HHH HHll BBS S3! HI hBw£s ■ ' I ■ 23 rSrSS ■ &-*-%■* * •f!*- ■ KMK ■ LIBRARY OF THE UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 510.84 l£6r no. 480-489 cop. 3* 3 UIUCDCS-R-71-486 f r COLORAMATRIX. A THERMALLY CONTROLLED LIQUID CRYSTAL ALPHANUMERIC DISPLAY by STAVROS ALEXANDROS HADJIS TAVROS June, 1971 UIUCDCS-R-71- i +86 COLORAMATRIX. A THERMALLY CONTROLLED LIQUID CRYSTAL ALPHANUMERIC DISPLAY by STAVROS ALEXANDROS HADJIS TAVROS June, 1971 Department of Computer Science University of Illinois Urbana, Illinois 6l801 This work was supported in part by Contract No. N000 14-67-A-0305-0007 and was submitted in partial fulfillment of the requirements for the degree of Master of Science in Computer Science, June, 1971* Digitized by the Internet Archive in 2013 http://archive.org/details/coloramatrixther486hadj Ill ACKNOWLEDGMENTS The author wishes to thank Professor W. J. Kubitz for suggesting the basic idea investigated in this thesis. He al- so acknowledges the useful discussions with and the encourage- ment received from both Professor Kubitz and Professor W. J. Poppelbaum, principal investigator of the contract under which this work was carried out. He would like to thank Miss Barbara Weeks for the typing and extend a special thanks to the fabrication group under Mr. Frank Serio for their patient efforts. IV TABLE OF CONTENTS Page 1. INTRODUCTION 1 2. LIQUID CRYSTALS . . . 5 2.1 Types of liquid crystals 35 2.2 Smectic liquid crystals ' 4 2.3 Nematic liquid crystals 4 2.4 Cholesteric liquid crystals 6 2.5 Encapsulated liquid crystals (E.L.C.) 11 5. LIGHT SOURCES 15 4. THERMAL ELEMENTS I7 4.1 Thermal considerations I7 4.2 Transition times 20 4.3 Dissipation of heat 21 4.4 Basic calculations 22 4.5 Thermal bias 29 5- ELECTRONIC CIRCUITRY J>0 5.1 General 30 5-2 Input system 25 Page 5.3 Memory 38 1 }A Character selector 58 5.5 Character generator 39 5.6 Clock and timing circuitry 40 5.7 A different thermal design 44 5.8 Drivers 46 5.9 Temperature control system 49 5.10 Power supplies , 52 6. CONCLUSIONS 5^ LIST OF REFERENCES 56 VI LIST OF FIGURES Figure Page 1. Smectic structure of L.C 5 2. Nematic structure of L.C 5 3. Cholesterol 5 k. Stereochemical module of Cholesterol .5 5. Cholesteric structure of L.C 7 6. Helical structure of cholesteric L.C 7 7. Circular dichroism 9 8. Response curve of cholesteric L.C. . 13 9- Switching mode of transistor .... 18 10. Thermal model of transistor 18 11. Ceramic microtransistor 23 12. Pellet resistor 23 13- A part of the matrix 23 lh- Deration diagram 25 15- Matrix of microtransistors 31 16- Block diagram of the electronic circuitry 3^ 17- Input system, memory and character selector J>6 Vll Figure Page 18. Clock and timing circuitry 41 19. Modification of the Clock and timing circuitry 45 20. Horizontal driver Vf 21. Vertical driver 47 22 . Temperature control system 50 23. Variable voltage power supply 53 2 4 . -5 volts power supply 53 Vlll LIST OF TABLES Table Page 1. Thermal transitions of L.C. and required power 27 1. INTRODUCTION The basic idea employed by COLORMATRIX involves using the thermal properties of cholesteric liquid crystal s in or- der to display information in color. Cholesteric liquid crystals have the property of selectively reflecting differ- ent wavelengths of incident white light as their temperature is changed. This phenomenon is reversible and the spectrum of the reflected light is a function of both the chemical composition and the temperature. Since liquid crystals are light reflecting devices,, low power consumption and high con- trast are their main advantages; however, their reflectance is low. Thermal elements , assembled in the form of a two- dimensional matrix, are used in order to generate the ap- propriate temperature and optical patterns. Any heat generators, such as resistors, transistors or diodes, can be used as the thermal elements. The present display uses transistors. An alpha-numeric display has been constructed to illustrate that this type of display is possible. Howev- er, for an efficient display, the isolation of the thermal elements which are, of necessity, in close proximity, repre- sents a difficult problem. The present study can be divided into the following parts : 1. The properties of the cholesteric liquid crystals which determine the selection of the thermal ele- ments and the associated electronic circuitry. 2 . The properties of the thermal elements both as iso- lated elements and as elements in the matrix. ^. The design and implementation of the electronic circuitry . 2 . LIQUID CRYSTALS 2 . 1 Types cf liqu id crystals Liquid crystals are generally organic substances which, in a specific temperature region, possess many of the properties of a crystalline solid. Most chemical com- pounds are characterized by a well-defined melting point. However, the melting point for liquid crystals (L.C.) is not well defined since prior to their real melting point they pass through an intermediate state (turbid and viscous or mo- bile) between solid and isotropic liquid called a mesophase . That is T T SOLID (crystals) — * MESOPHASE — * ISOTROPIC LIQUID It is this mesophase which is of interest. A more detailed representation splits the mesophase according to the following scheme: T, T, ' NEMATIC MESOPHASE T p — > SMECTIC MESOPHASE — — * OR — ^-> CHOLESTERIC MESOPHASE Some of the above transitions may be present or ab- sent depending on the chemical structure of the L.C. These transitions are generally enantiotropic although complicated mono tropic transitions are known, and they take place at the sam e temperature. Of prime interest to us is the cholesteric mesophase which is characteristic of cholesteric liquid crystals , de- rivatives of cholesterol. However, because of the interre- lations between the three classes, the smectic and nematic "J iquid crystals will also be described briefly. 2 .2 Smectic liquid crystals In this mesophase the rod-like molecules are arranged with their long axes approximately parallel but also strati- fied in planar layers (Figure 1). These layers cause the substance to exhibit the properties of a two-dimensional (2 k) liquid since they are free and can slide over one another. v ' ' 2 .J> Nematic liquid crystals In this case we again have rod-like molecules but bi- lateral branches are also present. They keep their long axis approximately parallel but there is no stratification in lay- ers (Figure 2). This kind of parallelism is attributable to the mechanical and electrical forces between the molecules which generally exhibit a dipole character.^ ' III! ' J! mill j! 1 ! 1 1 1 : I i ; t Hill :i:ii!i;Miiiiiii:;;ii:!!ii i ill 1 1 1 1 1 j ml illii Figure 1. Smectic structure of L.C, Figure 2. Nematic structure of L„C, R-cofoiTiTlo Figure 3. Cholesterol CH3 22 21\20 r-co |qhh] o s Figure 4. Stereochemical module of Cholesterol 2 A Cholesteric liquid crystals These are usually esters or derivatives of the cho- lesterol. A pronounced long axis is present. However,, lat- eral branches are of prime importance because either they exhibit permanent dipole monents or they are easily polariza- ble. The lateral chains are above or below the average ring level of the molecule. The long lateral branch, CI7-C27, as indicated in Figures 3 and k, is downward relative to this average plane so that two molecules, one below the other, cannot have the same space orientation. This and the fact that the molecule itself exhibits a well-defined dipole mo- ment (not generally parallel to its long axis) produces a helical configuration in the aggregation of the cholesteric L.C. molecules (Figure 6). These molecules are arranged in layers as in the smectic L.C. but within each layer they have the nematic structure. Their long axis changes from layer to layer (for the reason given before) in an additive way (Figure 5) • The cholesteric L.C, because of the previously de- scribed properties, are optically active ; also they exhibit the phenomenon of double refraction . When unpolarized light is directed at the surface of a cholesteric L.C. preparation, two different rays are produced; one reflected and the other transmitted. The electrical vectors of these light compon- ents are rotating clockwise and counterclockwise depending on ml IliiiUrl Figure 5. Cholesteric structure of L C« Distance between ^ Planes I Angle between Successive Long Axis Pitch of the Helix P Figure 6. Helical structure of cholesteric L.C. 8 the chemical structure of the cholesteric L.C. (Phenomenon of c ircular dichroism ) (Figure 7). ; The reflected part of the light is responsible for the iridescent colored appearance of the cholesteric L.C. Generally,, the intensity of the light of the reflected beam is a_ small part of the incident white light . Therefore, cholesteric L.C. are poor light reflectors. The color observed by reflection depends upon the an- gles formed by the incident and reflected rays with the normal to the L.C. preparation. and A are approximately related^ ' by the BRAGG equation: 2dsin9 = nA which is indicative of the layered structure. The spacing between these layers is calculated to be 0.2-0.8 \ebl. That is,, it is of the order of the wavelength of visible light. A model proposed by C. W. OSSEElT ' for the geometry of the aggregation of the cholesteric L.C. is the helical structure (Figure 6). According to this model, an analysis of the phenomenon of the circular dichroism gives the result that the wavelength A of the scattered light (maximum scat- tering) in the material is equal to the pitch of the helical (3) structure. K ^ ' Therefore: P = nd , n = 2g0f. and p = ^60^ = A White Light Irridescent Reflected Light Wavelength, X Transmitted Light Figure 7. Circular dichroism 10 When external energy is delivered, the weak molecular forces are disturbed and so the pitch of the helical structure of the cholesteric L.C. is changed. This changes the wavelength A and thus also the color of the reflected light. This ener- gy can be delivered as thermal, mechanical or electrical en- ergy; in the latter case the electric field interacts with the dipole moment of the molecule. For this transformation, which is generally reversi- ble, a very small amount of energy is required. ^ ' From x-ray diffraction measurements it is calculated that the av- erage molecular thickness of a cholesteric L.C. molecule is approximately J>A . Therefore, this is the average distance, d, between the layers of the helical structure. In order to cover the visible region of the electromagnetic spectrum (4000-7000 A°), has to be changed by 1 -0 2 - = d3 60°[^-^] or = 310" 8 560° I —1 - — 1 ] L 410 3 710"^ where io"5 1 = 4.70 1U radians 2 = 2.70 10 radians and = 0.116° or = 0.002 radians. 11 If a force moment M = aMt (where a -- area of the lay- ers and k = force constant) is exerted between two successive ' 01 , " ' change of p and is E = Md0 = ^t0 2 K ^0o 0, layers of the helical structure, the energy E required for a ! 1 From magnetic '2 y 2 resonance measurements for p-azoxian isole it is calculated that K = 210"' dynes.' ' If we take this order of magnitude as being valid for the cholesteric L.C.,. then E - ^K(0 1 2 -0 2 2 ) , and for a = 1cm 2 E = — i-n2.10" 7 [(4.70) 2 10" 6 - (2.70) 2 10" 6 ] 6l0" b and E = 4 . 90 10" 5 ergs, or E =4.90 10" 12 Joules. Therefore, the display of information utilizing different colors can be controlled by minute amoun t s of energy . The response time of these transitions is also of in- terest. Response times of 10-500ms are reported for choles- teric L.C. in electrical fields. The thermal response is of the same order. ' Therefore, relatively fast display of information is possible . 2 . 5 Encapsulated liquid crystals (E.L.C.) Cholesteric L.C, because of their chemical struc- ture, are sensitive to oxygen. Also, they undergo perman- ent transformations when exposed to U.V. radiation and va- rious chemical vapors. Another disadvantage for display 12 purposes is the dependence of the observed color on the angle. 0, of the observed light. "Encapsulation" techniques offer a solution to these problems. The L.C. are enclosed in small capsules (4-30u. diameter^') so that they are protected from oxygen^ U.V., chemical vapors and dust. The encapsulation also yields a pseu do-solid such that the angular dependence of the observed color is mostly eliminated. Because the light scattered by the cholesteric L.C. is only a small portion of the incident light while the larg- est portion is transmitted through the L.C, a black back- ground is needed in order to absorb this transmitted light. This improves the purity and the resolution of the reflected light. The temperature range of the cholesteric E.L.C. is -20° to 250 c for the visible region of the electromagnet- ic spectrum. This range can be modified in a predictable way by changing the proportions of the different components of the cholesteric L.C. mixture. A general diagram indicating the relation between A and T is as indicated in Figure 8. A non-uniform distribution of colors in the spectrum exists since A is generally a non-linear function of T. This favors some colors for uniformly changing T and also affects the purity of the colors which appear. The energy and the time required for the transitions to occur are approximately pro- portional to the change in temperature required for the tran- sitions. Also for the wider temperature E.L.C.'s because the 13 T t Temperature Threshold—* Temperature Viol Blue Grn Yel Red Wavelength, X — * Figure 8. Response curve of cholesteric L.C< H time required to cover the whole spectrum is longer,, the ob- server is more likely to see intermediate colors during the transition. Therefore, E.C.L. of narrow temperature range are preferred. Because L.C. displays are based on light reflection, the contrast is high, and they can be viewed under a wide range of lighting conditions. However, the reflected light is of low intensity. Resolution for cholesteric L.C. of 200 lines/cm are reported for the thermal mode. ' 3. LIGHT SOURCES Up to this point it has been assumed that the choles- t.eric liquid crystals were illuminated by a light source pos- sessing characteristics closely that of sunlight so that none of the colors reflected by the L.C. will be favored due to the light source used. However, this is not actually the case. The light of commonly available light sources differs from that of sunlight. In every case some of the colors in their spectrums are favored over others. Thus, the color of the light reflected by the L.C. is directly affected by the light source used. Moreover, the eye is more responsive in the 5000-5500 A green-yellow region. In the case of incan- descent lamps the relative amount of emitted energy (as indi- cated by their spectral energy distribution curves) increases monotonically toward the red and infrared radiation portion of the spectrum. Therefore, the light reflected by L.C. ap- pears somewhat stronger in the red region, accompanied by a considerable increase of temperature of the L.C. For fluor- escent lamps the green and blue colors are favored. As mentioned the reflectance of the L.C. is low. Thus in order to have a good optical result, we actually need a strong light source. In this case (particularly for 16 incandescent lamps) it is inevitable that the infrared radia- tion will be strong enough to saturate the L.C. matrix with heat to the extent their thermal threshold is shifted and the displayed patterns become confused. The use of an infrared absorbing filter does not completely eliminate this problem. 17 k. THERMAL ELEMENTS ^ • 1 T he riTi n 1, c o n s 1 d e r a t 1 on s For the generation of the required temperature pat- terns, the area of the display is divided into small elements which are activated independently. The resolution is in- versely proportional to the cross section area of the indivi- dual element, assuming that each element remains "thermally" independent (i.e. not affected by the conditions of its neighboring elements). These elements can be transistors, resistors, etc. In the case of transistors the heat is gen- erated mainly in the base-collector junction. The common base configuration (Figure 9) offers a simple way to control the corresponding power P. For pulsed operation we have P = f(I F ,V rR ,D), where D is the duty cycle, and when D = 1. the generated heat if given by: (VV^ + (W^ or approximately by Wc A simple and adequate model for the thermal behavior of a transistor is as indicated in Figure 10. ' In this 18 Vc V C Re Vcc -o Ic i Figure 9. Switching mode of transistor Collector -Base Case Junction j #JC C ^CA f WvWV f VWAV- © + Cj ♦ T A 4=c c £jc,0CA,Thermal Resistances in °C/W Cj,Cc i Thermal Masses in J/°C Figure 10. Thermal model of transistor 19 thermal model, in strict analog)? with electronic circuits, the product C,0 Tr - t t , is the the rmal time constant of the junction. Because we are interested in the thermal behavior of the transistor as a whole (including its case),, we can ap- proximate a thermal constant for the transistor as *— 'rn^rp I rp For square wave power pulses of period T two important cases must be considered: 1. When T « t t In this case there is insufficient time for the heat introduced to dissipate. Therefore, the temperature of the case increases as though the average power were being deliv- ered statically. The average temperature approaches asymp- totically the final equilibrium temperature. 2 . When T ~ t , In this case the change of temperature of the case can be significant during each pulse because there is enough time for the dissipation of the introduced heat. The result is an obvious temperature increment for each pulse as well as a continuous overall increase of the average temperature to- ward to the final equilibrium point. Therefore, thermal elements with small t t are appro- priate for fast displays. In order to have an optical result 20 free of flicker the power pulse must satisfy the first of the above conditions thus, T V EB0 = ^ volts 23 Ceramic H .-' Case 0.036" L_ |--^--ZT7^^_r-d 0.025' / E C B ' Epoxy Cover Figure 11. Ceramic microtransistor Figure 12. Pellet resistor Screen Phenolic Board Aluminum Thermal Plate Microtransistor Copper Screen 3 ' 16 " JO O O O O O !°-' 00 F^er-istor Figure 13. A part of the matrix 24 From the deration diagram (Figure 14) 9 TC and Tj are cal- culated as e Tr = i = 9l0°c/w and S = max Tn , To = 25°c soT T = I61°c. Jmax Experimentally it is found that in order to change the temperature of the ceramic case of the transistor by I6°c (from 25°c to 41°c), 70mW are required. This allows one to calculate 0, 'CA" o T C " T A = T = e CA P and 6 CA " ^§W = 22 5° c / w The 9_ r , 9^. thermal resistances permit the quick estimation of the power requirements for temperature transitions. In order to change the temperature of the Junction by 1 c 1° T T = 1 c = (0 Tr +9 rA )P and P = — ±~ = 0.88mW J JL LA 11J5 c/w However,, for the case, a larger amount of power is required because 6 C » is much smaller than 9 TA . Using E.L.C. of the range 35 -36 c this amount is determined to be: P, R = s 2 - - a ~P = 44.4 mW ^ 6 CA 225°c/w P ,- = gl- = ?? -25 = 48.9mW and P = 4.5mW ^ b B CA 225 c/w Tn T Smar. Figure 14. Deration diagram 25 26 For safe operation of the transistor,, the maximum temperature of operation of the case is estimated to be: T C " T A = l50mW ' 22 5° c / w = 53.7°c where T A = 25°c and T cmax = 5 8.7°c For the E.L.C. used in the form of a "paper" (range 35° to J>6°c) , the relationships which exist between the col- or transitions and the temperature differences T are as in- dicated in Table 1. The required power P for these transitions is estimated as above and verified experimentally in the case of the 2N]5l28 microtransistors. These values of P as well as the experimental values for the pellet resistors are tabu- lated in Table 1. For reasonable reliability and for low complexity and cost of the associated electronic circuitry, X-Y addressing and a "time sharing" technique were adopted for the alphanumeric display . The required energy for a thermal transition is given by W = P-t Joules. For pulsed operation with a duty cycle D, during each period T energy is delivered for DT sees. Therefore the time, t, of delivering energy during 1 sec is t = DTf , where f = ~ and is thus t = D sees For the transistor . P = O^t, and V '.:"'..:£ V CB " I C D 27 MICROTRANSISTORS PELLET RESISTORS TRANSITIONS T 2N2128 R = 250 TO P V CB P V R RED 0.25°c 1 . 56mW 5-7v 7 . 48mW 8.65v YELLOW 0.45°c 1 . 99mW 7.25v 9.20mW 9-70v GREEN 0.725°c 2.25mW 11.65v 11.65mW 10. 8v BLUE 1 . 000°c 4.20mW I5.lv l5.20mW 12 . 35v VIOLET >1.000°c >5.00mW >18.2v >2 1 . 50mW >H.60v Table 1. Thermal transitions of L.C. and required power V R and V R " \/¥ 28 For the resistor , P using the previous formulas, a quick calculation of Vpg, I I„, V DJ I D can be made in order to specify the requirements for the R R driving circuitry of the matrix. For the display of eight 5x7 modules using horizontal "strobing" the required duty cycle is' D = l/hO. The minimum 1^ required for the 2N3128 microtransistor (V nT>r » = 20V ) considering a thermal transition to the violet (W - 5 mjoules) is X C = V^D and X C = 10mA A value of I r = 11mA is selected. An estimation of Vp R and V R for the previous temperature and color transitions , as verified experi- mentally, are tabulated in Table 1. A quick estimation of the thermal time constant t t can be made by feeding the thermal element with variable frequency square wave power pulses and noting the frequency for which the flicker of the color of an E.L.C. indicator just becomes perceptible. The pe- riod corresponding to this frequency is of the same order as the Tm of the thermal element. It is found that for the 2N3128 microtran- sistors, T = t t = 330 ms, and for the pellet resistors t t = 250 Thus, t = t f = 2.2t t = 725 - 550 ms; this limits the speed of the display of information by such a hybrid matrix to 1.5 - 1.1 sec per character. 29 h . 5 Ther m::! bias In using E.L.C. of the range 35 *"36 c for fast display of information we must keep the liquid crystals "thermally" biased very close to their threshold temperature of 55 c. In this way we save precious time and driving power since it is then not necessary to travel the unreasonable distance of 35 ~2 5 c for every thermal transition. This thermal bias can be implemented: 1. By biasing each element separately. The required power must be adjustable for each element independently because of the existing differences in their thermal parameters. 2. By a temperature control system for the whole matrix of the el- ements. For the first method and to a lesser extent for the second method adjustments are required when the ambient temperature changes. In the second case, because the thermal biasing system plays the role of a reference temperature system, the heat re- leased during the pulsed operation of the thermal elements must be small, so that the temperature may be maintained constant. This fact and the requirements of speed and uniformity of thermal be- havior of the thermal elements, which are very important for dis- playing information, suggest an integrated type of structure rather than the present hybrid-type matrix. 30 5. ELECTRONIC CIRCUITRY 5 . 1 Gene ral The properties of the liquid crystals as well as those of the thermal elements directly influence the kind of the display and., therefore, the associated electronic circuitry. An alphanumeric display formed of a 5 x 7 array of elements has been constructed using the 2N3128 ceramic micro- transistor. The elements of this display matrix , X-Y addressed, are connected as indicated in Figure 15- A va- riable voltage source (^-I5v) connected to the collectors of the transistors covers the total range of the liquid crystal colors. The mechanical construction and the appropriate dim- ensions of the matrix are indicated in Figures 1J> and 15 . The X-Y addressing scheme offers the following advantages : 1. Simplicity of the matrix itself as well as of the corresponding electronic circuitry. 2. The possibility, by using "time sharing'' techniques, of sharing the same character generator (R.O.M.) for several display matrices. This also leads to simpler and more reliable electronics. 31 R El pgr *-w»- E2 P?Z r^- > o • c o o X r^:: ■ ♦ wv- • -^^w- -*— «^v- E 6 r^r r^z ~^_ Bi B2 ~^r r^r* r©~" r€r^ rt r^: rfc rr r¥: r^_ Vertical Drivers • • B 3 • ~^z' r^z' p?z z?z ~^~' r^~' r©~' p?z r©-; Z¥l B4 ii + V p "^z n?z r^z 1 r^z' r^H B5 ^ ^_ E 4 - 0.800 — E3 E2 El 0.160 ■ a a m~ B □ B ■ IB1I IBBI ■ bi a ■ 3 U E3 a B n □ e___ +V Bi B2 E7 B3 B4 E6 B5 E5 Figure 15. Matrix of microtransistors 22 In the present case there are two kinds of time shar- (11) ing — " strobing — techniques: v ■L* He r i.7. e n ta 1 s tr ob ing . In this case the character gener- ator "scans" the 5 x 7 matrix horizontally . When the charac- ter generator accepts six bits of information (in US ASC II code) at its six inputs for a definite time t, it generates sequentially at its seven outputs five equally spaced groups of pulses each of duration t/5. Using horizontal scanning the character is displayed from left to right. Because five descrete fields are displayed sequentially the repetition rate must be kept high enough to avoid flicker. For n successive display matrices and repetition time 1 , each character lasts t = T/n and, therefore,, each of its five columns T/5*i. Thus the duty cycle D for each column is (T/5n)/T = l/5n. The required current is inversely propor- tional to D for the same average delivered power so T P eak = W and thus I peak = ^V 5 *' It: is advanta g e " ous to use microtransistors possessing high BVp R0 so that smaller currents are possible. This allows the use of simpler and less expensive drivers. In the present matrix BV CBO = 20v ^ -^cmax = • L00mA anci ' f° r tne violet color transition, p AVERAGE = 5-OmW. Therefore, Average = 20 = °- 25 ^ and I peak = 100TnA = 0.2 5mA- 5n. This means n = 80 is the upper limit on the number of matrices which can be served by the same character generator. 53 £ . Vertical strobing . The character generator scans the 5x7 matrix vertically from top to bottom. In this case it sequentially produces seven equal length (t/7) groups of pulses at its five outputs. If the repetition time is T, each character lasts t = T/n and each row T/711. The duty cy- cle of each row is thus (T/7n) /T = l/7n. This method introduces more severe limitations on the number of displayed characters. For this case 7n = n PR A and n = 57. Thus, horizontal strobing is selected with a to- tal capacity of eight matrices. The electronic circuitry associated with the L.C. display can be divided into six parts plus the temperature control system and the power supplies (Figure 16): 1. Input system . The generation of six bits (in US ASCII code) corresponding to each character takes place here in a serial form and is transferred to the memory in a parallel form. Signals enabling the memory, character selector, R.O.M. and the drivers are generated here also. 2. Memory . It accepts six bits of information in descrete groups and stores them for temporary or permanent display. Therefore, it consists of two successive groups of memory elements . 3. Character selector . It consists of six 8:1 multiplexers which sequentially select the stored characters. The clock and timing circuitry deliver the appropriate selection signals 34 Character Selector Character Generator ( R.O. o Clock and I iming Power Supplies Temperature Control System r Vertical Drivers Horizontal Drivers Matrix No. 1 Matrix No. 8 Figure 16. Block diagram of the electronic circuitry 35 k. Clock a nd timing circuitry . A pulse generated by a master clock times the character selector, the R.O.M. and the horizontal and vertical drivers. 5. Character generator (R.O.M. ) . It accepts six input pulses plus one enable pulse of information and produces at its seven outputs five sequential groups of pulses scan- ning the character in an horizontal manner. 6. Vertical and horizontal drivers . These drive the corre- sponding X and Y address lines of the display. 7. Power supplies . There are +^v, +I5v, and one of varia- ble voltage. 8. Temperature control system . It is a "proportional" con- trol system used to maintain the "thermal" bias of the matrix. A detailed description of the electronic circuitry follows: 5.2 Input system (Figure I7) A simple input system generates, by means of Sj. and S r) the "l n, s and the "0"'s needed to produce the character- inputs to a six bit serial-in, parallel-out shift register consisting of three SN7^76 J-K F/F s. Only the "1" pulse is connected to the J, K inputs and the clock of the shift reg- ister. For well-shaped input pulses two bounce eliminators, 36 1-1/2 SN7475 1-1/2 SN747[> , ^ , \':'i i it- 1 :" SN7400 1' JK FN 1. rl" , £jj6^ ..-V IK ^ SN74121 I 3*SN7476 Si 3 1 ..!..- f+TT 7 + 5V MC3003 ^F Write "0" , — \A/v IK S 5 — vw- t IK + 5V SN7493 I SN7445 I 15 14 13 12 * *■ ■-<=, X 47 K 2.2^F ~1 1 1 C " D R SN7400 A -5-, 4.7K? + 5V SN7493 n u 14 13 12 -Jr SN74121 12 + 5V SN74 45 n A B (. 4 ♦ 5 8 6 7 e 4.7 K + 5V* ww 5 N 74 04 < 2XSN7402 SN74121H £ F^# ^^ 5-* 6— ■ 7 — ^£ £^£n £^£-n SN7-102 7&>- f-w — LqV i ik " + 5V CM K) o o S ro 2 O SN7404 K 2 l-.< - - •- 0. r, 6 SM?-;i?! (] ,-1 F^ I 4,13 4 SN74151 F J F ,_F 4 - F J,i F\. F\. F F 2.2/iF 4.7 SN74121 HE 5 1 - 3 F F "CJ F ilT To Inputs of R.O.M. From ♦ SN7493 ET From K 2 V •5V Figure 17. Input system, memory and character selector consisting of two SNy^OO's and the monostable SN7^l2l I (D - 50|xs), are provided. Each six bit character is fed to the memory elements, 1 1/2 SN7475 data latches (1-8). Only one is clocked at a time, accepting and storing the informa- tion while all the others remain closed. A switch S-. is provided to manually clear the shift register in case of errors. Two groups of counters ( SN749J3 I, II) and decoders (SN74H5 1j> II) introduce the information in groups of six bits by selecting sequentially every one of the 1-8 memory elements. This is done by using the eight groups of gates (SN7402, MC3OO3). The decoder ( SN7445 I) produces a short pulse A for every six counted pulses by SN7H93 I. The SN7 1 l i l-5 II produces eight sequential pulses B. When A and B are in coincidence the appropriate group of gates, SN7402 and MC^OOp^ in sequence, generate a clocking pulse and the information is entered into the corresponding 1 1/2 SN7445 memory element. At the eighth B pulse ( the 1-8 memory is then com- plete) the SN74121 II monostable is triggered (D =2-6 sees) which generates the K-, and K~ pulses and enables the charac- ter selectors SN74151, the R.O.M. TMS 4103 and operates the logic circuitry connected with the horizontal and vertical drivers. Simultaneously, the SN7 i ll2l III is triggered; so that the present information which is clocked by its output pulse is introduced and stored permanently by the l'-8' memory elements. 38 In this way the information can be displayed as long as the K-, and 1C pulses exist, i.e. for 2 to 6 sees. The switch S_, is provided for permanent display or "stand by." Mien the memory (1-8) is not complete, S^ can trigger the SN74121 II monostable producing all the previous actions. The monostable SN7^T2l IV is provided to clear the shift, register and the 1-8 memory elements at the end of each sequentially displayed character. 5.3 Memory (Figure I7) It consists of two groups of eight memory elements, 1-8 and 1-8', each consisting of 1 1/2 SN7^75 data latches. As described before, each of the 1-8 memory elements accepts and stores the information available at the output of the six bit 2 x SN7^76 shift register. This information is clocked just as it is ready to be displayed by a pulse from the monostable SN7^l2l III and is stored permanently in the l'-8' group of the memory elements. 5-^1- Character selector (Figure I7) Each character is selected sequentially by using six SN7HI5I one out of eight multiplexers. The required timing pulses are generated by a three bit SN7493 IV counter of pe- riod 10T (= 1 ms). Each character lasts 1 ms . The same counter controls the vertical drivers also. 39 5 . 5 Ch arac ter g enerator ( R . . M . ) This is an MOS Read Only Remory (TMS 4103) and is or- ganized as 64 v/ords of 55 bits (5x7) with a total capacity of 22*1-0 bits and is appropriate for horizontal strobing. It accepts six input pulses in US ASC II code (plus one enable pulse) at its character address inputs and pro- duces sequentially j at its seven column outputs, five groups of pulses , each of equal time length (t/5 = 200 [is). These pulses, fed to the horizontal drivers, are connected to the corresponding rows of the matrix (the emitters of the 2NJ3128 micro trans is tors ) as current pulses. At the same time these output- pulses appear at the appropriate rows, the timing cir- cuitry connects sequentially by means of the vertical drivers the bases of the 2N3128 (the columns) to ground as indicated in Figure 15. Therefore, only the microtransistors which are in coincidence are fed with power pulses and heat up at that time. For the horizontal strobing, five column select in- puts arc provided which are pulsed sequentially by T/5 = 200(_ls pulses generated by the SN7445 III decoder. Thus, each character, as explained previously, is completed in 1 ms. The field rate is 1/8 ms = 12 5Hz and is far above any flicker rate. This period also satisfies the condition T« x T . 40 The R.O.M. and the associated input-output interfacing circuitry are indicated in Figure 18. The input gates (SN7401) are able to withstand the +15 volts required by the R.O.M. In the output either low power TTL SN74LOO or descrete component inverters can be used. In the second case the fan-out is increased (e.g., for 2M5225 inverters). The 15K resistor (Figure 19 ) provides a path for the reverse currents of the R.O.M. and the 2N5223 inverters which is approximately 20uA. Therefore, its value must be R v < — ° ' ^ V g and R V <"20K . The 10K resistor provides ^20 10" 6 ^ enough of the amplitude of the input signal for the base of 2N5223 and also limits the power dissipation of the MOS driv- er in the R.O.M. Maximum access time of the TMS 4103 is 700ns. 5.6 Clock and timing circuitry (Figure 18) An as table multivibrator of frequency lOKHz (T - 100u.s) serves as the master clock. Frequency sta- bility is immaterial since it has the same effect on all "timed" operations. This clock times three channels: 1. By means of an SN7493 HI binary counter it feeds 2T = 200[is pulses to an SN7445 III decoder which pro- duces five sequential 200(_is pulses for the five column se- lect inputs of the R.O.M. 41 -15V + 15V TMS 4103 From the Outputs of < SN74151 From Kj<>- SN74LOO SN'7404 OH>~o To ► Horizontol Drivers BO R R R R SN7490 SN7493 IE 14 15 M 13 12 UJjeUJr- lEp RCD SN7445 3T. 2l_J3 9811 To SN 74151 ■{ B SN7445 12 1? 15 14 13 J_l_l From D A BC 12 3 4 5 6 7 1 I I I I I I ! 12 3 4 5 6 7 9 unwn CKj CK 8 Matrix 1 Matrix £ Figure 18. Clock and timing circuitry h2 2. The same counter also feeds the decoder (SN7445 V) which drives the logic circuitry of the vertical drivers with 100[is pulses. Every two successive pulses of this decoder (depend- ing on the final thermal design of the L.C. display) can be combined into one of 200|.is, or they can remain separate with the second of the two active only for timed periods of opera- tion. In the second case a monos table (SN74121 V) triggered at the beginning of each displayed character,, controls the length of these periods. Therefore, during that time the duty cycle and so the power for the thermal element is doubled. This is needed for fast thermal turn-on-transitions J>. An SN7493 IV three bit binary counter is fed with 10T = 1 ms pulses by means of an SN749O decade count- er working as a 10:1 frequency divider. The BCD out- puts of this counter are driving the selection inputs of the SN74151 multiplexers and a decoder ( SN7^45 IV). This decoder produces sequentially 1ms pulses CK-, - CKn which select each one of the eight groups of the vertical drivers for 1ms . As pointed out before it is important: 1. To preserve the !I thermal" independence of the ele- ments in the matrix. 2. To dissipate effectively the average heat generated during the operation of the display. M> 3. To have the capability of switching as quickly as possible between successive characters thus short- ening the turn-on time just as the display of each character starts. This can be done either as explained above or by us- ing a second group of horizontal drivers delivering an addi- tive current to the micro transistors of the matrix for timed periods of operation and controlled as indicated above. In the present matrix, because of the required close proximity of the thermal elements (0.16"), the hot elements influence the neighboring elements thus increasing their tem- perature to a considerable degree. In the worst case this is 0.4 - 0.5 c for the 1 c range L.C. To reduce this interac- tion a screen is used as indicated in Figure 13 • The metal- lic part of this screen is connected to a much lower temper- ature. In this way the hot air current is disrupted and diverted and is thus cooled somewhat before it reaches a neighboring element. Also, due to this screen, heat is continuously dissi- pated outside the matrix which prevents the accumulation of heat on the matrix and an accompanying uncontrolled increase of its temperature. Using the screen results is an improve- ment of 50 1 ^ leaving a remaining influence of 0.1 - u.25 c. This thermal influence can be further decreased if the ratio (Area displayed) / (required power) H is increased. In this way the dissipation of heat will be much easier. This leads to the conclusion,, that an integrated matrix of transistors instead of the present hybrid scheme would improve the situation. 5.7 A differe n t thermal design (Figure 19) The inactive elements of the matrix normally are at their threshold temperature so they are colorless. However, the interaction between elements shifts their color toward the violet, and so they appear as active confusing the dis- played pattern. A good solution to this problem would be the delivery of a negative thermal pulse to the non-active elements when the active elements are receiving their positive thermal pulses. In this way the non-active elements will always re- main below the thermal threshold since the actual influence of the active elements in the worst case is found to be about 1/2 T (= 0.5 c) in the present matrix. This idea can be implemented by either of the follow- ing schemes: 1. Divide the basic 200 h is pulse into two lOOu-S pulses C and D. When all the elements are non-active, the}' receive pulse C which serves as a thermal bias pulse. When the information is to be displayed only the active elements receive both C i i i : c :• -] < 5V 5V — j 2N5223 200 pF' Hh 1 1 v% I c A 3 10 K k — 5 5 P9 15K * 6 13 ~ 45 To ^ Horizontal Drivers Matrix 8 Figure 19. Modification of the Clock and timing circuitry 46 and D while the non-active ones no longer receive pulse C. The pulse C acts like a "negative pulse" for the non-active elements . This switching of pulses is done easily by two MC5OO5, SN7HOO gates, using the enable -disable pulses K~ and K-j from the input system. For fast turn-on time another group of seven horizon- tal drivers (group 2) is used. This part of the circuitry is as indicated in Figure 19 . 2 . By keeping the thermal bias lower than the thermal thresh- old of the L.C. as in the first case and using only one pulse to drive the active elements directly. This method leads to simpler logic for the drivers. However , it does not permit the exact regulation of the thermal bias because no sharp color change occurs in the blue -violet region for the present L.C. The first method was implemented and resulted in the easy elimination of the interaction between elements. In this case the duty cycle is 1:80. 5.8 Drivers These are designed using descrete components (2N^64 i l PNP transistors). They are described briefly as follows: (Figures 20 and 21) 47 5V 4 K 200 pr 32K I h.6. r i i i Rl vsy v^ 220n 16kft 2N3644 2N3644 0.033 pF -va ►To 2N3128 =!= ^e Emitters 50 pF "2 18K Jill Ri7.5K 5V ii R 3 IK To 2N3128 Bases 2N3644 Figure 20. Horizontal driver Figure 21. Vertical driver 48 - • Horizon tal dri vers. In order to be compatible with the SN7400 interfacing gates {T-j-. ay = 0.4uA), we must have R i + R 2OK . Values used: R x - 16K , Rg = 5JK , EU = 1.6K . In the case of D = \/KQ } the corresponding values are I 7? = 9.5mA, R^ = 36OSI, R^ - 3.3K . The R 1 and R^ calculated are different. However , because the previous driver satis- fies these less severe requirements, they are retained for this case too. 2. Vertical drivers . Taking for the 2N^l28 (in the worst case) hfc pulsed = 25* its base current is pV^m A = 0.6mA. Therefore, when the seven horizontal drivers of the matrix are activated simultaneously (worst case), each vertical driver has to be in saturation, a source of 7 x 0.6mA = 4.2mA. A direct circuit design leads to R, = 7 . 5K , R~ = 18K , R v = IK . Speed up capacitors are provided to increase the speed of the drivers so that the on and off portions of their pulses do not overlap and generate uncontrollable spurious 49 switching currents and heat on the 2NJ128 microtransistors of -t- 1 - ae matrjx. 5-9 Temperature c on trol system (Figure 22) It accurately controls the temperature of the "thermal plate (Figure 1J>) which thermally biases the elements of the matrix. The required accuracy is +0.05 c. A "proportional" temperature control system was adopted. Such a system has the characteristic that, at least for a part of its opera- tion range , the power delivered to the thermal load is pro- portional to the error signal. That is (V s -V A )aP where Vq is the voltage corresponding to the temperature set point Tg and V* is the voltage corresponding to the actual temperature as determined by and fed back by the thermal sen- sor. This power is given by P = IVD. Therefore, if D, the duty cycle, is changed as a linear function of the error sig- nal (V and I being kept constant) we can have a proportional control system for its whole region of operation. The circuit satisfying the relationship Da(Vg-V» ) consists of an integrator (1/2 MC1457L I, 2W$6h2 , 2N2646) generating ramps of constant period (T = 10ms) and a com- parator (1/2 MC1457L II) which compares the ramp to the error 50 2N3642 + 15V + 15V i 1/2 MC1437L H + 110V Thermistor 500&i -15V- 18K 10K > 500 Temperolure s* Adjustment A " -15V pfi ' MC1439 nr 0.002/xF 150K I — wv- 150 K — vw- T^Pf~^ 150K 1C1349 W. = = | 390ft 0.002/iF 110V BB527 110V o.c. o+15V o + HOV Tr4 Figure 22. Temperature control system signal continuously as it is amplified by the operational am- plifier MC1^39a III. The higher the amplitude of the error signal the larger the duty cycle D of the rectangular pulse at the output of the comparator. In this way (V q -V.) and D are linearly related. Temperature errors in this system can be produced: 1. By offset and drift of operational amplifiers. 2. By self heating of the sensor, in this case a PTC thermistor. 5. By drift of the reference voltages. Low drift operational amplifiers are needed to elimin- ate the first problem. The thermistor used is a PTC type (Ro S o = 500#) with power dissipation 0.7u.W. Its thermal time constant is small enough (2 sees) for fast response. In this way we can mini- mize "thermal" overshoots which are caused by the time lag between changes of the thermal load and that of the sensor signal. In order to avoid self heating of the thermistor, a relatively small current (0.8mA) and thus a power of 0.^2mW is provided. The corresponding slope of the voltage - temperature curve in its middle region of operation is ap- proximately 4.5mV/ c in the present circuit. Therefore, any shift of the +15 volts reference voltage greater than k RmV / c <* ' -, 0.225mV is not acceptable. For this reason a 0.05 c 52 Burr-Brown 527 power supply with good load and temperature regulation Is used. This proportional control system approaches the tem- perature set point as3nnptotically . In order to shorten the required time an extra comparator (MC14;59G IV) is provided to deliver in parallel with the main control system a D.C. exci- tation to the output power transistor (GE 12). In this way the temperature changes rapidly initially. The MC1439G IV stops delivering that excitation when the control voltage across the thermistor drops below a preset point. This point is set appropriately by a IK pot 171 order to avoid thermal overshoots. For the present thermal plate the required power for the range 25 -45 c i- s approximately l5w. This heat is gener- ated by a resistor Ky = 760ft connected as load to a switching transistor (GE 12) with a high breakdown voltage BV^q. A high voltage,, low current operation is selected in order to decrease R.F.I problems. 5 . 10 Power supplies For the +5v source an external power supply is used. For the -5v and the variable voltage source (4-l5v)j the de- signed power supplies are as indicated in Figures 23 and 24. 53 1N4004 — of- CH — ' Tr l 1NI4004 5K 100 K -15V o y& — ■ 2 N 3642 100 K 15K tifctf ] I Vf 3 1/2 MCK .E~T4 I 3 1/2 MC JL S390JI 0.002 M F Figure 23. Variable voltage power supply Tr2 INI 4004 MC1461R 30V J = 2 5k|'L 1100 M F J 6.8K* 2 o.6n 1 =h0.1uF *~r -v 5V -r 100/iF 35V Figure 24. -5 volts power supply 54 6. CONCLUSIONS The present study demonstrated that the construction of an alphanumeric display is possible using the thermal properties of liquid crystals. The quality of the colors de- pends on the liquid crystals chosen; some colors appear weak because of the poor reflectivity of the L.C. The problems of "thermal" isolation as well as those of dissipation of the heat of the matrix are crucial ones for this kind of display. Because of small variations in the thermal parameters of the thermal elements as well as effects due to location,, obtaining an absolutely uniform thermal bias is not possible in the present hybrid type matrix. These non- uniformities cause small differences in the displayed colors from element to element. The present speed of the display is of the order of 3 sees. These problems will be greatly reduced if the size and the thermal parameters of the thermal elements is decreased; this will decrease the required power per displayed area and it will increase the speed of the display. Therefore, an integrated circuit type of matrix rather than the present hybrid matrix will yield better results from the electronic point of view. However, many improvements involving higher reflectivity, better purity of colors and a decrease in the angular dependence must be made for the 1 i q uid c ry s t a 1 s . 56 LIST OF REFERENCES 1. M. V. Joyce, K. K. Clarke, "Transistor circuit analysis," Addis on -Wesley Publishing Co., I96I. 2. J. L. Fergason, "Liquid Crystals," Scientific American 211, 77 (1964). 3. J. L. Fergason, "Cholesteric structure-I, Optical proper- ties," "Liquid crystals," G. H. Brown, G. J. Dienes, M. M. Labes, Gordon and Breach Science Publishers, 1966. 4. G. N. Gray, "Molecular structure and the properties of liquid crystals," Academic Press, London, I962 . 5. J. L. Fergason, N. N. Goldberg, R. J. Nadelin, "Choles- teric structure -II, Chemical significance," "Liquid Crystals," G. H. Brown, G. J. Dienes, M. M. Labes, Gordon and Breach Science Publishers, I966 . 6. J. L. Fergason, "Liquid crystals and their applications," Electrotechnology, 1970. 7. M. Kutz , "Temperature control," John Wiley and Sons, 1968 . 8. C. W. Osseen, "The theory of liquid crystals," Transac- tions Faraday Society, 29, 883 (1958). 9. "Encapsulated liquid crystals," N.C.R. Bulletin. 10. "MOS Character generators," TEXAS INSTR. INC., Bulletin CA-145, I97O. 11. "Solid state alphanumeric display," HEWLETT-PACKARD, Application Note 931. mSICLASSIEEED Security Classification DOCUMENT CONTROL DATA - R&D (Security classification ol title, body ol abstract and indexing annotation must be entered when the overall report /« claesilied) I 1 ORIGINATING ACTIVITY (Corporate author) Department of Computer Science University of Illinois i Urbana, Illinois 61801 2a REPORT SECURITY C L AS S I F I C A T I ON Unclassified 2b CROUP E 3- REPORT TITLE COLORAMATRIX. A THERMALLY CONTROLLED LIQUID CRYSTAL ALPHANUMERIC DISPLAY ! 4. DESCRIPTIVE NOTES (Type ol report and inclusive dates) I Master's Thesis, Technical Report June, l£7l 5. AUTHORS (Last name, tirst name, initial) Hadjistavors, Stavros Alexandros , 6- REPO RT DATE June, 1971 7a. TOTAL NO. OF PACES 6k 7b. NO. OF REFS 11 8a. CONTRACT OR CRANT NO. N000 1A-67-A-0305-0007 6. PROJEC T NO. i c. • d. 9a. ORIGINATOR'S REPORT NUMBER'S,} 9 b. OTHER REPORT NOfSJ (A ny other number* that may be assigned this report J 10. AVA IL ABILITY/LIMITATION NOTICES 11. SUPPLEMENTARY NOTES 12 SPONSORING MILITARY ACTIVITY ■ Office of Naval Research 219 South Dearborn Street Chioagn, TJLLLnQia 6060^ 13. ABSTRACT The basic idea employed by COLORMATRIX involves using the thermal properties of cholesteric liquid crystals in order to display information in color. Cholesteric liquid crystals have the property of selectively reflecting different wavelengths of incident white light as their temperature is changed. This phenomenon is reversible and the spectrum of the reflected light is a function of both the chemical composition and the temperature. Since liquid crystals are light reflecting devices, low power consumption and high contrast are their main advantages; however, their reflectance is low. I U 1 JAN 64 J A? /O UNCLASSIFIED Security Classification EaBnol HUH BEL ■F UNIVEHSrTYOFILUNOIS-URBANA J u'n.VWS.TY OF .LL.NO.S-UBBANA m ^HBHI BUM KUH H BE 510 B4IL6Rno.C002no.4B0«6(19'1 SynchronlMtlon system lo.. revision ■■I 3 0112088400061 ■t " 'a I H ■ ■ ■ : ■ ■^■1 ■ ■ V m HJ i ■ ■