IHHBRffiHIiisiIl HflHS EflnlHlMiiKIImJ JfifisBL HHM Ss tamStSSBiAv AVr l WmM UUU33 Utf Wtttitisffl ig|||lH| MB ■BHHORnmltll^miral HHHRQ LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 510.84 X-Gfer -no- 629 -5 34 co/o. 2. CENTRAL CIRCULATION AND BOOKSTACKS The person borrowing this material is re- sponsible for its renewal or return before the Latest Date stamped below. You may be charged a minimum fee of $75.00 for each non-returned or lost item. Theft, mutilation, or defacement of library material! can be cause! for student disciplinary action. All materials owned by the University of Illinois Library are the property of the State of Illinois and are protected by Article 16B of lllinoit Criminal Law and Procedure. TO RENEW, CALL (217) 333-8400. University of Illinois Library at Urbana-Champaign 2001 5 2001 MAY 2 6 2002 When renewing by phone, write new due date below previous due date. L162 77}* ZL -r .£ mucDcs-R-72-532 COO 1^69-0210 PERSPECTIVE TRANSFORMER FOR THE STEREOMATRIX 3-D DISPLAY SYSTEM by SHIV PRAKASH VERMA October, 1972 UIUCDCS-R-72-532 PERSPECTIVE TRANSFORMER FOR THE STEREOMATRIX 3-D DISPLAY SYSTEM by SHIV PRAKASH VERMA October, 1972 Department of Computer Science University of Illinois Urbana, Illinois 61801 This work was supported in part by Contract No. AEC AT (11- 1)1^69 and. was submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering, October, 1972. Digitized by the Internet Archive in 2013 http://archive.org/details/perspectivetrans532verm Ill ACKNOWLEDGEMENT The author wishes to express his sincerest gratitude to Professor W. J. Poppelbaum for suggesting this thesis topic and for his continued guidance, friendship and support. He is also very grateful to Professor W. J. Kubitz for his invaluable aid, timely suggestions, friendship and patience. The author wishes to thank the other senior staff members of the Hardware Research Group and his fellow students in the group for their friendship. Thanks are also due to Barbara Bunting for typing the thesis, to the drafting department, to the technical staff for circuit fabrication and to the offset department of the Department of Computer Science. IV TABLE OF CONTENTS Page 1. INTRODUCTION 1 2. THE STEREOMATRIX SYSTEM 3 2.1 The Perspective Transformer 5 2.2 The Rotation and Translation Coefficient Generator ... 5 2.3 Cursor and Inverse Transformer 8 2.4 Programmed Storage 9 2.5 Observer Position Detector 9 2.6 The Laser Display 11 2.7 Physical Layout of the System 11 3. PERSPECTIVE TRANSFORMATION l4 4. SPECIFICATIONS AND BLOCK DIAGRAM OF PERSPECTIVE TRANSFORMER . 19 4.1 Specifications 19 4.2 The Overall System Description 20 5. SYSTEM DESIGN OF PERSPECTIVE TRANSFORMER 22 5.1 Digital to Analog Converters 22 5.2 Coordinate Transformer Unit 25 5.3 Scaling Unit 25 5.4 Windowing Unit 29 5.5 Perspective Generator Unit 32 5.6 System Control 35 6. ANALYSIS AND DESIGN OF HIGH SPEED ANALOG MULTIPLIER AND DIVIDER FOR THE PERSPECTIVE TRANSFORMER 38 6.1 Analysis of the Multiplier Circuit 38 6.2 Circuit Description 45 6.3 Multiplier Scale Factor 47 6.4 Test Results 48 6.5 Analog Divider 48 7. DESIGN OF THE FAST ANALOG SWITCH 51 8. RESULTS AND CONCLUSION 57 8.1 Results 57 8.2 Conclusion . . 57 LIST OF REFERENCES 60 APPENDIX . . . 62 VITA 71 V LIST OF FIGURES Figure Page 1. Block Diagram of STEREOMATRIX System k 2. Block Diagram of Rotation and Translation Generator . . 7 3. Block Diagrm of the Observer Position Detector .... 10 h. Layout of the Optics for Laser Display System 12 5. Physical Layout of the STEREOMATRIX 13 6. Formation of Stereoscopic Pair 15 7. Overall Block Diagram of Perspective Transformer ... 21 8. Fast D-A Converter 2h 9. Block Diagram of Coordinate Transformer Unit 26 10. Actual Diagram for One Coordinate Transformation ... 27 11. Resistance Characteristics of Vactec's Vactrol Type VTL2C3 30 12. Schematic Diagram of Scaling Unit 31 13. Block Diagram of the Windowing Unit 33 Ik. Schematic Diagram of Windowing Unit 3^+ 15. Block Diagram of the Perspective Generator 36 16. Logic Diagram of System Control 37 17. Multiplier Circuit Model Diagram Uo 18. DC-20 MHz Four-Quadrant Multiplier k6 19. Output of Multiplier and Level Shifter U9 20. Analog Divider 50 21. Circuit Diagram of Fast Analog Switch for D-A Converter 52 22. Circuit Diagram of Analog Switch for Multiplexor ... 55 23. Analog Gate Switching Characteristics % 2k. Schematic of Perspective Transformer System 58 1. INTRODUCTION ■ In the last decade two-dimensional displays (CRT terminals) have been accepted as an effective tool for better man-machine interaction^ . In these systems the graphical information is input via a light pen or a tablet and commands are invoked by either pointing at a light target displayed on the screen or making an entry on a keyboard or a pushbutton device. Many of the potential applications of these interactive displays may be classified under the general heading of computer aided design or design automation. Since the human observer is used to interacting with three- dimensional surroundings and much of the data involved in these interactions (sketches, drawings, diagrams, curves, etc.) are three-dimensional in nature, scientists have continually stressed the need for a good three-dimensional man-machine interface. The lack of such a device has forced users into an unnatural compromise in data processing. For example, the three-dimensional view of a building or a bridge is presented on a two-dimensional interface - a cathode ray tube (CRT). A fighter pilot in training who operates in a three- dimensional environment, would probably be happier if he could replace his two-dimensional CRT display displaying combat maneuvers with an equivalent three-dimensional man-machine interface. Furthermore, the displays on which many kinds of information are currently being displayed are also limited as to size and thus the number of observers who can efficiently use the dis- play is also limited. Therefore, one may conclude that in order to provide an efficient interaction between man and machine, it is highly desirable to have an on-line, large screen, three-dimensional interactive display system. The system at hand is an effort in this direction. There are many possible methods available for implementing three- dimensional displays x ' ' ' '. However, under the real time and interactive requirement of a 3-D man-machine system, our choice is limited to a stereo- pair type of 3-D display. In STEREOMATRIX the stereo-pair method of 3-D display has been used in designing the first large screen on-line three-dimensional man-machine interactive system for displaying computer generated transparent "wire-frame" line drawings . 2. THE STEREOMATRIX SYSTEM The STEREOMATRIX system, proposed by Dr. W. J. Poppelbaum, is the world's first large screen, real time, three-dimensional, interactive display system which presents computer generated three-dimensional line figures. This system has the capability of providing rotation, translation and scaling of the displayed figure in virtual 3-D space. The perspective of the figure changes as the observer moves about. The observer can interact with the computer by identifying a point in the display volume or create his own draw- ing with the help of a 3-D cursor and a joystick. The display volume is a cube with 500 elements on a side. This system has the capability for displaying, in 3-D, x, y, z analog signals obtained from the outside world or those from a 3-D analog pattern generator. The principle used for creating the 3-D illusion is that of the stereoscopic pair. High speed hardware in the system generates the point seen by the left eye and the right eye for each incoming triplet (x, y, z) obtained from the computer. These two points are projected onto a rear-projection screen by two laser beams which are orthogonally polarized. The observer's left eye is constrained to see the left point, and his right eye is constrained to see the right point by means of a pair of polarization analyzing glasses. Thus, the observer perceives a single point formed at the intersections of the extended lines where the image of a point seen by the right eye coincides with the image of the same point seen by the left eye. A block diagram of the STEREOMATRIX system is shown in Figure 1. The basic components of this system are the perspective transformer, the rotation and translation coefficient generator, the cursor and inverse transformer, the pro- grammed storage, the observer position detector and the laser display. §e UJ o X UJ h- CO lll tr n UJ > z cr o UJ CO h- GO co O o Q- I 2.1 The Perspective Transformer As the name implies, the transformer accepts at its input a graphic element (x, y, z) and transforms the element into a stereoscopic pair, (X T , Y) and (X_, Y). Due to the number of points required in the display and real time nature of the system, the speed of transformation is very high. During rotation, trans- lation and scaling, the perspective transformer operates on the input x, y, z coordinates using the twelve linear transformation coefficients from the coef- ficient generator, scales them and generates the stereoscopic pairs. It also uses the observers X and Z position from the screen in changing the perspective. During the input mode it multiplexes -in the cursor position and displays its position in the 3-D space. In the interactive mode the perspective transformer provides a transformed X, Y, Z signal in order to obtain a coincidence with the cursor position so that the observer can store this point in the computer and may attach some importance to this point for his design manipulations. The perspective transformer also communicates with the computer. It accepts the X, Y, Z triplets and signals back to the computer for a new point after displaying the previous one. If the point exceeds the window size, it does not send the intensity signal to the display but instead immedi- ately signals the computer for a new point. A detailed description of this unit will be given later. 2.2 The Rotation and Translation Coefficient Generator The coefficient generator supplies the twelve elements of the rota- tion and translation coefficient matrix to the perspective transformer. The coefficient matrix can be represented as t ll t 12 t l3 t lk T = t 21 t 22 t 23 t 2k t 31 S2 t 33 Sh j (1.1) This matrix is not orthogonal due to terms t,, , t , and t , . Therefore, it does not commute. This makes it necessary to store the previous matrix in order to obtain the new one. Initially, all the diagonal elements (t,,, t . t ) are one and the rest of the elements are zero. When the observer chooses to rotate the figure in viewing space, this unity matrix is multiplied by the appropriate operator matrix and the new matrix is stored back. That is, for a X rotation of 9 degrees, this unit generates Zt ij from previous Ztij as: t 11 t 12 t 13 t Ik t 21 t 22 t 23 t 2k t 31 t 32 t 33 t 3^ 10 Cose Sine -Sine Cose *u '12 \ 3 hh t 21 t 22 t 23 t 2i+ Si t 32 *33 V . . . (1.2) This unit rotates and translates the figure incrementally in order to achieve a smooth rotation and translation. It is intended to rotate the figure by 0.2 degree per frame of the 3-D input picture. A block diagram of this unit is given in Figure 2. The coefficients are stored in twelve ten-bit registers, and the outputs of these registers are connected to ten-bit D-A converters. The D-A converter outputs are sup- plied to the perspective transformer. The observer selects any desired operation by pressing a button on the control box. For example, he may want to rotate the figure about the X axis of the viewing space by a certain angle. For this case the coefficients which correspond to X rotation (t , 1; , t , t^, t^, t , t and t ^) are operated on by the multiplier and adder unit, and the new values corresponding to the desired X rotation are written back into the storage registers. This process Ill o z < o 3 | ITT 2 S S LLL -e- P 3 nr m nr m 88 - s gvi i s °2S- ■9- -£■£-«• . tttt T tt! ft! O ly K O K o o -p CO U CD sd cu G O •H -P cO H w EH d CO o ■H -P CO -P O « O I bO cO ■H Q o o H pq OJ CD •H *i HI 8 is performed incrementally. That is, an X rotation of 12 would be performed in sixty frame periods, since during each frame the figure rotates by 0.2°. Pro- vision is made for the observer to operate this unit in two modes, fast and slow. 2.3 Cursor and Inverse Transformer The cursor is a 3-D cross positioned in the display volume by means of a joystick which generates a static voltage corresponding to its X, Y, Z position in the viewing space. Thus, the cursor is an input medium for this 3-D man-machine interface similar to the light-pen in a 2-D system. Since the cursor is generated in the viewing space coordinate system, the X, Y, Z analog signals corresponding to the cursor's position must undergo an inverse transformation in order to obtain the cursor position in computer j space. The analog X O 5 NJ X t c * o — «- o. ui - c — O 3 <~> 5 T X 1 a> < t o < r o % a § o CD j: 5 ■n o o SL o CO CO c UJ I I J < t o to u T (!) < t °3 CO X (Li 2; ■o O O C" (_> (0 c LU w CD CD to r o o u .c <-> CJ C) <"£ J k. ^_ ^~ T3 o O .c O (0 C UJ 8 -p o CD -P cu w -Q O CO O faO CO •H Q o o CD I •H a> .c CJ o» 3 _J o CO 11 2.6 The Laser Display This unit projects stereoscopic pairs generated by the perspective transformer on a 3 x h rear project screen. A one-watt argon laser at 51U5A is used as the light source. The laser beam is first intensity modulated by an electro-optic modulator with the intensity signal of the input picture and then it is vertically deflected by the Y signal of the sterescopic-pair . Next, the beam is split into two beams. These two beams are polarized per- pendicularly to each other and serve as the right eye and left eye image light source. [The observer wears a pair of polarizing glasses in order to separate the two images.] These two beams are horizontally deflected by the appropiate X-signal. The scanning is random and thus requires high speed deflectors with a resolution of 500 points. State-of-the art ultrasonic deflectors have been developed for this purpose. The analog deflection signal is used to frequency modulate an ultrasound signal in a lead molybdate crystal which has been placed in the path of the laser beam. The light is directed onto the crystal at the Bragg angle. The ultrasonic signal has a central frequency of 120 MHz with an 80 MHz bandwidth, figure h illustrates the layout of the optics for the laser display system. 2.7 Physical Layout of the System Figure 5 illustrates the physical layout of the STEREOMATRIX system. All the electronic hardware for the perspective transformer and the coefficient generator and cursor is housed in two cabinets. The observer position detector hardware is mounted under the Display Screen desk. The two shaft encoders are mounted on the either side of the top of the screen. The laser and its pro- jection optics are behind the screen. The observer is allowed to move upto 10 feet away from the screen and 5 feet on either side of the center of the screen. The screen lies on the coordinate axis of the display volume. -p w H ft W •rH o h > o •H w >s CD tiO 11+ 3. PERSPECTIVE TRANSFORMATION Generating a perspective image of three-dimensional information is relatively easy. The formation of the stereoscopic pair is shown in Figure 6. The image plane is Z - 0. The observer is at a distance (X , Y , Z ) from the image plane: PL and PR constitute the stereoscopic pair formed for the point P(x, y, z) in the display space. These points are the intersections of the lines joining PL and PR to P with the image plane. The geometry is elementry and one can write down the relation between the coordinates of the stereoscopic pair and the point P(x, y, z) as: X L = (X Q - a/2) ♦ _2-j (x - X ♦ a/2) . . . (3.1) X R (x o + a / 2) + z~T7 (x " x o - a / 2) • ' • (3>2) Y - Y o + z^Tl <* " V ^ The above equations describe the relationship between a triplet (x, y, z) and its stereoscopic pair. However, STEREOMATRIX has a built-in facility for rotation, translation and scaling of the figure in the virtual 3-D space. Therefore, every point obtained from the computer undergoes the following transformations before it is displayed as a stereoscopic pair: Stereoscopic Pair = [Windowing] [Z Division] [Perspective] [Scaling] or X -WDPSTX. out m Rotation and Scaling [x. in" 15 P(x,y,z ) L(X ,%rZ ) P L (X L ,Y,0) R (Xqi ^2 »"Zo' Pr(X r ,Y,0) LEFT EYE RIGHT EYE Figure 6. Formation of Stereoscopic Pair 16 If we denote the elements of the rotation and translation matrix T as '11 L 12 l 13 Ik t. T = '21 22 23 24 t 31 *38 * 33 '3^ 1 and scaling matrix S as s x S Y S 3 1 and writing the perspective matrix P, as P = z o ^0 - a/2) z o <*0 + a/2) z o Y c 1 z (3.5) (3.6) (3-7) Then, by applying the rotation and translation operation to a point (x, y, z) obtained from the computer we have : r ' X = '■ T out j r i x i y z 1 t i: x + t 12 y + t 13 z + t lk t 21 X + t 22 y + t^z + t^ t 31 X + t^y + t 33Z + t^ 1 17 Then applying scaling, we have X out s X s y s z 1 hi x + \ 2 y + t 22 z + hk *21* + *&? + t 22 Z + t 24 t 31 x + t^y + t 33 z + t^ S x (t ll X + \& + \3 Z + V S y (t 2L x + t^y + t^z + t^) S z ( V X+ V + Ss 24 " V applying P we have out Z, (x Q - a/2; (*n + a/2) s y (t 21 x + t 22 y + t 23 z + t £4) s z " ftt !1 Is *5, 5 ci -o V— ft *5 _ o w • g; II -O \ k MI ' £<* M &£ . '1 ♦. * X >» M >. 9 C . R 2 i *. c — t> c o •= •s o I 5 M ■si " 1 ; «•< t »' i «' 1 X >> N k. o o ^. o> a> 01 c „_ c a 1 ^ o 3 a> O k» 1- C/) L„.. icient isity s a> z i ^j i jn X >Z N 7i Coeff Intei o O a> r oc ._ O 3 c o . o » V) a. ( f "5 a> o 3 H X / Coo Tran - o CL o * o — . k J L J I ■D C X >» N * 5? k. o 4> "5 < r o i a> ^_ Q c J*) o Q. O / / ■, / V a> <_ o 3 o a. t w c a> M " .. / v / ... t s 1 t A X >n N > 0) E.r V^, hi a> / S E a. U- **> E *^ o o °.° (_> c E L J o 0. Q 3 O X CD o ra CO EH CD > •H -P O (L) ft M a, o CO •H Q o o H rH H CO Ih 0) > o 0) bO •H p4 22 5. SYSTEM DESIGN OF PERSPECTIVE TRANSFORMER From the overall block diagram of the system (Figure 7), we can list the various subunits of the perspective transformer as; 1. D-A Converters 2. Coordinate transformer unit 3. Scaling unit h. Windowing unit 5. Perspective generator unit 6. System control We will consider the detail design considerations for each unit and their actual implementation. 5. 1 Digital to Analog Converters The inputs to the D-A converters are straight binary with the most significant bit being the sign bit. The main criteria in designing the D-A converter is the accuracy and conversion speed. Accuracy of +1/2 LSB and settling time of less than a micro-second was the main consideration in choos- ing the organization for D-A converters. ( f\ i \ Different method of D-A conversion were studied . Digital-to analog converters consist essentially of two parts: the actual decoding unit which is usually a high-output- impedance low current device and a high gain summing amplifier with low output impedance. There are two types of decoding networks commonly found in fast D-A converters: the weighted resistor network and the ladder network. However, from the accuracy point of view ladder network is superior for the following reasons: The ladder network has a more uniform current distribution and all 23 switches can be designed for approximately the same current: For example, 9 in a 10-bit weighted resistor network the bit currents vary as 2 :1 or 512:1, while in a ladder network the bit current vary by less than 14:1. The ladder networks smaller variation in current eases the switch design and can improve switching accuracy. Finally, an n-bit ladder network uses only two values of precision resistors; an n-bit weighted resistor network, n different values, The second important criterion of the D-A converter is the decoding speed. If the logic waveforms controlling the decoder are assumed perfect with no delay between switching edges, the the ladder network introduces some delay due to wiring capacitance between each resistor junction point. This is overcome by using a thin-film monolithic ladder network and designing a very fast analog switch. For the above reasons a ladder-type R-2R network decoder is chosen over the weighted resistor type. A fast single pole double throw switch which can switch a +10V reference signal is used for switching. Figure 8 shows the actual arrangement for the D-A converter. Double rank registers are used to store the digital information in order to avoid any transition noise going to the analog switches. Edge triggered flip-flops (type SN7^107) are used for reliability. The outputs of the line receivers are loaded first into the lower rank of the register and then its contents are transferred into the upper rank. The sign bit flip-flop is used to switch the proper polarity of the reference signal. That is, if the sign bit is one, then the -10V reference is switched, otherwise the +10V is switched. A high speed single-pole double-throw type of analog switch is used to switch the appropriate bits of the R-2R ladder. 2k Figure 8. Fast D-A Converter 25 From the accuracy and speed point of view the R-2R ladder of R = 10K is used. Allen-Bradley thin-film R-2R ladder networks are used. These ladders are noninductive (risetime less than 100 nsecs), possess small size and have excellent temperature coefficient tracking. In order to maintain the accuracy of 0.1 percent and the settling time of 250 nsecs, data device corporation high performance fast settling inverting amplifiers (type FS-23) are used. These amplifiers settled to 0.2% in 100 nsecs and have a slew rate of 1000V/usec . The D-A converter is con- structed on three standard size cards. 5. 2 Coordinate Transformer Unit The transformed triplets (x , y , z ) are formed from the analog signal of the triplets (x, y, z) and the twelve rotation and translation coefficients. Thus, this part of the system implements following equations: x t = \l x + \& + V + hk y t = t 2L x + t 22 y + t 23 z + t 2h z t = Si x + ^2 y + S2 Z + V A block diagram of this unit is given in Figure 9* High speed analog multi- pliers and high speed inverting operational amplifiers are used to obtain the coordinate transformation. Figure 10 shows the multipler and operational amplifier connections for one of the coordinates. All three analog multipliers and operational amplifiers are laid down on one standard size card. Thus, this unit has three such cards. 5.3 Scaling Unit This provides the observer with a control over the size of the figure in the display. Since this is a facility provided to the observer, the hardware 26 n -P •H a s o 'I 1 > 4 6 1 s ; > 4 6 1 r : ► 4 6 8 1.0 10.0 100 INPUT CURRENT - MILLIAMPERES 31 DC O K o f- p . !2 4 se 35 z o x l ■ (x o - a / 2) + rfr K - x o + a / 2) s 2 x r " (x o + a / 2) + 27TT" < x s " x o - a / 2 > s Z Y = Y + (v - Y ) Z. + z u s V s A block diagram of the implementation of these equations is given in Figure 15. This unit uses three fast analog multipliers and one divider and several summers to generate the sterescopic pairs. 5.6 System Control This unit provides the handshaking signal between the PDP8 e and STEREOMATRIX as well as the signal needed for transferring information into the D/A converters and generating the intensity control signal. Figure 16 shows the logic diagram of this unit. After receiving the PDPSe ready signal from the computer, this control transfers the first triplet (x, y, z) into the lower rank of the D/A storage registers and then into the upper rank. Then, the control waits for 7 u-secs before it turns on the intensity gate because of aperture time of ultra-sonic deflectors. After that it signals the computer for a new point. It also provides the multi- plexing signal for the cursor multiplexor and the frame pulse to the other units of STEREOMATRIX. 36 2 uj £ 2£ cr if) o a p uj > UJ cr h- UJ uj o z o t ce 1/5 ^ o a U o -p > •H -P O > GO o in M cti •H Q O •H O VO 01 u •H (O oj O V) V) 38 6. ANALYSIS AND DESIGN OF THE HIGH SPEED ANALOG MULTIPLIER AMD DIVIDER FOR THE PERSPECTIVE TRANSFORMER It is obvious from the diagrams of the preceding chapter that we need a high speed analog multiplier for the coefficient transformation and perspective generation. Multipliers with a full output frequency response of 10MHz are required for the system. The cost of commercially available 10MHz multipliers is very high. Therefore, it was decided to design a high speed multiplier for this system. During this process various methods of analog multiplication were studied ' , The speed requirement can only be met by the trans- conductance type of multiplier. However, the available circuits for linear transconductance multipliers could only be used in integrated version i . Therefore, a new linear transconductance multi- plier circuit was developed. This chapter deals with the analysis and design of this circuit. 6.1 Analysis of the Multiplier Circuit The following is an approximate analysis of a new circuit for an analog multiplier which can linearly multiply two voltages to produce a pro- duct with linearity approaching 0.1$. The operation of the circuit is unusual in the sense that it responds linearly to either of two inputs x or y, yet it provides a product, xy, as a result of nonlinear processing of both x and y. In the analysis which follows, we will derive equations which show that the signal distortion and error arise primarily from the effects of voltage offset between the various devices used in the multiplier. 39 The circuit model configuration for the multiplier is shown in Figure 17 where it has been assumed that all the transistors are identical and have small voltage offset generators in series with each of their emitters. Other than these voltage offset generators the devices are assumed to be perfectly matched and behave according to the ideal Ebers-Moll model. We assume that the lower differential amplifiers have been sufficiently linearized by the emitter resistors, R^. Thus, we can write V - (I + x)R + (I - x)R^ + V =0 X v ' E ' E x V X X = RE (1) Therefore, we see that the currents I + x and I - x are linearly dependent on the input voltage V . x We will consider, therefore, only the current transfer from the collectors of the multiplier outputs. That is, we want to calculate the (17) ratio I /xy. We begin by writing the Ebers-Moll expressions v ,; relating out' the collector currents and the base emitter potentials as shown below. (lb is assumed that leakage currents can be ignored and a's are absorbed into constant a^. In the analysis e ^ corresponds to exp (^).) h = a n e k¥ I - a e SL23 3 11 k.T h - hi ^ C6.D Uo u bD c6 0) O -P •H O Jh •H O V •rH H ft •H ■P rH s & hJD •H kl These expressions can be simplified by dividing some currents by others as follows : I x l(cp x - cp 2 ) - = e r-= We also note that I 2 1 3 q ^3 " V r k = e *t h + X 2 - z o + x y • *! - V + ^02 + ^2 = ° y - \ - %k + % 3 + ^3 = ° Simplifying Equation (6.3) we get: 1 + V 1 ! 2-r+i^ i Io " x 3 "" 1 + I,,/I V"3 x o - x >* " i + ijAi, From Equation (6.^+) we get cp x - cp 2 = y - cp Q1 + cp Q2 (6. 2 ; (6.3) (6.V (6.5 h2 q qy -q \ e ^ (cp x - cp 2 ) = e - e - (cp Q1 - cp Q2) = - q -qy q i 2 and e - (cp 2 - cp^ = e — e - (cp Q1 - cp Q2 ) = - Similarly: (cp^ - cp ) _■ (y - cp Qi+ + cp ) q qy -q \ e - (r Pi+ - q) 3 ) = e - e - (cp^ - cp^) = - -q -qy q i. e S (cp 4 " cp 3 ) = e — e - (cp Ql+ - cp^) = ^ . . . (6.6) Substituting the above values into Equation (6.5) we get l -, -qy q / \ 1 + e KT 8 i^T (c P l - V I.. 1 + e i e § Cq> 01 " cp Q2 ) I - x I. = • • • (6.7) 1 + e s e a <<& - V 4 " . -qy q / 1 + e Tt e M ( 'V " ^03' The multiplier output currents can now be obtained by noting: H =I 1 + I 3 (6 . 8) which produces ^3 J o + x . z o ~ x °^ = 1 + -=g e |- (cp Q1 - cp 2 ) 1 + egeg(cp ,-cp 03 ) »'» i out = x o + x T o " x (6.10) OU 2 1 + e i e S (( P 1 - V X + e ll e kT Kk ~ %1 ] Now assuming the offsets are small compared with KT/q, we can simplify the offset terms as follows : S fe (C P 1 - CP 2 ) = X + fe ( V " V + •'• = (1 + £ 1 } e q (kT e qy ~ (1 + —) 2 v kT assuming e , e very much smaller than one and y^ is also smaller than one, 1 2 Kl then above expression can be simplified to the form AI q (e l - € 2 } ( £ l + e 2 } out „ - - xy + - i o + x (6.14) This is the general expression for the differential output current including the error terms which arise from the presence of voltage offsets. Hence, we see that the differential output current is a sum of a product term, a constant offset and a linear error in terms of x. As an example of the effects of these error terms, consider first the case when x is a constant, C . Then the output contains only terms in y or constants as below: c o < I o o -p •H ? co bO O H a -p w h O •H o o •H O H OJ bO En Q. ui 3 o b£«- g°9: 53 diode D is forward biased, and the FET gate diode is reverse biased. The FET leakage current I (OFF) maintains a slight forward current in D and the gate is clamped near -15V. Thus, the pinch-off voltage is exceeded for any source voltage more positive than -10V (maximum negative voltage to be switched) and the switch is OFF. The gate to source capacitance is thus charged to +15V + V . s The positive transient at the collector of the 2N3905 (i.e. when it starts turning ON) reverse biases the diode D , and the transient then appears c i at the gate reduced by the capacitive division factor - — -— ; — . Therefore, 1 gs it is necessary that the voltage change at gate fully discharge C or the gate will not turn ON. Consequently, C must be large with respect to C 1 gs to permit a change equal to (15V + V ) <; 25V. Alternately, D must have leakage current capable of charging C within a very short time after the transient. At the time when the 2N3905 has fully turned ON and its collector is at +15V, D, is reverse biased and the gate-to-channel junction is forward biased by the D leakage current. Then the gate is decoupled, V = or 1 Gs slightly positive, and the switch is ON for d-c voltages of +10V. When the 2N3905 drive transistor tries to turn off (i.e. at the negative transient) D again becomes forward biased. This decouples the trans- ients to the gate of the JFET, and the switch is turned OFF when the gate becomes more negative than the source by the value of pinch-off voltage of JFET. Another switch across the output of JFET performs the complemented function to the series JFET. Therefore, it insures that the output of gate is grounded when the switch is OFF. Th 54 e value of C. in the circuit is determined for the 2N4093 JFET. It has C =10 pf. Therefore, in order to switch +10V with +15 volts drive, gs C must charge to 25 volts so that V__ = for an input of +10V. Since the gs ub input drive swing is 30V, C is allowed to charge to 5V. To solve for C we have : AVC, = AVC 1 gs 5 x C - 25 C 1 gs C. = 5 C 1 gs Since C =10 pf, C, > 50 pf. Since the diode has some capacitance, the gs 1 available value of ^3 pf was chosen for use in the circuit. One of the advantages of using a capacitive or diode drive is that for any source voltage, it maintains the gate to source voltage equal to zero (i.e. V =0). If we look at the ON resistance versus gate to source voltage ^ gs characteristics of the JFET, we will notice that the ON resistance is mini- mum for zero gate to source voltage. Therefore, we always get a constant and minimum ON resistance for the switch even for different input voltages. This also produces the same switching times for different input levels. In the analog multiplexor the shunt switch is not used and the capaci- tive drive had to be changed so as to avoid clipping of the a-c signal. Figure 22 shows the complete switch for the analog-multiplexor. These switches were tested upto 6.6 MHz for satisfactory operation. The photographs of Figure 23 show the switching times of the gate. 55 3 O o 3 < Z < o. m ro o> Z 1-AA/V— I > = *2 +i o o _l < z < B 0) H ft •H s o -p o o U •H ■H O (M CV| bO •H 56 Input and Output of Analog SWITCH HORIZONTAL 20 nsec/div and Vertical 2 V/div Input and Output of Analog Switch for Control Signal Rate of 6.6 MHz (Horizontal 50 nsec/div, Vertical 5V/div) Figure 23 Analog Gate Switching Characteristics 57 8. RESULTS AND CONCLUSION 1,1 Results Figure 2^4- gives the complete schematic of the perspective transformer. The x, y, z analog inputs were obtained with the three-dimensional analog pattern generator (PAGAN) used as the input to the system. The stereoscopic pairs generated by the system are displayed on two CRTs. We could see perspec- tive of the patterns changing with the observer's movement. We also see rota- tion and translation of the figure as commanded by the observer from his control box. We have been able to align the optics of the laser display to project one of the pairs on the screen. Therefore, all the goals outlined for perspective transformer have been met. 8. 2 Conclusion Analog circuitry has been commonly used in display applications, where speed is not a major factor. Building a high speed analog perspective transformer has illustrated the feasiblity of using high speed analog cir- cuitry in a display application where speed is more important than accuracy. In fact great accuracy could not be appreciated by the observer. The system is passively interactive in that the observer's position is monitored, and the perspective of the view is corrected for his particular viewing location. Thus the perspective transformer provides not just a static stereo but a dynamically corrected stereo view. Since the observer is to be allowed to move about over any reasonable area (10 ft. x 10 ft.), this made it necessary to construct the system with a large screen. This necessitates, of course, a laser light source and a sophisticated deflection system. All of the components of the laser display 58 3 8 3 5 s 0) -P w >> CO a; a c o w in En 0) > •H ■p V ft w (U Pn o o •H -p s 0) o CO C\J 0) 3 •H in«N03 >■ «0a 59 have been procured, and soon it will be possible to see a 3-D display. The perspective transformer has been tested to work at the input rate of 5 MHz with input supplied by a pulse generator. It can display all the pattern generated by PAGAN which are generated at 1 MHz rate. There has been no noticeable distortion of PAGAN'S patterns after being transformed by the per- spective transformer at this input rate. This insures that the system main- tains the required accuracy for a good display. Thus, all of the goals outlined at the beginning of this thesis have been met. There are two additions which would enhance the purpose of STEREOMATRIX which can be added as another phase of the research project: a) The window size in the system is fixed. However, it would be nice to have the window size change with the observer's position. The implementation of this is somewhat complicated and is thus left as a future addition. b) During scaling the picture is simply magnified or demagnified and thus the number of points in the display remains the same. Therefore, after magnifi- cation the points in the picture will be sparsely located and some degradation of the resolution of the picture may be noticed. It would be also helpful to have a line generator in order to fill up the space between two points during magnification. To summarize, this effort in building the interactive large screen 3-D display for computer generated transparent "wire" drawings will be defi- nitely a milestone in providing a better man-machine interface. It is not too difficult to imagine that in the not-too-distant future an architect will be able to put his own imagination into a reality, a mathematician may see solutions to three-dimensional partial differential equations, and an air traffic controller may see the three-dimensional position of an aircraft in the vicinity of an air- port. 60 LIST OF REFERENCES (1) Lewin, Morton H, "An Introduction to Computer Graphic Terminals", Proceedings of the IEEE . Vol. 55, No. 9, pp. 154I+-52, September 1967- (2) Tilton, H. B., "Stereoscopic CRT Displays", Design News , Vol. 20, No. 26, December 22, 1965, pp. 88-96. (3) Wolvin, John, "Analyph Stereoscopic CRT Display System", National Symbosium on Information Display Technical Proceedings, I967. (k) Rawson, E. G. , "Vibrating Varifocal Mirrors for 3-D Imaging", IEEE Spectrum , September 1969? PP- 37-^3 • (5) Sutherland, Ivan E., "A Head Mounted Three-Dimensional Display", Proceedings of the Fall Joint Computer Conference, 1968. (6) Schmid, Hermann, "An Electronic Design Practical Guide to D-A Conversion", Electronic Design , October 2k, 1968, pp. ^9-88. (7) Pearman, C. R. and Popodi, A. E., "How to Design High Speed D-A Con- Converters", Electronics , February 21, 1964, pp. 28-32. (8) Sedra, A., and Smith, K. C, "Simple Digital Controlled Variable Gain Linear d-C-Amplifier", Electronic Engineering , March 1969' (9) New, F. D., "Voltage Controls Solid-State Non-linear Resistance", Electronics , February h, I96U. (10) Bilotti, A., "Operation of MOS Transistor as a Variable Resistor", Proceedings of IEEE, August 1966, pp. 1093-109^. (11) Newinixe, L. J., "FET Ride Operational Amplifier Gain, Minimizes Offsets Voltages", Electro-Technology , March 197 0, P- 31- (12) Bilotti, A. and Pepper, R. S., "A Monolithic Limiter and Balanced Dis- criminator for FM and TV Receivers", IEEE Transactions on Broadcast and Television Receivers , Vol. 3TR-13, pp. 5^-65 , November 1967. (13) Bilotti, A., "FM Detection Using Product Detector", Proceedings of the IEEE , Vol. 56, pp. 755-757, April I968. (lit) EEE Special Survey, "Packaged Analog Multipliers", EEE, November 1968. (15) Gilbert, B., "A New Wide-Band Amplifier Technique", IEEE Journal of Solid State Circuits, Vol. SC-3, December 1968, pp. 353-365. 61 (16) Gilbert, B., "A Precise Four -Quadrant Multiplier with Subnanosecond Response", IEEE Journal of Solid State Circuits, Vol. SC-3, December 1968, pp. 365-373- (17) Renschler, E. L. , "Theory and Application of a Linear Four -Quadrant Multiplier", EEE, May I969, pp. 6O-67. (18) Thornton, R. D. and others, "Multistage Transistor Circuits", Semiconductor Electronic Education Committee, Vol. 5, John Wiley and Sons, New York, I965. (19) Hoeschle, David F., "Analog-to-Digital, Digital-to-Analog Conversion Techniques", Wiley, New York, 1968. (20) Wolleson, Donald L. , "Analog Switching - High Speed with JFET's", EDN, January 15, 1970, pp. 38- kO. (21) Kvamme, E. F. "Analog Switching - High Complexity with MOS", EDN January 15, 1970, pp. 33-37- 62 APPENDIX PICTURES OF PERSPECTIVE TRANSFORMER CARDS 63 D-A Register Card 6k Six Fast D-A Switch Card 65 D-A Five Fast Switches, Ladder and Operational Amplifier Card 66 Analog Multiplexor Card 67 Coordinate Transformer Card 68 Z Division Card 69 Stereo Generator Card 70 Summer Card 71 VITA Shiv Prakash Verma was born in Tikamgarh, India on May 22, 19^2. He graduated from Government Intermediate College, Panna, India in 1958. In 196^, he received his B. E. (Hons ) in Tele -Communication Engineering from the University of Jabalpur, India. He completed M. Tech in Electrical Engineering from the Indian Institute of Technology, Bombay, India in 1966. He joined the Hardware Research Group of the computer section at Tata Institute of Fundamental Research in November 1965. In September 1966, he joined the Electrical Engineering faculty of H. B. Technical Institute, Kampur, India. In June 1968, he joined the Circuit and Hardware Systems Research Group of the Department of Computer Science, University of Illinois under Professor W. J. Poppelbaum to work toward a Ph.D. degree. He has held a joint appointment with the Department of Computer Science and Department of Psychology since January 1971- Form AEC-427 (6/68) AECM 3201 U.S. ATOMIC ENERGY COMMISSION UNIVERSITY-TYPE CONTRACTOR'S RECOMMENDATION FOR DISPOSITION OF SCIENTIFIC AND TECHNICAL DOCUMENT ( See Instructions on FCvtrte Side ) 1. AEC REPORT NO. coo 1I469-0210 2. TITLE PERSPECTIVE TRANSFORMER FOR THE STEREOMATRIX 3-D DISPLAY SYSTEM 3. TYPE OF DOCUMENT (Check one): E a. Scientific and technical report I I b. Conference paper not to be published in a journal: Title of conference Date of conference Exact location of conference. Sponsoring organization □ c. Other (Specify) 4. RECOMMENDED ANNOUNCEMENT AND DISTRIBUTION (Check one): 5c1 a. AEC's normal announcement and distribution procedures may be followed. ~\ b. Make available only within AEC and to AEC contractors and other U.S. Government agencies and their contractors. 3) c. Make no announcement or distribution. 5. REASON FOR RECOMMENDED RESTRICTIONS: 6. SUBMITTED BY: NAME AND POSITION (Please print or type) Shiv Prakash Verma Research Assistant Organization Department of Computer Science University of Illinois Urbana, Illinois 61801 Signature **/, c/j r^y/tUs^c:*,-, Date October, 1972 FOR AEC USE ONLY 7. AEC CONTRACT ADMINISTRATOR'S COMMENTS. IF ANY. ON ABOVE ANNOUNCEMENT AND DISTRIBUTION RECOMMENDATION: PATENT CLEARANCE: I I a. AEC patent clearance has been granted by responsible AEC patent group. I 1 b. Report has been sent to responsible AEC patent group for clearance. I 1 c. Patent clearance not required. BIBLIOGRAPHIC DATA SHEET 1. Report No. UIUCDCS-R-72-532 3. Recipient's Accession No. 5. Report Date October, 1972 PERSPECTIVE TRANSFORMER FOR THE STEREOMATRIX 3-D DISPLAY SYSTEM 7. Author(s) Shiv Prakash Verma 8- Performing Organization Rept. No. 9. Performing Organization Name and Address Department of Electrical Engineering University of Illinois Urbana, Illinois 61801 10. Project/Task/Work Unit No. 11. Contract /Grant No. 46-26-15-301 12. Sponsoring Organization Name and Address US AEC Chicago Operations Office 98OO South Cass Avenue Argonne, Illinois 60^39 13. Type of Report & Period Covered Thesis research 14. 15. Supplementary Notes 16. Abstracts STEREOMATRIX is a large screen interactive 3-D display system which presents computer-generated transparent "wire-frame" drawings stereoscopically. A 3-D cross called the cursor can be moved around in the viewing space by the observer by means of a joystick. This feature allows the observer to select a point in the viewing space for graphic manipulation and relay it to the computer space. He can also use the joystick to draw pictures in the 3-D viewing space The perspective of the figure changes with observer movement in such a way that the figure appears to be stationary in space. 17. Key Words and Document Analysis. 17a. Descriptors Stereomatrix 7b. Identifiers/Open-Ended Terms 7c. COSATI Field/Group 8. Availability Statement 19.. Security Class (This Report) UNCLASSIFIED 20. Security Class (This Page UNCLASSIFIED 21. No. of Pages 77 22. Price ORM NTIS-35 ( 10-70) USCOMM-DC 40329-P7 1 s ep '% > UNIVeMITYOf ILUNOirUK"***! ■■■■ ■ ■ ■ m Ha si ■ Ural ■ HH mm m SHU fflB bH 91 Ml ■I ■MEMO KM nm ■■1 Sl*Jij ■^^ Ban wMm