Some Vector Generating Techniques for Cathode Ray Tube /CRT/ Display NASA Technical Memorandum ~ I I I ' i t X-521-67-271 ' SOME-VECTOR GENERATING I TECHNIQUES FOR CATHODE RAY TUBE (CRT) DISPLAY SYSTEMS I I , I , , p , I N67- 3 1 3 0 2 (ACCESSION NUMBER) (THRU) L e / - (CODE) 0 i 04 (NASA CR O R TMX O R AD NUMBER) (CATEOORY) " I \ lUNE 1967 , I ' ' , GENERAL DISCUSSION This report will be confined to considering vector generating procedures f o r non-scan CRT displays, which constitute the majority of CRT display sys- tems available today. In these systems electrical signals a r e supplied to hori- zontal (X) and vertical(Y) deflection circuits and the electron beam is moved to a position on the CRT screen, determined by the magnitude of these signals. Normally, digital X,Y information is generated by the computer to which the display system is attached, or by internal logic circuitry, and passed through D/A converters, amplifier and compensation circuitry to the deflection system of the CRT. The electron beam is moved to this position and a symbol, speci- fied by an input code from the attached computer, is drawn by the character generator. In order to produce a line segment on the CRT screen, the initial X, Y co- ordinates and the terminal X,Y coordinates must be presented to a vector gen- e r a t o r by the attached computer. The electron beam is then moved from the initial to the terminal position in such a manner a s to produce a straight line segment. 1 Alternately, the initial X,Y coordinates and the horizontal (Ax) and vertical (AY) increments can be presented to the vector generator with the same effect. The vector generator can produce line segments in two different ways. It can generate a series of points between the initial and final coordinates of the vec- tor, the incremental point positioning method, o r it can continuously provide the deflection signal required to move the electron beam smoothly in a straight line path from the initial to the final location, the stroke method. INCREMENTAL POINT POSITIONING VS STROKE METHOD The incremental point positioning method has the following advantages: complete compatibility with the original display system requiring no modifica- tion of the display system analog circuitry, accuracy and inherent stability, complete endpoint matching for connected vectors. Furthermore , the intensity compensation that is required for some stroke type vector generators, when drawing vectors of different lengths, i s not needed when using the incremental point positioning method, since all vectors, regard- less of length, have equal brightness. The incremental point positioning method has, however, the following dis- advantages: line segments a r e a collection of points and not continuous, the time required for vector generation is high, there i s a noticeable departure from a straight line for some of the simpler point generating techniques, there is little flexibility for adaptation to more complex curve generation (circular, parabolic, hyperbolic, etc.). The quality of the vectors generated by the incremental point positioning method depends primarily on the point generating algorithm, the spot size of the electron beam, and the r a s t e r matrix used for point positioning. By the r a s t e r matrix we mean here the matrix of individually addressable discrete points, formed on the face of the CRT. Roughly, if the spot diameter is equal to o r larger than the distance between adjacent r a s t e r points, good line conti- nuity is achieved by a vector generator that plots points in near adjacent loci- tions in generating a line. Of course, the more r a s t e r points that a r e address- able the better a straight line segment can be approximated, but on the other hand, the more points must be produced to construct a given vector. Most CRT dis- play systems have a 1024 by 1024 r a s t e r matrix. The principal reason for having a vector generator of the incremental point positioning type, rather than none at all, is to remove the necessity of storing 2 all the intermediate points of a given line segment in the display computer, o r the refresh buffer memory. No significant writing speed advantage is achieved by this method, except when the display computer o r buffer refresh memory have a very slow memory cycle time. AN INCREMENTAL POINT POSITIONING METHOD FOR VECTOR GENERATION In this section one scheme of vector generation, using the incremental point positioning method will be presented. For the purposes of this discussion let us zssume that we a r e given a typical display system having the following characteristics: Random Positioning time of the beam on the CRT face 1 5 psec Time Required to generate an adjacent point 1 psec CRT display area 10" x 10" Raster matrix 1024 x 1024 Display refresh rate 30 times/second In order to keep the vector drawing time within reasonable limits let us assume that the vector generator will produce up to 64 points per given incre- ments AX and AY. This corresponds to 6 bits and one sign bit in the AX and AY registers, respectively, and a vector write time of 64 psec plus positioning time. F o r this system up to approximately 333 inches of connected vectors can be generated per flicker f r e e frameo* The basic circuit requirements for this vector generator are: (a) a seven Sit AX register (b) a seven bit AY register (c) a 1 megahertz clock *This approximation i s obtained by noting that for the system specified, the distance between ad- jacent raster points i s approximately 0.01 inches. Since points are generated at a rate of one per microsecond for one thirtieth of a second (then the display must be refreshed), a total of .0333 s e c = 3.33 x 104 points can be generated. This corresponds to 333 inches of line display. IxlO-Gsec 3 (d) a six stage counter (e) logic gates required to implement the point generating algorithm (this will be described in more detail in Appendix A) (f) The ten bit X and Y registers (normally found in the display sys- tem) which specify the instantaneous location of the electron beam will be replaced by two ten stage up/down counters with parallel input. This will not affect the normal operation of the display SYS- tem. The functional diagram of the vector generator is given in Figure 1. The display computer loads the X and Y up/down counters (by parallel in- put), and the AX and AY registers. The coordinates of the starting point of the 1 DISPLAY COMPUTEWBUFFER I 1 t D/A - D/A CONVERTER 1 CONVERTER + I TO X-DEFLECTION AX ALGORITHM GATES 4 I TO Y-DEFLECTION CIRCUITRY - UP/DOWN- CIRCUITRY t t * - UP/DOWN -, - GATE u’ ~- GATE 4 START TO CRT UNBLANKING CIRCUIT Figure 1. 4 vector are stored in the X and Y up/down counters, respectively, and the X and Y increments are stored in the AX and AY registers, respectively. The internal logic of the display system will generate a signal to start the clock counter se- quence. The successive states of the clock counter are gated with the AX and AY registers through the AX and AY Algorithm Gates and the resultant pulses are ei- ther added t o o r subtracted from the numbers stored in the X and Y up/down coun- ters, depending on the state of the sign bit in the AX and AY registers. Each time a pulse passes through either the AX or AY Algorithm Gates a signal is routed to the CRT unblanking circuitry and a point is generated on the face of the CRT. The START pulse is also passed on to the unblanking circuits and is used t o produce a point at the starting position of the vector. By this procedure a set of points is generated which will closely approximate the desired vector. The comparison between the actual vector and the generated set of points will be made in Appendix A, where the Algorithm Gates are discussed. The gating f o r the START pulse, the STOP and RESET of the clock counter at the end of the 64th count is not shown, but implied in Figure 1. At the end of the 64th count the clock counter is reset and clock pulses to the counter are inhibited. The display computer can now load the X and Y counters and AX and AY registers again and a new vector generating cycle can begin. For connected vectors only the AX and AY registers will be loaded by the computer, since the electron beam i s already at the starting point of the vector. The main limitation of this method is the time required to generate a given vector. Other considerations a r e the fact that for the system just described the time required to draw a short and long vector is constant (64 psec + positioning time) and that the maximum vector length for one set of computer instructions is only one-sixteenth of the full diagonal distance on the CRT face. These last two limitaticns can be overcome by additional logic circuitry as will be shown in Appendix B. But it is unlikely that the time required to plot adjacent points can be reduced much below 1 psec; s o that little can be done to lessen the total vector generating time. When the requirement for endpoint accwacy, stzbility, and compatibility with the original display system a r e the over-riding consid- erations, a longer vector generating time might be tolerable (especially if only a small amount of graphic information h a s to be displayed at one time). The extreme accuracy and stability of this method for vector generation makes its use desirable in computer aided design systems. GENERAL DISCUSSION O F STROKE TYPE VECTOR GENERATORS The two major limitations of the incremental point positioning technique, excessive vector drawing time and lack of continuity of the generated vectors, 5 can be eliminated with the stroke method of line generation. Here a continuous signal is sent to the deflection circuitry of the CRT by the vector generator and the electron beam is moved smoothly from the initial to the terminal point. There a r e basically two different types of deflection systems available for CRT's: electrostatic and electromagnetic.* The characteristics of both types of deflection systems a r e summarized below:? Characteristic Electromagnetic E le ct r o static Beam focus Well focused through- Beam defocused at out tube a r e a tube edges Internal Structure Simple Complex Tube Length Short Long Spot brightness High Moderate Deflection Sensitivity High Low Complexity of Amplifiers High Low Deflection Speed Low High Weight Heavy Light Since electromagnetic deflection offers the advantages of a well defined beam over the entire tube face, high spot brightness, and high deflection sen- sitivity most present day display systems employ t h i s mode of deflection. For both types of systems, however, the shapes of the signals that must be produced by the vector generators a r e very similar. The difference between the two systems a r e found primarily in the different types of amplifiers that a r e required to drive the deflection circuits. +There i s also a hybrid electrostatic-electromagnetic system where the electromagnetic portion is used for large deflections and the electrostatic for small deflections. In this case the electro- static deflection i s utilized to generate alphanumeric symbols at locations specified by the large scale electromagnetic deflection. The electromagnetic deflection is used for vector generation. tCharles W. Adam Associates, Inc., Computer Display Review, 11. 11.0 6 For electrostatic deflection* D'K, - Vl V where, D = distance the beam is deflected from the center of the CRT face by the deflection voltage VI. v, = potential difference between deflection plates V = accelerating voltage of the electron beam K, = constant factor that is determined by the tube construction For electromagnetic deflectiont where D = distance the beam is deflected from the CRT face center by the mag- netic field H H = magnitude of the magnetic field at the center of the deflection coil V = accelerating voltage of the electron beam K, = constant factor that is determined by the tube construction But since for a current carrying coil (containing no ferromagnet) the mag- netic field in the interior is approximately proportional to this current.$ H = K" i and, K, i D =- fi where i = current through the coil K, = tube parameter Kf' = deflection coil parameter *Theodore Soller, Merle A. Start, George E. Valley, Jr., Editors, MIT Radiation Laboratory Series, Cathode Ray Tube Display, McGraw Hill Book Company, 1948, p. 64 tIbid, p. 304 SIbid, p. 3 7 Thus it can be seen that in the electrostatic case the deflection of the elec- tron beam is proportional to the potential difference between the deflection plates and in the electromagnetic case the deflection is proportional to the cur- rent in the deflection coil. The defl.ection speed of the beam is limited primarily by the capacitative time constant of the deflection plates in the electrostatic case and by the inductive time constant of the deflection coil in the electromagnetic case. The function of the vector generator is to present linearly changing elec- trical signals to the X and Y deflection circuitry of the CRT, such that the am- plitudes of the signals are proportional to AX and AY respectively. There are two ways i n which the vector generator can do this. It can pro- duce deflection signals which will cause the electron bes.m to go from the initial to the final point in a constant time interval, regardless of vector length - COQ- stant time vector generator. O r it can produce deflection signals which will cause the electron beam to travel from the initial to the final point with a fixed velocity - constant velocity vector generator. The constant velocity method is pr‘eferable for two reasons. F i r s t , since the beam travels with a uniform speed over the face of the CRT, all vectors, regardless of length, will have the same brightness. Second, since most graphic information consists of a large collection of short length vectors, o r combina- tions at short and long vectors, a more efficient use of the display system is made. In contrast, for the constant time vector generating scheme it takes as long to draw short vectors as it does long vectors, and the vector drawing time must be chosen to be of such a magnitude s o as to allow the generation of the longest vector in the fixed time interval. Furthermore, since both long and short vectors are drawn in the same time period, the short vectors will be gen- erated more slowly across the CRT than the long vectors and thus appear brighter, unless intensity compensation circuitry is provided. However, it appears that constant time vector generators of good quality are easier to implement than the corresponding constant velocity generators. For constant velocity vector generation the following vector generzting procedure is followed (after the coordinates of the starting point and A X and AY have been loaded into the appropriate registers of the display system by thc dis- play computer) : 1. Cornpute the total vector length “ L ” L = r‘(AX)2 + ( A Y p 8 2. Compute the time "T" required to draw the given vector T = L/S where S = constant vector speed 3. Compute the X and Y components of velocity 'rSll' and "S2" where S , = AX/T S, = AY/T 4. Generate a sawtooth signal for the X-deflection circuitry whose slope is "S,", for a time interval rrT". 5. Simultaneously generate a sawtooth signal for the Y-deflection cir- cuitry whose slope is ' I 5,'' for a time interval "T". 6. At the same time unblank the CRT beam for a time interval "T". Even though the required calculations can be performed by analog, digital, and/or approximational techniques, all of the above functional steps must be carried out by the constant velocity vector generator. On the other hand for the constant time vector generator the electron beam unblanking time is usually fixed. In that time interval sawtooth waveforms, whose slopes depend on the values found in the AX and AY registers, a r e pre- sented to the X and Y deflection circuitry of the CRT. Various methods are used t o accomplish this, from the charging of delay lines t o the use of opera- tional amplifiers as high quality integrators. Perhaps the simplest, but not necessarily best, method is to convert the AX and AY values (by D/A converters) to two square wavepulses whose amplitudes a r e proportional to the magnitudes of the numbers stored in the AX and AY registers respectively. Then these pulses a r e presented to the X and Y deflection circuits and the triangular wave- forms (whose forward slopes a r e used to deflect the CRT beam) a r e generated by the integration of the applied pulses through the inherent impedances of the deflection circuits. The main problem in this case is the lack of control over the integrating time constants, since these a r e determined by the structure of the deflection circuits. Also, unless the X and Y deflection circuit time con- stants a r e identical, amplitude compensation circuits must be provided after the D/A conversion of AX and AY. Normally, vectors generated by this method are excessively distorted. 9 A constant time method of vector generation that provides control of the integrating time constant and yet is fairly simple to implement will be discussed in a section below. The requirement of intensity compensation for constant time vector gen- erators generally involves the determination of the vector length, since the intensity modulation of the CRT beam is proportional to this length. Mathematic ally, I = k L where, I = intensity modulation required L = length of the generated vector k = parameter that depends on the CRT characteristics A s before, For the purposes of intensity compensation the following crude assumption can be made, So that, 1 2 k (AX t AY) This approximation will not provide perfect intensity compensation, but will considerably facilitate implementation. One method for intensity compen- sation is then, clearly, to pass the magnitudes of AX and AY through D/A con- verters, sum the result, and use it to modulate the amplifier that controls the intensity of the CRT beam. A t present the majority of the commerically available vector generators is of the constant time stroke variety. However, this is only a slim majority and, as component prices drop and the demand f o r display system increases, the constant velocity vector generators a r e becoming more and more popular. 10 A STROKE TYPE VECTOR GENERATOR In this section a fairly simple and easy to implement approach to constant time stroke type vector generation will be presented. Basically, this method consists of producing sawtooth waveforms whose slopes depend on the magnitude of the numbers stored in the AX and A'Y regis- t e r s by means of two digitally controlled sawtooth generators (DCSG's) (the de- sign of these DCSG's will be discussed in Appendix C). The resultant sawtooth waveforms a r e either inverted o r not, depending on the values of the sign bits in the AX and AY registers, and then passed on to the X and Y deflection cir- cuitry of the CRT. A block diagram of this system is given in Figure 2. The pulse generator will produce a timing pulse of fixed duration, starting at the beginning of the vector drawing sequence. In this time interval, the SIGN BIT SIGN BIT TO INTENSITY COMPENSATION CIRCUIT 11 11 - ANALOG GATE 4 ' v --j ANALOG GAT E - TO X-DEFLECTION TO Y-DEFLECTION OF DISPLAY UNIT OF DISPLAY UNIT Figure 2. 11 variable slope AX and AY waveforms a r e produced and the CRT beam is un- blanked. The time interval, which is essentially limited by the electrical inertia of the deflection circuitry, must be of sufficient length to allow the CRT beam to be deflected through the maximum vector increment. Each analog gate in Figure 2 is really a combination of 2 analog gates, both controlled by the logical state of the respective sign bit, allowing either the sawtooth o r its inverse to be transmitted to the CRT deflection circuitry. The logic gates indicated in Figure 2 were implemented (see Appendix C) using Packard Bell Transistor Logic Cards. If digital logic cards, operating between different voltage levels a r e used, then the base biasing resistors of Figure C-1 will have to be adjusted accordingly. The inverters shown in the functional diagram of the system (Figure 2) should be high quality DC amplifiers having a s small a DC offset a s possible. This can be achieved with a good grade operational amplifier, whose feedback loop is adjusted t o provide a gain of unity.* The analog gates required to switch either the sawtooth or its inverse into the deflection system of the CRT can be implemented using Diode Bridge, tran- sistor, o r FET switching methods given in Johnson** o r Korn and Kornt. Good quality analog switches a r e also commercially available at reasonable pricest t . The intensity compensation required can be of the simple summation type mentioned at the end of the last section and is shown in Figure 3. A s can be seen the modulation of the electron beam is only present in the vector drawing time interval a s specified by the vector pulse generator. This method offers the advantage of being able to display continuous vectors at high rates (being able to display a large amount of graphic information at one time). It can be readily applied in madmachine command and control consoles, where overall status is presented graphically and parameters requiring a high degree of accuracy a r e presented numerically, o r in time shared remote con- soles for the handling of mathematical problems, requiring the analysis of com- plex curves. 'BurrBrown Research Corporation, Handbook of Operational Amplifier Applications, pp. 8- 10 **T. E. Johnson, Analog Generator for Real Time Display of Curves, MIT Lincoln Laboratory, Technical Report 398, 28 July 1965 tGranine A. Korn and Theresa M. Korn, Electronic Analog and Hybrid Computers, McGraw Hill Book Company, 1964, pp. 219-233 t tPhilco, System 4-D Digital System Building _____ Blocks Applications Manual, SU-02, SU-04, SU-06, SU-03 analog Switch units specification sheets 12 w SIGN BIT AX REGISTER - "r' D/A CONVERTER D/A CONVERTER ' SUMMING CIRCUIT ' GATING PULSE ANALOG (FROM VECTOR PULSE GENERATOR) TO INTENSITY AMPLIFIER OF THECRT BEAM Figure 3. CONCLUSION It can be seen that vector generators, either of the incremental point posi- tioning o r stroke type, depending on the user's requirements can be readily con- structed. The incremental point positioning type vector generator can be im- plemented using any standard commercial digital logic line , requiring a minimum amount of system development. The stroke type vector generator, while probably needing less hardware requires more development, because of its hybrid nature, but has the potential of providing more effective utilization of the graphic capability of the CRT. 13 APPENDIX A Algorithm Gates for the Seven Bit Incremental Point Positioning Vector Generator A s was mentioned, when the functional diagram of this vector generator (Figure 1) was discussed, the AX and A Y Algorithm Gates basically extract point generating pulses from the clock counter states, depending on the values stored in the AX and AY registers. Since tihe operation of the AX and A Y Algorithm Gates is identical, only the A X Algorithm Gates will be described. The gating action will take place between the AX register and the clock counter, and the resultant point generating pulses a r e passed on to the X up/down counter. Let A = X up/down counter B = AX register c = clock counter Let the letters with subscripts indicate the condition of a bit stage of a given counter or register. Thus A, represents the first stage of the X up/down coun- ter. The convention followed here is that the largest digit is stored in the lowest numbered stage. Thus A, stores a digit that is twice a s large a s an equivalent digit in A,, and A, is least significant digit of the A up/down counter. B Register C Counter In the 3 register Bo will represent the sign of AX. Let ftO" represent a "+" sign and "1" represent a ti-tf sign, where t r + t c causes the X up/down counter "A" t o be set to count up and the t f - T t sign causes the X up/down counter A to be set t o count down. Basically, the higher digits of the AX register a r e logically "anded" with the lower digits of the clock counter and the resultant pulses a r e either added A-1 to o r subtracted from the X up/down counter, depending on the value of the sign bit in AX. For example, if the AX register consists of all r r l c ~ "every then clock counter pulse will pass through the Algorithm Gates and produce a point on the CRT. If Bl is "1" and the rest of the digits are all zero, then every other counter pulse is passed through the Algorithm gates and 32 points a r e produced. If any of the more significant bits in the AX register a r e "0's'' followed by cclrs" then we say that there exists a "gap" at the location of the first l e s s sig- nificant lfl". Example, 000110 t\ Gap position In this case all the counter bits up to the corresponding gap a r e "anded" with zeros and pulses are passed on to the X up/down counter, if an 7Tand/orff condition exists for all consecutive r r l r s r (in r the B register) following the gap. In this way the correct total number of pulses a r e passed on to the X up/ down counter (as will be illustrated by some examples in this appendix), pro- ducing the required vector increment AX. , ' Sign bit 6 For example if the AX register reads 0 0 0 0 110 then only the following clock counter states a r e passed through Algorithm Gates to produce point pulses: co CI c2 c, c, c, Decimal Equivalent of the clock coun- ter state 0 0 1 0 0 0 8 0 1 0 0 0 0 16 0 1 1 0 0 0 24 1 1 0 1 0 0 0 40 1 1 0 0 0 0 48 1 1 1 0 0 0 56 As can be seen, the 8th, 16th, etc. clock counts produce points. Of course, another point is produced at the starting location by the start pulse. Clearly, 6 incremental points a r e generated corresponding to AX = 6. A nice feature is the fact that the generated points a r e fairly evenly dis- tributed over the clock counter cycle, corresponding to a nearly uniform change A- 2 ' I in the X coordinate with time. Since the same procedure is followed by the Y portion of the vector generator, the requirement of a constant slope for the generated line is almost satisfied. Let us exprese the logical gating required in t e r m s of Boolean Algebra. The most significant bit of the AX register is "anded" with the least sig- nificant bit of the clock counter, the next most significant bit of AX with the next least significant bit of the clock counter, etc., providing that no "gap" con- dition ed&. If a TTgapTq is present then in order for a point pulse to be pro- duced, all the less significant bit positions of the clock counter up to the r*gap'' mmt be zero, and all tke following bits of the clock counter a r e "anded" with the bits of the B register, as before. is produced when any one of the following conditions A point pulse TcP,rc Bit configuration of the AX register (X here indicates that a given bit is either " 0 " or "1") Bl B2 B3 B4 B5 B6 1 1 1 0 1 1 1 1 0 1 1 X 0 1 1 1 1 1 0 1 1 1 X 0 1 1 X X 0 1 1 1 1 1 1 E5 0 1 1 1 1 - c5 X 0 1 1 1 c5 X X 0 1 1 c5 X X X 0 1 A-3 B1 B2 B3 B4 B5 B6 ‘0 1 1 1 1 1 1 ‘1 B2 B3 B4 B5 B6 ‘0 ‘5 0 1 1 1 1 1 - - x o 1 1, 1 1 B2 B3 B4 B5 B6 ‘0 4‘ ‘5 B3 B4 B5 B6 ‘0 ‘3 4’ 5‘ x x o 1 1 1 B4 B5 B6 ‘0 €2 €3 4‘ 5‘ x x x o 1 1 E, CO €2 €3 €4 c5 x x x x o 1 This expression represents the action of the X Algorithm Gates. It can be reduced to: Before actual implementation, this Boolean expression should be minimized by using a suitable computer minimization program. It must be remembered that identical Algorithm Gates must be provided for the Y portion of the vector generator. A -4 At this point it will be useful to consider some examples of vector generation by this method Bl B2 B3 B4 B5 B6 LetbX=20(Decirnal) = 0 1 0 1 0 0 where the primes indicate digit stages of the AY register. The following clock pulses will be passed through: AX AZgmithm Gates A Y Algorithm Gates 0 0 0 0 1 0 0 0 0 1 0 0 (4) 0 0 0 1 1 0 0 0 1 0 0 0 (8) 0 0 1 0 0 0 0 O * l 1 0 0 (12) 0 0 1 0 1 0 0 1 0 0 0 0 (16) 0 0 1 1 1 0 0 1 0 1 0 0 (20) 0 1 0 0 1 0 0 1 1 0 0 0 (24) 0 1 0 1 1 0 0 1 1 1 0 0 (28) 0 1 1 0 0 0 1 0 0 1 0 '0 (36) 0 1 1 0 1 0 1 0 1 0 0 0 (40) 0 1 1 1 1 0 1 0 1 1 0 0 (44) 1 0 0 0 1 0 1 1 0 0 0 0 (48) 1 0 0 1 1 0 1 1 0 1 0 0 (52) 1 0 1 0 0 0 1 1 1 0 0 0 (56) 1 0 1 0 1 0 1 1 1 1 0 0 (60) 1 0 1 1 1 0 1 1 . 0 0 1 0 14 pulses pass through the 1 1 0 1 1 0 (54) AY Algorithm Gates 1 1 1 0 0 0 (56) 1 1 1 0 1 0 (58) 1 1 1 1 1 0 (62) 20 pulses pass through the AX Algorithm Gates A-5 Each one of these pulses produces a point on the CRT. The result is shown in Figure A-1. Figure A-1. For a second example B~ B2 B3 B4 BS B6 let AX = 18 (Decimal) = 0 1 0 0 1 0 B;' B\ BS Bb B; Bh AY = 5 (Decimal) = 0 0 0 1 0 1 A-6 the following clock pulses will pass through AX Algorithm Gates AY Algorithm Gates 0 0 0 0 1 0 ( 2) 0 0 1 0 0 0 (8) 0 0 0 1 1 0 ( 6) 0 1 1 0 0 0 (24) 0 0 1 0 1 0 (10) 1 0 0 0 0 0 (32) 0 0 1 1 1 0 (14) 1 0 1 0 0 0 (40) 0 1 0 0 0 0 (16) 1 1 1 0 0 0 (56) 0 1 0 0 1 0 (18) 0 1 0 1 1 0 5 pulses pass through the (22) 0 1 1 0 1 0 (26) AY Algorithm Gates 0 1 1 1 1 0 (30) 1 0 0 0 1 0 (34) 1 0 0 1 1 0 (38) 1 0 1 0 1 0 (42) 1 0 1 1 1 0 (46) 1 1 0 0 0 0 (48) 1 1 0 0 1 0 (50) 1 1 0 1 1 0 (54) 1 1 1 0 1 0 (58) 1 1 1 1 1 0 (62) 18 pulses pass through the AX Algorithm Gates The resultant vector is shown in Figure A-2. A s can be seen from the two examples, the more nearly the two numbers stored in the AX and AY registers agree in magnitude, the better will be the approximation to a straight line, and vice versa. It should, however, be kept in mind that the two illustrations a r e highly magnified in size and that on the CRT screen the deviation from a straight line would be hardly noticeable. If we take the CRT specifications to be the ones given at the beginning of this discussion, the maximum length of our vector corresponding to 64 points would be 10 /2/16 inches % 0.885 inches. The magnitudes of the vectors in our two illustrations would be correspondingly smaller. A-7 A X (IN RASTER COORDINATES) Figure A-2. A-8 APPENDIX B AdaDtation of the Incremental Point Positioning Vector Generator to the Production of Larger Vector Increments The vector generator discussed in Appendix A can produce vectors corre- sponding to 7 bit increments AX and AY (6 bit magnitude plus the sign bit). In order to extend this generator to 8 bit increments the method of arriving at the required logic gates in Appendix A must be re-examined and extended. The AX register in this case will be the 8 bit B register with Bo again rep- resenting the sign of AX. The clock counter C will have seven stages: B Register Clock Counter Again the 1 megahertz clock will drive the clock counter, in this case through 128 steps, beginning with the arrival of the "start" pulse. A s in Appendix A, a point pulse " P X " will be passed on t o the X Up/Down Counter when one of the following conditions exists: B i t c o n f i g u r a t i o n o f AX r e g i s t e r Bl B2 B3 '4 B5 B6 "7 Bl 6' 1 1 1 0 1 1 1 1 0 1 1 x o 1 B-1 B-2 A s i n Appendix A, this Boolean expression should be minimized by a computer minimization program before implementation. Again - identical Algorithm Gates must be provided for the Y portion of the vector generator. By comparing the Algorithm Gates given in Appendix A and Appendix B, it becomes easy t o generalize this vector generator for any number of bit position of the incre- mental registers AX and AY. Thus any vector length can be generated by this method provided that sufficient logic circuitry is found in the system. The advantage gained in being able to draw longer vectors with a single set of instructions in this extended version of the incremental point positioning vec- t o r generator appears to be offset by the fact that the clock counter now has to cycle through 128 counts for all vectors, o r that the vector drawing time has been increased to 128 psec., regardless of vector length. One way to overcome this disadvantage is to have the clock counter cycle through a smaller number of counts for short vectors than for loilg vectors. For example, if the most significant bits (not the sign bits) of the AX and AY registers are both zero, then the clock will drive stage C5 (instead of c6) of the clock counter. Thus the clock counter will only cycle through 64 counts, since its least significant stage is bypassed. B-3 The following terms will produce counts through the Algorithm Gates in this configuration: Bit configuration of AX r e g i s t e r Bl B2 B3 B4 B5 B6 0 1 0 1 1 x o 1 0 1 1 1 x o 1 1 x x o 1 0 1 1 1 1 x o 1 1 1 x x o 1 1 x x x o 1 - - 0 1 1 1 1 1 B1 BZ B3 B4 B5 B6 ‘1 ‘6 - - x o 1 1 1 1 ‘2 B3 B4 B5 B6 ‘1 ‘5 6‘ - x x o B3 B4 B5 B6 C, - c6 c4 -c5 - 1 1 1 - x x x o 1 1 B4 - B5 B6 ‘1 - _4‘ _‘5 _‘6 ‘3 x x x x o 1 B5 B6 ‘1 2’ ‘3 ‘4 ‘5 ‘6 0 1 1 1 1 1 1 x o 1 1 1 1 1 x x o 1 1 1 1 x x x o 1 1 1 x x x x o 1 1 x x x x x o 1 B-4 Since B, and c6 a r e both always zero in this case B, and E 6 will be logical "1" and can be ignored. If the terms of the B register a r e relabeled s o that B, corresponds to B, of Appendix A, B, corresponds to B,, and so forth, then we see that the pulse producing terms here correspond to the ones found for the 7 bit AX register of Appendix A. In summary, if the most significant bits of AX and AY a r e both zero, then the clock pulses a r e routed to the next least significant stage of the clock coun- ter. If the most significant bits of AX and AY a r e not both zero then the clock pulses a r e fed to the least significant stage of the clock counter, which then cycles through 128 counts. This modification allows the system to draw short vectors in 64 psec and longer vectors in 128 psec. There is no reason why this procedure should not be applied to all stages of the AX and AY registers. If the more significant bits of both registers a r e zero, then the clock pulses will bypass a number of least significant stages of the clock counter equal to the number of more significant bits that a r e zero in both AX and AY registers. This will allow a variable vector drawing time, depending on the vector length, and thus a more efficient production of line segments. B-5 APPENDIX C Ditzitalls Controlled Sawtooth Generator (DCSG) The Digitally Controlled Sawtooth Generator (DCSG) can be used to produce a sawtooth voltage signal whose slope is controlled by the number stored in a given register. This approach is based on the fact that in an RC sawtooth gen- erator the slope of the voltage output waveform depends on the time constant of the system. By digitally changing the value of the charging resistance one ob- tains a proportional change in the output waveform. The circuit used is shown in Figure C-1, where the incoming pulse turns off the normally conducting transistor Q, and turns on transistors Q2 and/or Q3 ............. Q9,depending on the condition of the flip-flops in the AX or AY register. At the end of the pulse all transistors Q2.. ........ Q, a r e cut off and Q, is turned on providing a discharge path for the capacitor C. The DCSG was implemented using Packard Bell Transistor Logic Cards for the digital portion. Transistors Q2 to Q9 were 2N703's, and Q, was 2N995. These transistors were chosen mainly for the low value of their leakage cur- rents. Base bias resistors R, and R,, were selected at 300 and 510 ohms re- , spectively. R, was 9.1 KR and R, was 3.6 KQ. The charging resistors R, , R, .......... R8 should form a binary sequence corresponding to the binary weighting of the different bits in the AX or AY register. The actual values chosen f o r the breadboard model of this system depended on the availability of resistors in our laboratory (without having to form resistor combinations) and a r e shown in the following table together with the peak output voltages of the generated sawtooth. Magnitude Peak Output Voltage 470 R - 6.4 V 910 R - 3.7 V 1.8 K R - 1.85 V 3.6 K R - 0.86 V 7.5 K R - 0.42 V 15 K R - 0.21 v 30 K - 0.092 V 62 K R - 0.04 V c-1 - - 0 0 0 R j j ! Rg)! R2ji Rgj: Y zz c 4 VECTOR PULSE - 0 sw1 sw2 GENERATOR i ,- 0 0 0 SIGN BIT Figure C-1. All of the above resistors were of standard (596) quality and the voltage meas- urements were made on a Hewlett Packard 175 A oscilloscope. The capacitor Vqt was chosen so that the sawtooth pulse could reach the largest distortion free amplitude in the pulse interval. The pulse interval was somewhat arbitrarily placed at 11.6 microseconds and for this interval the value of the capacitor was 0.015 pf. Other time intervals could be selected, probably requiring a change in the value of "C". The inverters after the logical "AND" gates in Figure C-1 and the values of the biasing resistors R, and R,, were chosen to be compatible with the Packard Bell Logic Levels, where the logical "1" is -10 volts and logical zero is 0 volts. c-2 Originally it w a s intended to construct a DCSG that would produce a variable slope sawtooth corresponding to 9 bits in the A X o r AY register. However, at an early experimental stage it was found that the unshielded circuit was subject t o noise levels on the order of magnitude of the amplitude of the sawtooth produced by the least significant bit in the register. It is a matter of conjecture at this point if the DCSG as it stands could have been refined by providing a cleaner pulse source, mounting the system on a card with the ac- companying shortening of w i r e leads, appropriate grounding, etc., to operate as a nine bit system since the DCSG never went beyond an early breadboard stage. However, it is felt that the DCSG could be made t o perform reliably as an eight bit system. The linearity of the output waveform was very good over this wide range of slope variations. The sawtooth start and stop points were well defined. A small amount of D.C. offset is present at the output due t o the action of transistor Q1 , manifesting itself as a constant amplitude pulse of about 0.01 volts when zero is stored in the A X and AY register and the vector pulse is applied. This can be overcome by the addition of some logic gates, which would allow the sawtooth waveform or its inverse to be transmitted through the analog gate, only if a non-zero value is stored in the AX o r AY register. The method used to change the resistance of the charging circuit, and thus the slope of the output sawtooth, requires that a binary sequence of precision resistors R , 2 R . . ... 128R be placed in series with the switching transistors, as shown in Figure C-1. It is often difficult to obtain precision resistors over this wide range of resistance values. However, it is possible t o substitute an R/2R binary ladder* for the charging resistors of Figure C-1. The modified DCSG is shown in Figure C-2, where the charging network now only requires two different high precision resistors R and 2R. Transistors Q1 and Q2 should be complementary. They are required at each node of the ladder network, since for proper operation either zero volts o r the charging voltage (-12 volts) must be present at each node point. When a given stage of the AX o r AY regis- ter is r'O" then zero volts are switched in at this node; when a given stage is "1" then the charging voltage (-12 volts) is switched in at the corresponding node. Otherwise the operation of this system is identical to the one shown in Figure C-1. "Alfred K. Susskind, editor, Notes on Analog Digital Conversion Techniques, Technology Press of MIT and John Wiley and Sons, Inc., 1960, pp. 5-32 - 5-35 c -3 -1 2v +I 2v I R R 2R 4 + - 0 00 Figure C-2. c -4