than other memory units 2 o GENERAL DESCRIPTION OF SYSTEM AND OPERATION 2 o 1 . Organization of The Opto-Elsctronic Switching Matrix Sys' The opto-electronic switching matrix system can be used to direct information from various input channels onto properly selected output channels. It can be divided into four major parts? (ij optical source, (2) optical beam scanner , (3) control signal processor a (h) switching matrix panel, as shown in the block diagram in Fig, 1. The optical source is required to provide an intense narrow light beam with small beam divergence . A continuous gas laser serves this purpose adequately., The function of the optical scanner is to modulate the propagating direction of the light beam from the light source according to the information received from the signal processor such that the trace of the beam spot on a plane perpendicular to the non-modula1 beam follows a pre -determined two-dimensional scanning pattern . The control signal processor is a unit which processes the informati concerning the switching between input and : bput channels on switching matrix panel, and it generates the proper signals to dri the optical scanner. The switching matrix panel is the most essent section of the system. It is an array of properly interconnected photoconductive cells, and it carries bcth input channels and output channels. Upon receiving the control signal carried by the light- team from the optical scanner, it responds by allowing information to be transmitted from the proper input channel to the selected output channel.. 3 ^ OUTP ' CHAN CO / tttii! _i . ui Z o z < 5- ITCHIN ATRIX 1- £ 2 ID CO Q. Z L B cu -P w CO •H -P (S d •H ,£5 o -p •H CO cc o CO o CO _J < DC UJ z 8 UJ o o cc H ^ Q. Q_ o 1- O CO o _l < z CD CO a o 5h •p o z> - 1 CO 0> •H 2.2. Structure of The Photoconductive Cell Matrix Panel The photoconductive cell matrix is in the form of a two-dimensional panel with an array of photoconductive cells „ Electrically, it is simply a cross -"bar type formation of input and output channels, as shown in Fig. 2. At each cross-point, there is a "basic photo -switching unit connected to the input and output channels as illustrated. The basic switching unit consists of three photoconductors in T-connection with one branch grounded , The branch connected to the input channel may be termed the "input branch' 1 , the one connected to the output channel the "output branch" and the grounded one the "center branch" , Physically, all photoconductive cells are located on the front surface of the matrix panel, and the relative location of the cells is such that the center branches are situated in rows which are separated from the rows of input and output branches, as illustrated by Fig. 3° The packing density of photocells is limited by the resolution of the optical scanner and the photo- conductive material. The relative location of the matrix panel, and the optical scanner is such that the panel is perpendicular to the undeflected light beam path. The distance between the panel and. the scanner depends upon, the dimensions of the matrix panel and amount of energy density fluctuations received by the photoconductive ceil surfaces between the edge un.it and center unit. 2.3. Principle of Operation. It is well known that the basic characteristic of photo - conductors is the change of its resistance when excited by light energy, In general practice, a phot oconduc tor can have a very high yet finite INPUT CHANNELS -A OUTPUT BRANCH INPUT BRANCH CENTER BRANCH BASIC SWITCHING UNIT Figure 2. Electrical Connection of Photoconductive Cell Matrix INPUT BRANCH OUTPUT BRANCH MATRIX PANEL INPUT 8 OUTPUT BRANCH ROWS GROUNDING BRANCH ROWS 7 resistance state when it is in dark, and can reach a low yet non-zero resistance state when it is excited by intense light. In comparison with a perfect switch, which is characterized by allowing total information transmission when closed and no information passage when open, a photoconductor acts as an imperfect switch where information may flow through it with attenuation caused by its non-zero resistance in the on state and a small part, though negligible, of information may leak through due to its finite high resistance in the off state. To maintain a photoconductor at the low resistance state, one can excite it either with continuous light or a train of light pulses. In the latter case, the resistance of the photoconductor fluctuates along an average value as illustrated in Fig. h. The level of the average resistance and the amplitude of fluctuation are determined by the transient characteristics of the photoconductor and the amplitude, pulse width and period of the light pulses. With a proper combination of these factors, a low average resistance with small fluctuations in amplitude can be obtained. The interaction between these factors will be detailed in Chapter 4. The switching action of a basic switch unit on the matrix pane] namely, the structure of the T-connected phot oconduc tors as shown in Fig. 5> can be phenomenolcgicaily described as follows % Assuming all photoconductors are identical, and the intensity of exciting light is high so that a large dark-to-on resistance ratio can be obtained, then, as far as terminals A and B are concerned, the information transmission will be negligibly small when the center branch is excited as shown in Fig. 5a, and information transmission occurs only when both input branch and output branch are excited simultaneously as in Fig. 5b. Therefore, U-UJ O (o Figure U. Steady State Photoconductance of Photoconductive Cell ■with Periodic Light Pulse Excitation 3 a. i- => O UJ z z < I o 13NNVH0 lOdNI VMOHJ UJ Q. UJ o o 13NNVH0 lfldNI IMOyd < h- (0 2 P p •H = c O £3 z M £ I ■H o & 1- p £ CO CO x> CJ •H W Ctf m

-P cti P CO bO C •H UJ O -p UJ M u I ► « 1 o H p c CD O CO o w CD W W CD o • o fin bO C •H P P •H ■a M a •H P 1— (/) o or 3 UJ Q O i- Z 2 or z O < I UJ O 00 (/) u UJ o H ^ < > OQ u •H P>4 16 into the conduction band as in process "b or it may capture a hole from the valence band as in process c. The capturing processes are characterized by capturing coefficients C,, C and the emission ir processes are determined by emission coefficients e , e . The capturing coefficients are proportional to the capturing cross sections cf the centers. The ratio of the C's and e's are related to the energy- level of the SRH center, E t , by e n = exp P - ji; C n L kT e exp [^ - E t" P G P kT L_ -i- (3 = 2) SRH centers may locate at any energy level in the forbidden band and may have different capturing coefficients which may not depend upon the energy level of the center. Although SRH centers may be located at the same level but they are spatially separated, so that the effective electron flow between centers may be ruled out. In addition to the recombination mechanism through SRH centers, ether recombination processes are also possible such as direct recombination or impact recombination. The direct recombination is characterized by the electron and the hole combining in a single step. In other words, the free electron "drops 11 back to the valence band right across the forbidden gap. However, In most phot oconduc tors, this process is not likely to be a dominant mechanism if it exists at all. From conservation of energy, the excess energy of a free electron must be given away when it is recombined. While no photon emission is observed for most phot oconduc tors and releasing energy in the form 17 of high energy phonons is also not likely, it is strongly suggested that the direct recombination across the gap is at least in a negligible amount . The impact recombination is a result of three-body inter- action. It involves either two electrons and one hole or two holes and one electron. When this process happens, two carriers recombine and the excess energy is transferred to the third carrier. However, this process may be observed only under proper conditions [3> ^] s such as high equilibrium carrier density. It, therefore, does not appear significant under normal conditions. Since SRH centers play a dominant role in both equilibrium and non-equilibrium conditions for most photoconductors, understanding their characteristics will, therefore, be most essential to understand photoconductors o From the four basic capturing and emitting processes it is obvious that, for an electron captured by a SRH center ; . its fate would either be to recombine by capturing a hole or to be thermally emitted back into the conduction band. The occurrence of these two possibilities has defined probabilities . One may, therefore, divide the SRH centers into four groups. Those centers, in which the pro- bability of a captured electron being recombined is higher than "Che probability of the captured electron being re -emitted back into conduction band, are termed electron recombination centers. The centers, in which the above ratio of probabilities is reversed, are termed electron traps = Similar arguments yield hole recombination centers and hole traps. For SRH center of the same capture cross-section, it apparently is the energy level that determines whether the center acts as a recombination center or a trap. The energy level, at which the 18 probabilities of the fates of the captured electrons are equal is, therefore, termed the demarcation level. The forbidden band can, consequently, be divided into three regions by the electron demarcation level and the hole demarcation level as shown in Fig. 8, The SRH centers located in region I and II act predominantly as electron and hole traps, respectively, and the SRH centers in region II act predominantly as recombination centers. As a reason- able approximation, one may consider that electrons captured by traps can only be thermally re-excited back to the conduction band and electrons or holes captured by recombination centers can only be recombined. The location of the demarcation levels are given by C p E, = F + kT In -£- dn n C n n (3-3) E, - F + kT In -2- dp p C n n where E, and E , are the demarcation levels for electrons and holes , dn dp respectively, and F and F are the quasi-Fermi levels for electrons J n p ^ and holes, respectively. The above equations show that the demarcation levels are not only determined by the capture cross-sections of the SRH centers, but are also closely related to the carrier densities. Therefore, one can see that as the densities of carriers increase, the electron demarcation level and hole demarcation level are moving towards the conduction band and the valence band, respectively. This fact has an important effect on photoconductivities when SRH centers are distributed in the band gap or concentrated near the equilibrium 19 < o an \- o UJ _l UJ u LlI c UJ Z 2 t-co < Ul o X oz OGQ z o o zo £ dt p^ t p^l^ t t' {3>*J dAP t — ■ C n n( V n t ) - C^ - C pP n t + Cy^-nJ 21 where f is the generation rate, n is the electron density in SRH centers , An and Ap are the excess electron and hole densities and n and p are defined as n n = N exp t c (3.5) 1 c * kT IE, - E : N exp |— — rr - 1 v kT with N and N being the densities of the effective states of the c v conduction hand and the valence hand, respectively. The system equations, Eq, (.3°^)? are completed with the space charge neutrality condition that An. + An = Ap . (3° 6) The solution to Eq. (3°*0 and Eq= (3-6) with the proper initial and boundary conditions should give a complete picture of the kinetic characteristics of this single level system. Unfortunately, the mathematical difficulties with higher order non-linear differential equation prevent a solution in the genera.i case. A treatment for the small signal case has been presented thoroughly by Fan [6]. First, one considers the case An ~ Ap This is a good approximation for low SRH center density cr after the initial stage of a large excitation. Then Eq= (3-^) and Eq« (3°6) can be reduced to 22 dAa = _ 1_ l+JoAn . dt T - L + aAn where C + C n p n + P. and 1_ Vp^vV T o "' C n (n i + V + c P U i + V in which p n and n are equilibrium electron and hole densities. Equation (3° 7) is valid for both rise and decay transients with appropriate initial conditions. For the rise transient the generation rate is greater than zero and the initial conditions is An = when t = 0. For the decay transient, the generation rate is equal to zero and the initial condition is An - An , when t = where st An is the steady state value before decay transient starts . S"C The general solutions to Eq= (3° 7) are then (i) The rise transient; bAn 2 + (fT a-l) An + f T I J |^=i I = C ± exp(- f) (3-8) where a oo Ob ; - [(iT a-.l) 2 + 4bf T J £ 7= 2b p = (fT Q a-l) + E fT Q a-l)' tei, 2b 23 and ¥ ,2 1 5 - «a + ? + ^ r=T C ;I = + fT Q (&)5 (ii) for the decay transient, one obtains: (i - =9 (bAn+.l j (An) = q exp (3-9) where <*W \ ll QO ( bAn , + l) st d - t 2 ) The solutions of Eq. (3-8) and (3 = 9) have such a complex form that. real physical meaning can hardly be deduced from them. To gain a clearer understanding of the transient behavior, one may investigate a special case where C ~ or C » C , and where the location of n p n p the SRH center is such that n ~ n . Considering that the excitation is larger so that only a short time after the excitation An » n Q , it arrives at the conditions aAn » 1 and bAn » 1 2k and then Eq. (3«7) has the form of classical recombination theory [7] dAn = f-i-An (3.10) dt " t where _1_ . b _1 T oo a " T it is apparent that both rise and decay transitions are characterized by same time constant, t and rise time equals decay time. However, if 00 the SRH center level is moved closer to the conduction band so that n becomes much larger than An, while all others remain unchanged, this becomes another extreme case bAn » 1 and aAn « 1 . It yields the form of classical quadratic recombination theory [7] = f An-— An . (3.11) l l The solution for the rise transient of Eq. (3.11) is An = -= & (3.22) with — + r Coth(rt/2) 1 25 And the solution for the decay transient is then similarly An = An exp[- -H st T Q 1 + An . b[l-exp(- — )] St T Q (3-13) From the above analysis, it is seen that the relaxation processes of photoconductivity are closely controlled by the location of the SRH centers. If the centers are located near the dark Fermi-level, they are predominantly recombination centers and the transient response to square wave stimulation has same time constant for both rise and decay. However, if the level of the SRH centers is moving toward the conduction band, the characteristics of the centers shift toward more trapping action and thus increase the lifetime of free electrons and prolongs the decay transient, and thus the relaxation for rise and decay becomes more and more asymmetry . If one defines the rise time t as the time required for An to reach ~— of its final steady state r 1-e value and the decay time t as the time necessary for An to decay to - of its initial value then from Eg,. (3.12) and Eq. (3° 13) one finds 1 2, 1 >2- _d _ l/^ol 2 ^.[e+efl + (l+^fl) J - in[l+2 fl + (l+lj-fl) ] L CL -, -1 J- Coth — [e + (1+hilf 2"] > (3.H0 with ft = fb-r, f[C n ( V n c ) + C p (p 1+ n )] t N, C C (n_+p.)' t n p v N ffC n-+C P, ) • n 1 p-l y t n r> 0' e - 2.713 26 Equation (3.1k) is plotted in Fig. 9 which shows that the ratio of decay time to rise time increases as the trapping action of the SRH center increases. The previous example is only for the case that the density of SRH center is relatively low. For the opposite case where the density of SRH center is relatively large, a mathematical form of transient response can hardly be reached. However , one can expect similar conclusion as the low density case, that heavier the trapping activity of the centers, the larger the ratio of decay time to rise time. 3 . 3 . Transient Problem^ W ith Multi-level SRH Centers In the previous section, a model of a single level SRH center has been considered under some restrictions. Unfortunately, few practical cases can be represented by this simple model, especially for the insulating type photoconductors . In the practical case, one may expect that there are SRH centers of different types located at different energy levels. An exact analysis to any such given case is mathematically impossible. To understand the mechanism of such a complex system, a phenomen-logical argument is needed to deduce an approximation for semi -qualitative analysis [8]. A relatively simple model, of distributed SRH centers of one kind is discussed. The energy diagram for this case is shown in Fig. 10. Assume that there is a high concentration of SRH centers N near the dark Fermi -level ,. These centers act as recombination centers . There are also uniformly distributed centers of lower 27 St o CO o o .. o OO'OS 00 "Oh 00 '0E (— 00*02 H CD O 2 rH CD s> CD 1-3 G •H CO o ■H EH CD W •H K o -p •H EH >> n5 o CD Q O o •H o\ •H 00 "01 00' ■o > ^ o LU > CD -P CD O T3 (U -P -P w •H P H O Ch •H O CD o \- o if O CD I LU o z CD o H 0) H •H En 29 density between the dark Fermi -level and the conduction band. Let the density of these distributed centers be N per kT. At any free carrier density level, there corresponds a demarcation level E., dn such that those centers above E, will act mostly as traps and those below E n will, mostly be recombination centers . Due to the fact dn that N is a large quantity, the shifting of E does not effectively alter the recombination center density. Also, as a result of Boltzman's distribution law, the density of electrons captured by the traps is always N during the steady state and is not affected by the location of the demarcation level. Then,, at steady state, the lifetime t of the electrons can be determined from the trapped electron density and the free electron density as: n t T = (1 + — ) T n. (3*15) where i is the lifetime of electrons when there are no traps but recombination centers, t~ is determined by N in this case. One r can see from Eq. (3*15) that the lifetime is inversely proportional, to the excitation if the excitation is low (N » n) and approaches the constant value T n as the excitation intensity is increased. The above argument on the lifetime t is valid in steady state. Equation (3. 15) may not hold during the transient process, however, t is valid as an instantaneous lifetime under certain- condition [9]» If this is the case, the rate of change of conduction electrons may be written as dAn f An dt t. , inst f -^T^7 •- • The solution to above equation can easily be found to be 30 (3.16) (i) for the rise trans ie nt An. - T Q f An - t q fN t An-q! An-p] ' t \ C exp^- — j T (3-17) with ^fr n • (fx n ) 2 + 4fT Q N t ] 2 V '0' = i^T 0+ [(fT ) 2 + ^f VtH • I i [(fT Q ) + ^Tq^] C .. (f N t ) 2 |a| The rise time constant can then be found as (1+A) 2 | 2 - 2(1+A)| 1 i4-N, 2 fT (1-A)(|-1) (I^TFTi^A) (3.18) 31 in which A = 1 5=1 "e (ii) For the decay transient, where f=0, and An^An , when t=0 An An exp N t An. .(1 2.) An v An ' s = exp(- — ) l (3-19) The above equation clearly shows a long decay tail due to the presence of the exponential term on the right side of the equation, The effectiveness of this prolonged decay depends upon the initial excess electron density and N . If An is in the neighborhood of "0 s or less than N, , a very long decay can be expected. Assuming that An is produced by a generation rate f , then ^s - 2 f T i l and the time for — - to decay to - — — of its initial value will be An J 1-e s u 2N 1 + li^ll f t q 1+A (3-20) 32 The ratio of the decay time to the rise time can then be plotted from Eq. (3J-8) and Eq. (3-20 ) as shown in Fig. 11. This plot reveals that a heavy trap density yields a long decay time. This is due to the fact that, during the decay process, there are almost a constant number of electrons freed from traps which are turning into trap action from recombination while the demarcation level moves toward equilibrium levels. The analysis of the two given simple models merely serves the purpose of demonstrating the effect of trapping on the transient responses of photoconductivity. They show that both heavy trapping activity and the existence of deep traps prolong the delay time. Phenomologically , one sees that if large quantity of electrons are trapped ;, then during the decay process, these electrons are released into conduction band by thermal process and. the decrease of conduction electron density is slowed down by the arrival of electron from traps. Also, if the electrons are trapped in deep traps, where the exchange rate between conduction band is slow, the latter part of the decay transient is, therefore, dominated by this slow process. In the practical case, few insulator photoconductors have the simple energy band structure as given. The SRH centers may have different capture cross -sections and their density distributions may be non-uniform. A more complicated transient curve can be expected. But one can expect that the dominating fact by SRH center has the similar trend. 33 f? z -- 4- o H SL'Z OS'Z 0) S •H fn Eh CD P CD £ CO CD ■H O w -P CO (L) 15 S 0> •H -P EH 3 >5 -H c6 U V -P 01 P •H >, P rH K rH o •rH c U P ^ -H •H fa se-c! oo-z SL'l OS' I S2"l 3^ 4. ANALYSIS OF PHOTOCONDUCTIVE CELL MATRIX k . 1 . Equi valent Circuit, of Pulsed Photoconductor For phot oconduc tors., which are under the excitation of periodic light pulses, their instantaneous conductances depends upon the transient characteristics of the photoconductor as well as the condition of excitation. As noted in the previous section, the transient response of a photoconductor cannot be expressed by any simple mathematical function of time, even for a simple model. Therefore, an exact analysis of the behavior of the photoconductor matrix under the given stimulation is beyond our scope due to mathematical difficulties and complexity. However, a simple analytical result with rough assumptions will serve as a guide for the evaluation of system design and performance. Assuming that the relaxation process of the excess charge carriers in the photoconductor follows an exponential function of time characterized by two time constants, the rise (or growth) time constant, i and decay time constant t.,, then the instant- r d' aneous conductance G of a photoconductor can be expressed as G = G + m s [1 - exp( - ;£-)] (k.l) 'r for rise transients, and G G_ + £G exp (- — ) ( (4.2) S T, d 35 for decay transients, where G„ is the dark conductance of the photoconductor and AG is the steady state non-equilibrium con- s ductance corresponding to the given intensity of the light. If the photoconductor is stimulated by periodic pulses, as illustrated in Fig. 12, then in steady state, the instantaneous conductance can be written as G=G^+Z^. + (£G - £G . )[1 - expf- — ) ] (4.3) mm x s min /L v T v r for < t < T and t-T G - G n + ZX> exp[ - (■ -)] (4. k) max T, d for T < t < T where £G and ZG . are the maximum and minimum points of the wax min fluctuating instantaneous conductance. Since G is continuous at t = T and t-T, one finds 36 O (o It T , sp_j~i n n r 1 1 1 II II II m Figure 12. Steady State Photoconductance of Photoconductive Cell with Periodic Light Pulses Excitation 37 T l Sinh' -U- -^.^L, "Wn ' ^s " \X T ^ 6XP( " ^ Sinh — i — + — 2 Vr T d, The peak-to-peak amplitude of fluctuation, AG „, is, therefore, AG = AG - AG . f max mm Sinh I =■ . — Sinh 0<7) 2 t / \2t, » r/ \ d sinh §k + v Since the conductance G is periodic, it may, therefore, be expanded in a Fourier series such that G = G + G (4.8) av ac where G is the average conductance given by av T + 2 G = ^ /' G(t)dt (l+„9) av T 't 2 and G is the combination of all harmonic terms given by ac 38 G = Z ac n=l ?mr 2nn . a Cos -7=- t + b Sin — 7- t n T n T (1+. 10) where the a 's and b 's are the Fourier coefficients n n + = - / G(t) Cos -~- t dt T 2 n T (4.11) 2 b n = |/ G(t)SinfL t dt >.12) Equation (4.10) may also be expressed as G = Z C Cos ac . n n=l T n (4.13a) with C = fa + b ) ■ n n ' (4.13b) and i b -- tan"~ — n b a (4.14) 39 Assuming that the excitation condition satisfies AG . » G„ , mm the Fourier components are then given by and AG 'av " T T T 2 1 Sinh t: — Sinh — 2t, t \ + 2(t,-t ) '1 ' d r' Sinhi(^ + ^-) 2 T T, r d (4.15) ey, T T T = 2AG . °^P. Sin e, exp( — ). Sin(nit+e, )+Sin(nit— ^ — --0, ) mm d | t j d T d exf T T -T 2fAG -AG . ) —Sine |exp(-~ )sin(n«-e )-Sin(mr 2 m 1 s mm T r|^T r T AC T -T s a . / 2 1 N — - Sin (nit — - — ) nit T . (4.16) b = 2AG . -^ Sine, n mm T d T T -I exp( — )cosfnir+e, )-Cos(nit— — e _, ) T, k d T d' d r „. + 2 (AG -AG . ) ~ Sin Q s mm' T r m T -T 1 2 1 exp(- — )Cos(nit-e r )+Cos(nit T ' -Q^) r AG nit T -T 2 1 Cos nit - Cos (nit ■ — — -) (4.17) with m "I T Q, = Tan d 2nn t, d ■>.18) -1 T 9 = Tan r 2nn t r Tf a voltage V is applied across the terminals of the phot oconduc tor which is excited "by periodic light pulses as shown in Fig. 13a, the instantaneous voltage -current relationship is then I = V(G + G ) ^ 19a ) av ac using the instantaneous conductance expression Eq . (4.8). Equation (4.19a) may be rewritten as G I = V(l + ^--) G . (4.19b' G av - av This equation shows that the photoconductor may be considered to possess only a constant conductance G . The fluctuating part, G , may be considered as a series "induced" internal voltage source ac and the magnitude of its voltage is G -S£ v G av Ui LU (/> _l 3 a. x H < CD g i- < ■ LU t or 5E o UJ I 3 o Ixl UJ u or 8 O > o •H P C5 P •H O T3 O •H P O -P o g C o o o p o > o CD -I Z o z LU (/> LU or Q. £ or o LU o or 3 O if) CO LU -J i 3 H O Z LU LU or or 3 O O w p •H o •H o p a H a > •H en ■H k2 By defining the inducing factor [i, that av and equivalent circuit for the photoconductor under periodic excitation may, therefore, be represented by an induced voltage source in series with a constant conductance as shown in Fig. 13b. Equation (4.19) 'may also be considered from the viewpoint of current flow. Since VG is the current generated by a voltage V across a 3. v conductance G , another equivalent circuit can be formed with the av parallel combination of an "induced'' current scarce and a constant conductance as shown in Fig. 1.3c where G i = VG and u = — - ■ av G aA^ The inducing factor u. actually consists of the Fourier components normalized with respect to the average conductance G , and is a av periodic function of time, and, therefore, the induced sources merely represent noises injected, by the periodic stimulation. Since p.. is the combination of all the harmonic terms in the Fourier expansion, it may also be separated according to +he frequencies of the Fourier components. The induced sources may thus be represented by either series :ombination of induced voltage sources of single frequencies T n . , m> as shown in Fig, 14a, or by a parallel combination of induced U3 <$ ® ® ® iilGov M4' ^3' M2 1 Ml 1 (a) CURRENT SOURCE REPRESENTATION 5>nV (b) VOLTAGE SOURCE REPRESENTATION Figure lU. Equivalent Circuit Representation kk current sources of single frequencies =■, as illustrated in Fig. 14b. The inducing factor for each single source is given by C n U- = ~ CO£ n G av g 2nr^ (4.21) In the practical application of the opto-electronic switching system, one requires that, with a periodic excitation, the photo- conductor yields a high average conductance and small amplitudes of inducing factors. To show the relations between the equivalent circuit quantities , the photoconductor parameters and excitation condition, Eq. (4.15) and Eq. (4.13b) are plotted in Fig, 15 through Fig. 17. Figure 15a to Fig. 15c show the normalized average conductance T d T 2 as a function of — with — as parameters . The normalization is with r 1 respect to the steady state non-equilibrium conductance zX> corresponding to same light intensity as the amplitude, of the light pulses. The four figures are for four different light pulse widths, expressed by T l — . One notices that the average conductance decreases as the pulse T r T 2 width increases if — is unchanged. This is due to the fact that one 1 allows longer T for the photoconductor to decay, but the decreasing of the average conductance is not so drastic that it would, be a condition for concern However, for the amplitudes of the inducing factors, which are shown in Fig. 16 and. Fig. 17 for first and second harmonics, one no ; Lc several important facts, (l) The inducing factors increase *•? U6 15.00 60.X 75.00 90.00 ^'T (a) -i — i — i i -> — i — i 00 15.00 30.00 MS.00 60.00 75.00 90.00 105.00 120.00 (b) ^'T 90.00 105.00 120.00 (c) Figure 16. Inducing Factor of First Harmonic Frequency 1+7 -t — i — i — i — i — i — i — ►- H 1 1 1 1 1 1 .00 15.00 30.00 HS.OO 60.00 75.00 90.00 105.00 120.00 (a) Mi- 1. ■05 io s . T, 30 ■ 20 Ti Ti ■10 ID-*- -• 10 ■ Ti • 1 io""- ,o". 1 1 1 I 1 1— — 1 1 1- — 1 1 1 1 1— — 1 1 .00 15.00 90.00 1(5.00 60.00 75.00 90.00 105.00 120.00 (b) I I I 1 1 I t 1 1 1 1 1 1 I .00 15.00 30.00 15.00 60.00 75.00 90.00 105.00 120.00 (c) Figure 17. Inducing Factor of Second Harmonic Frequency T l rapidly when the pulse width increases (for — unchanged). For T T 2 1 1 example from — = .1 to — = .1.0, the first harmonic inducing factor T r Tr +2 increases by a factor of nearly 10 , (2) For a given pulse width, the harmonic inducing factor increases as the pulse period increases, but the rate of growth decreases as period becomes large, (3) The inducing factor decreases rapidly as its order increases. For both the inducing factors and average conductance; it is found that they will always have a satisfactory value as long as T t T 2 \ Additionally, one may neglect the higher order harmonics when considering the inducing factor. Furthermore, although the term ''current source" and "voltage source" are used in the equivalent circuit presentations, they do not have the conventional meaning. It is clear that the sources corresponding to different harmonic terms interact with each other since they are directly associated with the current flow in G J av (in case of current source representation) or voltage across the whole element (in case of voltage source representation). These induced sources are not effective unless external forcing is applied, therefore, for further analysis, Thevenin's theory is not valid. However, in the practical case, |_i * s are designed to be small and the regenerative effect may be negligible.- If this is the case, then the conventional theory of voltage sources or current sources may be applied and should yield an approximate analytical result. ^9 k .2° Transmission Efficiency, Injection Noise and Gross- talk Under the operating scheme described in Chapter 1, the characteristics for a cross-bar type matrix with T-connected photoconductor units at each cross-point, as shown in Fig. 2, can be analyzed with the equivalent circuit developed in the previous section. It is understood that, during the system operation, it is either the center branch or both the input and output branches of the switching unit that are being stimulated by periodic light pulses. Since the light pulses to a given photoconductor result from scanning, the photoconductor instead of receiving a light pulse which fully illuminates its sensitive area at once sees a light spot moving across its sensitive area. Therefore, before and after the photo- conductor is fully illuminated, there is a period of time that the photoconductor is only partially illuminated. The effect of partial illumination not only complicates the treatment, but also introduces additional noise into the system if the period of partial illumination is comparable to that of full illumination. However, if the sensitive area of the photoconductor is small in comparison with the light spot size or if a step scanning system is used to increase the ratio of full illumination period to partial Illumination period, one may neglect the partial illumination effect. It is assumed that the scanning light beam has such an intensity and scanning rate that the inducing factor p. of each photoconductor is small and the ratio of dark resistance to the average resistance R is very large. The average resistance is defined, as the reciprocal of the average conductance G . Under the restriction that each output av 50 channel may not receive information from more than one input channel but one input channel may feed more than one output channel , the interaction among output channels can be neglected . Considering any output channel Y., which is receiving information from input channel X , one can draw an equivalent circuit m for the situation as in Fig- 18, in which u(n) denotes the inducing factor for the photoconductor of the switching unit associated with the nth input channel. Since p's have been assumed small, super- position is approximately valid and one can consider the effect of information voltage V 's and noise sources separately. From the equivalent circuit in Fig. 19* which considers V 's only, the voltage- current relations are given by V" - V . ' V s V . - V ' m i m oj m R R , R av d av V - V ' V 8 V.-V* r / m r r r oj r R^ R R., r = 1, . . . m-1, m+1 . „.n d av d V ' - V . n V ' - V av r=l d /m r^m (4.22) 51 Vm> • Vi> • V 2 > V 3 > v r > jVWin Figure 18. Approximated Equivalent Circuit at an Output Channel 52 A -• w •H O o T3 ■a. 'H^ » ||l = >> H ■P ctf C C •H o CM U w 4 "O < (T < > o ~^< > c "O < a: < < c IX o •H ra w •H e w § rl EH U O Ch w •h -P w PS H o cr3 •H < O -P a) rH > •H ON rH •H Pn 53 Equation (4.22) leads to the voltage V . at Y . as oj j n-1 d. 1 av n-1 + or with oj \R T R R , R 2R^+R R^ P^+2R ° \L av d av d av d d av. R, V R 1 n d * + J?L^L^ £ V (4.23) R 2R,+R R^ 2R +R^ . r av d av d av d r=l ^ r^ra n V . = TV + p E V (4.24) r=l r = A R (2R +R ) 1 _1_ n-1 _d_ 1 av n-1 aV aV R T + R + R, "R °2P,+F. "R, - R^+?R av a av d av d d av (4.25) R V /2-R^ + R In the practical case,, the condition R, » R d av .s always satisfied and yields ^ 2R X (4.27) ! + _±I + 2(n-l) av (W p*2 s^l r The current flowing through the branches of R can then te found cL V ,'R, + P R, 1 ( d av d „ ml P V 2P + R i 2R + R oj av V d av d av _A_(____^ rv + v.) - -\ m2 R \2R J + R ^ oj V oj av V d av (k.29) V . + V r = 1, r R , + 2R ' d av . , m-1, m+1, . . . n For the case R , » R , Eq. ( J 4.29; "becomes d av i -, = i ml ' m2 2R av n (i - r) v - p Z v ^ ' m r r=l ~~ r^m — (4.30) i = ~ 1 rv - v r R . V m r d n P Z V r~l r/-m 55 When the currents are known- the effect of the "induced sources" can then he evaluated with the help of the equivalent network of Fig. 2C. The system equation V . V V " sj m m P~ " FT " R~ av d av V sj r r u(r) i = ' 4.3i; R, R R^ r d av d r =■ 1, ... m-1, m+1, ... n r fc m n V - v " V V - " v s j r sj si ra \ , _ r-1 d '- av r^m leads to the solution for the voltage V . appearing at the output channel Y. due to the induced sources. Since the inducing factors J M-(r) are out of phase with each other due to the fact that all the photocells are excited, sequentially, a complex form for V ., which s j can hardly be interpreted clearly, will be obtained if the phase shift is considered. However, one may create a worst case' by assuming that all |i ' s are in phase. This cannot happen to the first harmonic term, which is the most important one, with the given scanning scheme, but it is possible for higher order ones. The worst case analysis gives a possible upper limit of noise. By assuming that the |i(r)'s have the same amplitudes and. are in phase., the maximum noise at the output for the case R n » R becomes d av 56 n v„< = n< (i - r) rv m - pr(n-i; v^ + (1- p Z v sj m • m r or by const ip-ri n g t~ne practical case condition 6 « 1 > one has V . = n(l - r) T • V (^-32b) sj m Therefore, the total signal V appearing at the output is the superposition of V„ . and V Oj sj n v - rv + p Z v + J, - r) rv ^3; Oj m p r= , r • • m JJ/ r /'n A] examination of the terms on the right side of the above equation eals that the first term, TV , carries the information. V to be m m transmitted despite being modulated by a scaling factor I\ Therefore, may be defined as the 'Transmission Efficiency and it has a numerical value between and 1. The second term on the right clearly carries the information from input channels other than the mth one ; this inJ rmal ion is not supposed to be received by the output as it re}.' the i ross-talk from other channels. The constant 3 is, 57 therefore, defined as the "cross-talk factor". From Eq< (4,28), it is clear that the cross-talk may be suppressed to a negligible level if a large ratio of R, to R can be provided The Last term d av V . is the noise caused by the periodic stimulation. It may be s J termed "injection Noise" in the ser.se + hat it is injected by the scanning beam with a noise injection factor n. = (i(l - V) F, This noise contains a spectrum of u.. By examining the three quantities; F given by Eq. (4.27)> p given by Eq , (4.28) and t] = n(l - r) r, one finds that the matrix functions better when the load impede nee P on the output channel increases. Since the value R L of — — is usually very small, on the order of .10"'' or less, the R d R 3 V transmission efficiency F becomes highly affected bv - — , The larger the R T , the closer the F to unity. Improving the transmission efficiency F may also increase the cross-talk factor P, but it /R V 2 / av * becomes less important due to the fact (3 is governed by[ r — \ F d which is one the order of 10" or less. However, the injecting noise can be heavily suppressed when F approaches unity, The relation P R between ~- and ~- with V and n as parameters is plotted in Fig, 2.1 P- ^j L d to show their interactions. 4.3. I mportant Factors in Matrix Design From the above analysis, the basic characteristis of the switching matrix can be described by three quantities, namely, the transmission efficiency F, cross-talk factor P and noise injection factor T]« An ideal system should, have unit transmission efficiency, zero cross-talk factor and zero noise injection factor. By examining 58 o cr ■o -■ tr IfllOrfOl S m cj> 0> (7) CO CD OD d 6 6 6 d d II II II II II II C_UE-.UL.J-.-- i\\\\\ o -I o CM — o o r4 59 Eq. (4.27), Eq. (4.28) and Eq. (4.31), it becomes clear that a R d R L system requires large - — , - — and small u. av av It is not unusual to find photoconductors which can furnish h a ratio of 10 or higher for dark resistance R, to full-on resistance R (= 777^) with only moderate light intensity. From Fig, 18, it is s obvious that as long as the scanning satisfies r 1 the average conductance G satisfies av G > k £& av 2 s R d 3 Therefore, the quantity - — still is of the order of .10' or higher. R av To maintain this large ratio requires complete dark on the photo- conductor when it is not. scanned. Since the photo-resistance is very sensitive in low light intensity region, as shown in Fig. 22 for a typical photoconductor, any background light will reduce the R d effective - — drastically and degrade the performance of the system. R av For example, if the photoconductor has a resistance - light intensity relation as in Fig. 22 and the light beam has an intensity of 2 P d 22.9 mw/cm , with periodic excitation, R =10Ko This gives a - — ■ ' ' ' av R av 6o Hi o z S * ID 2 O -P O C o o o -p o o 6 8 d 2 H0/SllVM|-niN 6'22 01 103dS3W HUM Q3ZnVNM0N ' Ai.ISN31.NI 1H9H CVI CM f l ' 63 e CD -P w >> CO bo •H -P B •H H H O o 0) PQ •P ■a •H on CM CD S, •H 1 z H oo: uj o -Jflc Jc o 5 o2 the output beam diameter I , beam divergence angle 9 and energy density e of the collimated beam are given by 6k or and where 9 - d Z 3 \ i 9 = Tan (5-1) (5.2) l h £q Tan e = _ Ai __ (5-3) E = 2+jtE .i - (5-M f r f 2 d^d., c diameter of radiating source aperture of condensing lens effective focal lengths of condensing and collimating lenses, respectively distances, as illustrated in Fig, 25 radiating energy per ster radian from the radiating source . 65 A well collimated beam for moderate specifications can be obtained by optimizing Eq. (5-1) "to Eq. (5»^)» However, if the specifications of the required beam are extreme, for example very small £ and 9 and very large E, an optimum condition may not exist at all. From the above equations, it is seen that lenses of small f -number are required to produce a small, intense light beam. On the practical technology side, this imposes an additional limitation on the avail- ability of lenses with near ideal quality. Therefore, obtaining an intense, well collimated small light beam from a conventional light source is difficult and its broad spectrum also imposes difficulties in effectively controlling it (such as deflecting or modulating) without introducing color dispersion. The alternative is a continuous laser [10]. A laser beam is characterized by highly monochromatic and coherent light. The laser beam can be in different modes depending upon the resonant cavity structure. For the fundamental mode, which is the most important and most used, the laser beam is approximately defined by a Gaussian wave [11]. If the beam is propagating in the z -direct ion, it may be described by a wave function in cylindrical coordinates U = $(r,(j),z) • exp(-jkz) (5-5) where j = \l - 1 66 which satisfies the wave equation - 2 2 v u = k u = o (5-6) 2tt where k = - — is the wave propagation constant . This wave function A. has the approximate solution { U = ^exp4-j(kz- 4- x 72 z 2 c i? UJ UJ cr o c 30NV1SIQ "IVIlVdS O -P o ID H «H M •H 8.3 The elctro-mechanical deflector has the highest resolution potential due to its large deflection angle. But its scanning rate is the lowest of all. For the higher scanning rate units, the deflection angle is limited by inertia or other mechanical restrictions, thus the resolution decreases as scanning rate increases. Acousto- optical deflectors can reach a resolution comparable with TV scanning, their main restrictions are the frequency response of the ultrasonic transducer and the necessity of rotating the acoustic wavefront to keep track with the Bragg angles. Electro-optical devices have the highest scanning rate but their resolution is limited by the voltage gradients and crystal size that can be used. The range of resolution and scanning rate for the different deflectors is indicated graphically in Fig. 29. This indicates that, for all types of deflectors, the resolution and scanning rate are not independently controllable. One may have a wide range to choose deflector according to the requirement. 5.k. Comparison of Light Deflecting Methods Among the light deflection methods described in previous sections, none can be considered adequate for general practical use. Although each method offers some advantages over the others, they all have some serious disadvantages which limit their application. Bli < O CO OPTI CTOR i Ul O _i h- U. co cr O l- o _i < o CO -ACOUS DE _l 1- cr u. 0_ o UJ o \- Q 1 o o UJ h- cr _l UJ CD o Ul u. Ul Q Z _J UJ cr Ll UJ cr CD O n < -J 2 cr ui < o > CO J_ J- 0> o 00 o o CO o m O Q _l io o O CO '-* Ul a: C\J o ui -2L CO H Z Ul UJ _l UJ w fn o -p V 0> H 00 N CO IO «■ fO CM O Q O O O O O O 33S/S3NI1 * 31VU 9NINNV0S 85 The electro-mechanical scheme has a unique advantage in that it does not introduce color dispersion and its deflection is insensitive to the light frequency. It may also have a wider scanning angle and, therefore, higher resolution. But its frequency response is limited to a relatively low range due to the mechanical inertia present. The rotating mirror offers only one type of scanning -- namely, saw-tooth. Each trace cannot be triggered separately and, therefore, synchronization to an external time base requires good phase and frequency stability of both the deflector and the external time base. The vibrating mirror suffers from thermal effect since the expansion of the suspension system displaces the beam in the direction perpendicular to the scanning,, This becomes more severe for higher frequency units in which the suspension is stiffer, also, in high frequency units, the mirror is immersed in a damping fliud to provide critical damping operation which unavoidably creates color dispersion in a broad spectrum beam. Both the electro-optic and the birefringent methods are different applications of the electro-optical properties of some materials. The sensitive nature of the materials to the light frequency makes both deflectors usable for monochromatic light . Since they require a high electric field to produce the electro-optical effect, a high voltage control source is necessary and yet the result- is a relatively small deflection, approximately 1 . However, these two methods do have attractive aspects as high speed deflectors. Their frequency response is limited only by the thermal dissipation 36 due to dielectric losses and the driving electronics. A frequency 9 response up to 10 Hz can "be expected. Deflectors employing b.irefringent methods are the only ones that offer random access- ability without the need for an additional light shutter to blank out the light during the beam position transition. The acous to -optical deflection system has a frequency response into the MHz range. It is also rather insensitive to thermal effects but requires a critical alignment of the optical beam and the acoustic wavefront to maintain the beam intensity. To compensate for any deviations from the required alignment, an array of cells may be used to generate a rotating acoustic wavefront to match the angle. This method also su f fers from color dispersion. When the light power handling capacity is considered, the electro-mechanical types cf deflector emerges in a leading role among all deflectors. The main factor which limits the power capacity is the thermal effect which introduces a non-uniformity in either the electro- optical crystals or the transducers = The .^oss of light power through the deflector is not. a. sufficient reason to reject any of the methods becfe high optical quality components are available to insure high y in each case . 5° 5° leal Scanner The basic requirements for the light beam, scanning system in pto-electronic switching matrix are determined by several factors: (l) The beam diameter is specified by m>< criterion that at the edge 87 -3 of the beam the energy density is 10 of its value at the maximum point (usually the center of the beam). This definition is different from the conventional definition of ~ attenuation. The reason for e^ such a rigid criterion is that the operation of the system requires 3 the light on-off states having a contrast of at least 10 . A Gaussian beam with radius a for —r- attenuation, has a beam radius of 1.8.5a e in the new criterion. This also implies that the distance between photoconductors on the matrix panel cannot be less than the radius of the light beam. On the other hand, the size of the sensitive area of the photoconductor should be kept smaller than the -r attenuation e dimension in order to have the unit as uniformly illuminated as possible. (2) The system must provide a resolution of n-element per line where n is the number of photoconductive elements per line. (3) The system must provide a two-dimensional scanning pattern. (k) The periodic light pulses experienced by the photoconductors must satisfy Eq» (4.33) from which the requirement of scanning rate is determined. (5) The background light caused by scattering inside -3 the beam scanner should be less than 10 of the maximum bean intensity. Any higher background light intensity will degrade the matrix per- formance . The optical scanner used in the prototype system was constructed with two vibrating mirrors. The vibrating mirrors have a frequency response of 20 KHz with 5 deflect i.on angle and an aperture of approximately 1 mm by 1 mm. Two dimensional scanning is accomplished 88 by the optical arrangement shown in Figo 30. Both 1*41 and M2 are fixed first surface mirrors insuring proper incident and exit beam directions, GM1 and GM2 are the vibrating mirrors whose axes of vibration are at right angles to each other, and lens L serves the prupose of collimating and guiding the light beam. GM1 and GM2 are located in the conjugate planes of the lens L so that the beam deflected by GM1 is guided onto GM2 to be deflected again. The detailed optical path in the scanner is shown in Fig. 31' The laser beam with 0.7mm diameter and i.^ millradian divergence angle is converged by a lens with focal length of 203-2 mm. The optical distance between the converging lens and GMl is such that the focused light spot falls on the focal plane of the collimating lens whose focal length is 25^ mm. Therefore, the light spot on GMl is approximately 0.^ mm in diameter, just small enough to avoid diffraction by GMl, and the focused light spot in the focal plane is approximately 0*3 mm in diameter . The center of the light beam passes the conjugate point on the right side of the collimating lens and it is independent of the deflection angel — of GMl, Furthermore, the light beam image A or B in the focal plane is equivalent to a light source and radiates a cone of light rays which .is coliimated into a parallel beam by the lens, The coliimated beam has a diameter of approximately 0.4 mm and. a divergence angle of nearly 6 mi.liirad.ian due to the finite spot size at A or B. The direction of the coliimated beam d Ls on the location of the image in the focal plane, and in this case it has the same magnitude as the deflection angle 5- of GMl. 89 CVJ CD a o CO H c3 o •H -P ft O <+H o -p H n3 o •H P C5 o CD g> •H CO 90 CD O CO o -p P-' O -P o •H -p p< o H en bD •H 91 All of the center lines of the deflected beam from GM1 pass through the conjugate point 0. The vibrating mirror GM2 is located at this point to pick up the beam and deflect it in a direction per- pendicular to that of GM1, thus providing a two-dimensional scanning system. The use of a converging beam and a collimating lens reduces the diffraction caused by the small mirror aperture and removes the limitation on the deflection angle of GM1. Two -dimensional deflection could be obtained by having two mirrors face to face, but either the second mirror must have a larger aperture to accommodate the beam deflected by the first mirror or the deflection of the first mirror must be limited so that the beam will not sweep out of the aperture of the second mirror. The two lenses used in this canning system reduces the beam size, but increases the beam divergence by a factor of the ratio of their focal lengths which is 8 in this case. The resolution of the scanner will also be reduced by the same factor. The fact that the locus of the focused beam spot A is an arc instead of a line when GM1 vibrates, as shown in Fig, 32, introduces an additional divergence into the exit beam. To the first order approximation, this divergence is given by 5 = wl which results in a beam divergence of approximately 0,1 > mill! radian in the worst case. Therefore, the exit beam of the scanner will have 92 to Ixl 0) C 5 3 o TO O •H -P ft o c •H o -p •H > CD n OJ CO 0) M 93 a diameter of 0.4- mm with a divergence angle of 6.5 milliradian in the worst case. A resolution of 130 lines in either scanning direction may be expected if the remainder of the optical system is ideal . Unfortunately, the resolution of the system is less than expected due to the optical condition of the vibrating mirror structure. A cross-sectional view of the vibrating mirror is shown in Fig. 33 > the mirror is immersed in a damping fluid and sealed in a tube. The light beam exists through a window which is tilted to separate the first reflected beam as shown in Fig. 33a. However, successive reflections between the interfaces of the window as indicated in Fig. 33b, interferes with the main beam and results in interference fringing in the far-field pattern. The intensity of the fringing depends upon the reflective it ies at the interfaces. Unless the window is optically flat, the scattered light will also result in a serious background illumination. 9h i— DAMPING FLUID CYLINDER WALL INCIDENT BEAM AND EXIT BEAM MIRROR SUSPENSION CROSS -SECTIONAL SIDE VIEW CROSS -SECTIONAL TOP VIEW Figure 33. Cross-sectional Views of Vibrating Mirror 95 6. CONTROL SIGNAL PROCESSOR 6.1. Scanning Pattern and Driving S ign als To scan the switching matrix, several alternate sequences may he considered. In conventional raster scanning, in which the beam starts at one edge of the panel and retreats back to the same edge to start the second row, there is both horizontal and vertical fly-back. During the fly -back, the beam must be blanked so that it will not interfere with the switching units it passes over. This requires additional components such as a beam shutter. To avoid this, one may use an alternate interlacing scanning pattern in which there is no fly-back or retrace as illustrated in Fig= 3^a. Such a scanning pattern requires a more complicated driving signal than a simple saw-tooth wave for raster scanning and an. even number of horizontal lines to produce continuous scanning with no fly-back. Since the scanning beam spot is to be discrete in order to reduce the transition time between units and to provide a large ratio of full -illumination time to partial illumination time for the scanned photoconductor, a step-shaped driving signal is necessary for both horizontal and vertical deflections. To illustrate the sequential relation between driving signals, one may consider the scanning of a 4x4 matrix without loss of generality. The ideal signal for both the horizontal and vertical driving signals for a linear scanner are shown in Pig- 3^ . The driving signals are shown for the condition of no information transmission being allowed between input and output GATE n ANALOG ADDER CLOCK OUT FOR VERTICAL DRIVING (0) D* I / CLOCK OUT i m_ JLtL U (d) u .5. Block Diagram of Horizontal-driving Signal Generator and Wave Forms at Different Staees 99 the number of columns to be scanned, on the matrix panel, relative frequency drift between these signals cannot be tolerated. It would be very difficult to stabilize the beam position if free running oscillators were used. The method of generating vertical driving signals is similar to that for the horizontal driving signals and is shown in Fig. 36. The only difference is that the output D of the control un.it is a wide pulse which shifts one side of the staircase wave to provide an asymmetry staircase (.Fig. 3^c) for the interlacing scheme. The clock is fed by the control unit of the horizontal, driving circuit and its frequency is r— -- of that for horizontal driving. 6.3. Control Signal Memory and Regene ration The driving signals generated, by the above scheme provide the basic scanning, which maintains all the switch units in an off state To turn on a selected, switch unit, the beam, scanning has to be displaced in each scanning cycle until that switch unit is to be off. The information on selection may come on. a separate control channel or as part of the information on the input channels and its form may be in any code. It is reasonable to assume that it can be broken down to two parts which designate the input channel and the output channel to be connected. "Phastor" [19]j offers an elegant and. economical method, for storing this information and for regenerating the corresponding distrubing signal, every scanning cycle. C LOCK FROM HORIZONTAL DRIVING TRIANGLUAR WAVE GEN. 1 SAWTOOTH WAVE GEN. I 1 SAWTOOTH WAVE GEN.H CONTROL GATE I GATEH ANALOG ADDER ♦ t C* D* -►t Figure 36. Block Diagram of Vertical-driving Signal Generator and Wave Forms at Different Stages 101 Phastor is a high speed high precision analog memory unit in which the information is stored in the form of phase shift with respect to a standard reference. It is basically a regenerative constant delay loop. If the information is in analog form, the operation of phastor can he described with the aid of Fig* 37- The information voltage (Fig. 37d) to be stored is compared with a standard ramp voltage (Figo 37c ) to generate a single narrow pulse (Fig. 37e ) whose phase with respect to the standard reference pulses (Fig. 37b) is proportional to the amplitude of the information. This single pulse circulates in the active delay loop. The delay time of the loop is clocked so that its output pulses (Fig. 37-t ) has same period as that of the standard clock. Therefore, the output is a pulse train with same frequency as the reference pulses but with a phase shift between them, A phastor with an operating frequency in MHz region and ±00 memory levels has been reported. For n x m matrix channels, 2N phastors are used. For each input channel two phastors are used, to register the input and the output channel between which the information is to be transmitted. Those are designated as the X-phastor and the Y-phastor, respectively.. The X-phastor has N memory levels while the Y-phastor is divided into M levels. The X-phastor is controlled by the clock used in the horizontal driving and the Y-phastor by the clock used in the vertical driving. The output pulse widths of both phastors are set to equal the period of their respective clocks. Therefore, by sending the output of both phastors through an AND-gate , a single pulse with width 102 INFORMATION INPUT c COMPARATOR D ACTIVE E DELAY LOOP i u a « »B oA A* RAMP REFERENCE (a) ERASE CLOCK Figure 37. Phastor Block Diagram and Wave Forms 103 corresponding to the matrix cross-point is obtained. This pulse, if its amplitude is standardized, may be added directly to the vertical driving signal to produce displacement at proper location. Since geometrical position on the matrix corresponds to the time delay in the driving signal, and since the phastor registers information in the form of time delay, it offers a directly usable access signal thus eliminating the process of trans- formation required by other memory methods. 104 7. EXPERIMENTAL RESULTS To prove the feasibility of the suggested switching system, a prototype system with four input channels and four output channels has been constructed and tested. The light source used in the system is a continuous helium-neon gas laser with wavelength of 632.8 nm at TEM m mode. The output beam of the laser is measured at 4.5 milliwatts with the -2 beam diameter at 0,7 mm for e points at the exit aperture and the full divergence angle of 1.6 milliradians . The optical scanner is constructed as described in section 5«3« The phot oconduc tors used on the matrix panel are commercial grade EM1508 manufactured by Raytheon, They are caldium selenide cells primarily for control circuit use. The dark resistance of the 9 / cells has typical value of 10 ft in low voltage (around 50 volts or less) operation. All cells are selected so that their resistance and transient response are within 25$ of uniformity. The photo- resistance of the cell with response to the laser beam is measured at the room temperature and one typical case is shown in Fig. 38. Nature density filter and beam expansion are used to provide low energy density illumination. The transient time constants, t-,, t are d r measured with light pulses provided by rotating chopping wheel. The t and t is determined by comparison method. The circuitry for the measurement is simple as shown in Fig. 39- For such circuit the output 105 -i o.oi - 0.001 10 6 RESISTANCE fl Figure 38. Photo -response of a Photoconductor Id 60 50 - 40 - 30 - 20 - 10 - 1 RISE TIME CONSTANT T r : O DECAY TIME CONSTANT T d : ? : ' /x - x/ z 8 111 v> j -1 i 1 %/ — "0 ■ 2.0 - 1.5 ^*r / X / < 1.0 A. 0.1 .III! 1 1 ■ 1 1 - » 0.2 0.4 0.6 0.8 1.00 NORMALIZED LIGHT INTENSITY Pi mifo QQ t3iao o-n^ TIopqv Timo frvn c-han +■ :s nf a Phr>-hr>r>rmr1'nr > "hr)"r 106 voltage V n is R l v ___ v R(t) + R 1 in which R(t) is the instantaneous resistance of the photo- s' conductor, one set the value of R n being —r of the final steady 7 1 e-1 state value R(°°) of the photoconductor for rise transient time constant and set R "being eR.(°°) for decay time constant measure. With proper value R, one obtains V = — V v 2 v when t = t in rise transient and t = T n in decay. By this relation, r d both t and t, are determined on the oscilloscope with constant r d voltage supply. The measured data of a EMI 5 02 is given in Fig. 39- The scale of light intensity is normalized with respect to 22, 95 milliwatts/cm . The scanning pattern and driving signal wave forms are as that suggested in Chapter 5- But the circuits to generate the driving signals did not follow the recommended method because the total units to be scanned is only l6 that no economic factor is to be considered. To generate the horizontal scanning signal, a ring counter is used to register the corresponding horizontal beam position, as shown in the upper part of Fig. kO, outputs of each flip-flop of the counter processed by a group of NAND gates so that the output of the gate 107 E CD -P CO >> CO -p o -p o Ph o I u M •H O o o H o 0) •H Pn ±08 has pulse sequence as shown in Fig. kl. The output pulses of the four NAND gates activate four diamond gates which allow their analog input voltages transmitted to the output terminals. The analog voltages input to the diamond gates have precise step difference from one to another. Therefore, at the output joint of the four diamond gates, a staircase wave is obtained by their proper turn-on sequence. To drive the light beam off the area of photo elements during the vertical transition, an "edge-detector" is designed to generate pulses with correct polarities when it receives signal from the counter. The outputs from diamond gates and edge detector are added through a wide band operational amplifier whose output is amplified by a variable gain amplifier that provides critically damped driving current to drive the horizontal deflector in the scanner. The vertical driving signal generator, lower part of Fig. hO is basically same as the horizontal driving signal generator except it is clocked four times slower than the horizontal generator. The vertical displacement perturbation signal is added to the basic scanning signal through its analog adder. The channel switching information signal is simulated with a simulator of mechanical switches. The scanning rate is controlled by changing the frequency of clock, it ranges from few lines per second to 10 lines per second, limited by the response of light deflector. Since the light deflector ; electro-mechnical devices operated at critical damping condition, the transition time of light spot from, one switching unit to another constant, measured at 25-7 i^sec . becausi the driving signal 109 changes level faster than the deflector response. Then, the T 2 parameter =— is related to the clock frequency f by 1 T 2 16 T ' ^E 1 1 - f x 25.7 x 10 To measure the transmission efficiency T and injecting noise factor r\ , one applies a d-c voltage V to one of the input channels and grounds the rest of them. Then, switching the input channel to one of the output channels, the output voltage "becomes v Q = rv + t^V The first part on the right is then a d-c term (information transmitted) and the second part is an a-c term (noise). Therefore, by measuring both d-c and a-c component, one can determine both r\ and I\ The measurement of one of the channels is shown in Figc kl. Since rj is an a-c term, peak-to-peak value is plotted. To determine the cross -talk factor, the switched input channel is grounded and the rest of input channels are applied with d-c voltage V. Then the output is V Q = p(n-l)V in which n is the number of input channels. In the measurement one cannot determine the meaningful quantities, because it is smaller than the noise picked up by the panel and leads . It is negligibly small as expectedo 110 CJ 1 1 >1V3d-01-XV3d 1 a 1 O H 1 o t H 1 ^ ■ "i T t < O lO II Flu* °\ \ / x 1° X / I O UP- / II II / x O / /x "+ — X 1 1 1 1 6 CD 6 O <0 6 in 6 Ill 8 . SUMMARY The aim of this work is to present a new approach of switching matrix system. Theoretical analysis of various factors in this system has "been done. It leads to several general con- clusions, (l) The photoconductor must have large decay to rise time ratio. (2) The photoconductor must have large dark-to-on resistance ratio. (3) The system must not suffer from high background light, and (h) To have the system operate properly, high load resistance at output channel is required. An experimental model has been constructed and tested. It proved that the system is feasible indeed . The capacity of the system can be extended to the order 3 h with 10 ~ 10 switching unit per panel is the integration technique is used to construct the matrix panel. A high resolution and high speed light deflecting system is then necessary for proper system performance. LIST OF REFERENCES [1] Wo Shockley and W. T. Read, Jr., Phys . Rev ,, 87, PP. 835; 1952. [2] R, N. Hall, Phys. Rev ., 87, pp. 387; 1952 . [3] L. Bess, Phys. Rev ., 105, PP« 1469; 1957- [4] L. Bess, Phys. Rev .;, Ill, pp. 129; 1958. [5] Ao Rose, Phys. Rev ., 97. PP° 322; 1955- [6] H. Y. Fan, Phys. Rev., 92, pp. 1424; 1953 [7] R. Ho Bube , Photoconductivity of Solid , John Wiley and Sons, New York; i960. [8] A. Rose, Concept in Photoconductivity and Allied Problem , I nt e rscience; 1963 • [9] G. A. Dussel and R. H. Bube, J. of Appl . Phys., 37 . PP. 934; 1966. [10] A. L. Bloom, Proc. IEEE, 54, No. 10, pp. 0262; 1966. [11] T. S. Chu, B. S.T.J.,, 45, pp. 2.87; .1.966. [12] H. Kogelnik and T, Li, Proc. IEEE , 54, No. 10, pp. 1312; 1966. [13] K. Wo Boer, Physica S tatus Solid! , 8, pp. K179; 1965. [14] Midwestern Instruments Bulletin OE-1053> Tulsa, Oklahoma. [15] Ao Korpel, R. Adler, etc., IEEE J. of Quantum Electronics , QE-1, pp. 60; I96S o [16] A. Korpel, etc, Free.. IEEE, Vol. 54, Ko„ 10, pp. 1429; 1966. [17] W. Haas, R. Johannes, and P. Cholet, Applied Optics , Vol. 3, Nco 12, pp. 1504; 1964. W. Kulcke, To J. Harris, K. Kosanke and E. Max, IBM Journal , 65; January 1964. L , Wallman , Storage of Analog Vari ables in Delay Lines , M.S-. Thesis, University of Illinois; 196*8. 112 113 VITA Tuh Kai Koo was born on September lk, 1936, in KainSu, China . He received the Bachelor of Science degree from National Taiwan University, Taipei, Taiwan, China, i960. In 1962, he enrolled with the School of Mines and Metallurgy, University of Missouri, Rolla, Missouri, where he received the Master of Science degree in 1963° He joined the Circuit Research Group of the Computer Science Department at the University of Illinois in the fall of 1963. He has been a half-time research assistant to continue his graduate work since then. He is a member of Sigma -Xi, Kappa -Mu-Epsi Ion and the Institute of Electrical and Electronic Engineering. UNCLASSIFIED Security Classification DOCUMENT CONTROL DATA - R&D (Security claealtlcatlon of title, body of abatract and Indexing annotation muat be entered when the overall report ia classified) I ORIGINATING ACTIVITY (Corporate author) Department of Computer Science University of Illinois Urbana, Illinois 61803 2a. REPORT SECURITY CLASSIFICATION Unclassified 2b CROUP 3 REPORT TITLE Optoelectronic Switching Matrix 4 DESCRIPTIVE NOTES (Type of report and inclusive datea) Technical Report; Ph.D. Thesis 5 AUTHOR^) [Last name, first name, initial) Koo, Tuh-Kai 6 REPORT DATE June, 1^68 7a- TOTAL NO. OF PAGES 113 7b. NO. OF REFS 19 ■ 8a. CONTRACT OR CRANT NO. 6. PROJECT NO. 9a. ORIGINATOR'S REPORT NUMBEftfS.) 9b. OTHER REPORT NO(S) (A ny other numbers that may be assigned this report) 10 AVAILABILITY/LIMITATION NOTICES 11. SUPPLEMENTARY NOTES None 12. SPONSORING MILITARY ACTIVITY Office of Naval Research 219 South Dearborn Street Chicago, Illinois 60604 A cross-bar type switching matrix system with photoconductive cells as basic switching elements and an optical signal as the switching driving source has been studied and developed. The switching unit at each crosspoint in the matrix contains three photoconductive cells in T-connection, in which the center branch is grounded and the two arm branches are connected to input and output channels re- spectively. Switching action is accomplished by scanning the matrix panel with a programmed light beam. The state of a switching unit depends upon whether the center branch or the arm branches are being excited by the light beam. The equivalent circuit of a photoconductive cell under the specific application has been developed. The complete equivalent circuit contains harmonic terms which are the consequence of optical beam scanning. These harmonic terms are responsible for the noise appearing at the output. The performance of the system is analyzed with the equivalent circuit. Its characteristic is described by defining transmission efficiency, cross-talk factor and injection noise factor. One finds that satisfactory functioning of the system requires a photoconductive cell with a large ratio of off (dark) -resistance to on resistance and a large ratio of decay time constant to rise time constant. The scanning rate of the light beam is also an important factor in the good performance of the system. A prototype system with a capacity of four input channels and four output channels has been constructed to demonstrate the feasibility of the idea. Experi- mental results were very satisfactory. DD FORM 1 JAN 64 1473 UNCLASSIFIED Security Classification UNCLASSIFIED Security Classification 14 KEY WCRDS Optoelectronic Switching Matrix inducing factors LINK A ROLE LINK C ROLI WT INSTRUCTIONS \. ORIGINATING ACTIVITY: Enter the name and address of the contractor, subcontractor, grantee, Department of De- fense activity or other organization (corporate author) issuing the report. 2a. REPORT SECURITY CLASSIFICATION: Enter the over- all security classification of the report. Indicate whether "Restricted Data" is included. Marking is to be in accord- ance *ith appropriate security regulations. 2b. GROUP: Automatic downgrading is specified in DoD Di- rective 5200.10 and Armed Forces Industrial Manual. Enter the group number. Also, when applicable, show that optional markings have been used for Group 3 and Group 4 as author- ized. 3. REPORT TITLE: Enter the complete report title in all capital letters. Titles in all cases should be unclassified. If a meaningful title cannot be selected without classifica- tion, show title classification in all capitals in parenthesis immediately following the title. 4. DESCRIPTIVE NOTES: If appropriate, enter the type of report, e.g., interim, progress, summary, annual, or final. Give the inclusive dates when a specific reporting period is covered. 5. AUTHOR(S): Enter the name(s) of authoKs) as shown on or in the report. Enter last name, first name, middle initial. If military, show rank and branch of service. The name of the principal author is an ahsolute minimum requirement. 6. REPORT DATE: Enter the date of the report as day, month, year; or month, year. If more than one date appears on the report, use date of publication. la. TOTAL NUMBER OF PAGES: The total page count should follow normal pagination procedures, i.e., enter the number of pages containing information. 76. NUMBER OF REFERENCES: Enter the total number of references cited in the report. 8a. CONTRACT OR GRANT NUMBER: If appropriate, enter the applicable number of the contract or grant under which the report was written. 86, 8r. & 8d. PROJECT NUMBER: Enter the appropriate military department identification, such as project number, subproject number, system numbers, task number, etc. 9a. ORIGINATOR'S REPORT NUMBER(S): Enter the offi- cial report number by which the document will be identified and controlled by the originating activity. This number must be unique to this report. 9b OTHER REPORT NUMBER(S): If the report has been assigned any other report numbers (either by the originator or by the sponsor), also enter this number(s). 10. AVAILABILITY/LIMITATION NOTICES: Enter any Urn- ''•'' rthei