=======================LIQUID_CRYSTAL_DISPLAY========================================= LCD Liquid CrystalDisplay (FAQ) Frequently Asked Question two modes LCDs during absence of an electric field mode describes transmittance state of liquid crystal Normal White mode: display is white or clear,allows light to pass through Normal Black Mode: display is dark and all light is diffused. Virtually all displays are normal white mode to optimize contrast and speed. LCD Cell Fundamentals twisted nematic (TN) LC display consists of two polarizers, two pieces of glass, electrode to define pixels, and driver Integrated Circuits (ICs) to address rows and columns of pixels. To define a pixel (subpixel element for a color display), rectangle is constructed out of Indium Tin Oxide -- a semi-transparent metal oxide (ITO) and charge is applied to this area in order to change orientation of LC material ( change from a white pixel to a dark pixel). viewer ///////////////////////////////////// Polarizer _____________________________________ glass ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Liquid Crystal _____________________________________ glass \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Polarizer backlight Figure 1 cross sectional view of a simple TN LC display. as viewed without its metal module/case exposing IC drivers. Looking directly at display gate or row drivers are located either on left or right side of display while data or column drivers are located on top (and or bottom) of display. Figure 2: LCD panel and IC driver locations _______________________________________ | | | IC IC | Source/Column ICs | | | | | | |IC---------------------Pixel | | | |IC <---- Gate Line/Row IC | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Figure 2 depicts a dot matrix display An IC driver will address number of row/column lines and not just single pixel in diagram above Polarizers are an integral part of a LCD display, possessing unique property of only passing light if it is oriented in a specific (oriented) direction. To utilize this phenomena in TN LC displays, bottom polarizer orients incoming light in one direction. The oriented light passes through LC material and is either unaltered or "bent" 90 degrees. Depending on orientation of top polarizer, this light will either pass through or be diffused. If light is diffused, it will appear as a dark area. Figure 3 is a simple illustration of sequence of events that occur when light passes through a simple twisted nematic LC display. Figure 3: Polarized Light and its use in a TN LC display Light (unoriented) will be defined as: !#$%&|- Polarizer Orientation is defined by: ( $ or # ) ($ polarizer will only pass $ light) (# polarizer will only pass # light) THEREFORE: Light Polarizer result LC (90 result Polarizer Image Input type passed degree passed type output twist) !##$%&|-> |#|-> #### -> ~~~ ->$$$$ ->|#| -----> Black !##$%&|-> |#|-> #### -> ~~~ ->$$$$ ->|$| -----> White think of figure 4 as one thin slice (one layer of molecules ) of a block of material. If you examined another slice, molecules would still be oriented in Y direction, but they would be in different positions along X-axis. By stacking millions of these thin slices, Z direction is built up and as a result of change in relative position on x-axis, Z direction has no long range order. ^ Y | Figure 4 | | | | |||| || || | |||| | | ||||||| |||||||||||| ||| | | | ||||| ||||| ||| ||| | |||||| | |-----------------------------------------> X * The Z direction is coming out of page toward reader liquid crystals used for display technology are thermotropic liquid crystals; desired characteristics over a specific temperature range. If liquid crystals are too cold, will not twist too warm,resistance of liquid crystal material changes and this alters properties of display Liquid crystal material referred to as TN (STN, DSTN, MSTN, and etc.) or Twisted Nematic--sometimes known as TNFE or Twisted Nematic Field Effect. It is called TWISTED since crystals are twisted 90 degrees (or more for STN) from top piece of glass to bottom piece of glass. (TN usually refers only to a 90 degree twist.) polarizers are also one of major reasons LC displays require bright back lighting. polarizers and liquid crystal materials absorb more than 50% incident light. power hungry back lighting makes a LCD module one of primary causes of short battery life in notebook computers. Liquid Crystal Alignment Liquid crystals must be aligned to top and bottom glass in order to obtain desired twist. 90 degree twist is formed by anchoring liquid crystal on one glass plate and forcing to twist across cell gap (distance between two glass plates) when contacting second plate. process consists of coating top and bottom sheets of glass with a Polyimide based film. top piece of glass is coated and rubbed in orientation; bottom panel/polyimide is rubbed perpendicular (90 degrees for TN displays) with respect to top panel. It was discovered that by rubbing polyimide with cloth, nanometer (1 X 10 - 9 meters) size grooves formed and liquid crystals align with direction grooves. It is common that when assembling a TN LC cell, it will be necessary to eliminate patches of non- uniform areas. two parameters required to eliminate nonuniformities and complete TN LC display are pretilt angle and cholesteric impurities. TN LC cells commonly have two problems that affect uniformity following assembly: reverse tilt and reverse twist. Reverse tilt is a function of applied electrical field and reverse twist is common when no electrical field is applied. Reverse twist is eliminated by introduction of cholesteric additives and reverse tilt is eliminated by introducing a pre-tilt angle to LC material. The pre-tilt angle also determines what direction LC molecules will rotate when an electrical field is applied. Pre-tilt angle can be visualized by considering normal position of LC molecule to be flat against glass plate, by anchoring one edge and forcing other upward by a specific number of degrees, a pretilt angle is established. 1.23 Liquid Crystal Display Names and classes Before discussing different types of LC displays topic of Birefringence must be explained. When a light ray strikes a crystal ( or crystal-like material), it will be split into two separate light beams; with one beam perpendicular (offset by 90 degrees) from other. Since beams travel different paths, they reach viewer's eyes at slightly different times. This is an essential point, it may cause color or polarity of display to change when viewed at angles where viewer may see both rays. For active matrix displays, in order to maximize contrast and gray scale reproducibility, Twisted Nematic (TN) is utilized. This material is twisted 90 degrees from top to bottom glass panels. STN or Super Twisted Nematic is chemically distinct from TN and twist angle is usually greater than 200 degrees. Furthermore, due to large twist angle, actual alignment of polarizers for STN LCDs are not perpendicular, but adjusted to find best direction (rotation) for optimum display characteristics. The STN material is rotated in a way so change from transmission to dispersion is very abrupt and therefore can respond quickly to small changes in voltage. Figure 5 illustrates response characteristics of a TN curve and Figure 6 shows response characteristics of a STN curve which will further clarify these points. 100%| Figure 5 | Typ Response of Normal White TN Display T | R |********** A | * N | * S |<- Zone I-> * M | * I | * T | * T | * A | * N | <----Zone II--- * ---------> C | * E | * <-Zone III-> | * | * | **** 0% |__________________________________________> Vt (Threshold Voltage) Applied Voltage 1.24 The TN Liquid Crystal Response Curve The most prominent feature of TN response curve is central linear region between two flat areas (Zone II). Zone I describes white color of display when no electric field is applied. In other words, display will transmit virtually all introduced light. On other hand, in Zone III, display will diffuse light and appear dark. The middle region can display gray scale or an image somewhere between White and Black. The key point here is that you must be able to very carefully control voltage applied to LC cell and maintain it for one duty cycle (before that pixel is addressed again) in order to produce accurate colors. For this reason, this type of LC material is primarily used for active matrix LCDs. COMMONLY ASKED QUESTION NOTE: because LC material is partially twisted in gray scale area, when looking at a display at an off angle colors tend to shift and sometimes invert due to birefringence. COMPUTER APPLICATION NOTE: The TN response curve does not have to be utilized for gray scale, in order to make a simpler display, improve viewing angle, and use cheaper IC drivers; Apple Powerbook 170's TFTs (thin film transistors) drive TN response curve directly into region 3. This gives all speed/contrast advantages of a TFT display and cheaper manufacturing cost, but provides no gray scale. 1.25 The STN Liquid Crystal Response Curve The Key to understanding STN curve is simply that due to addressing method applied, only a small amount of voltage is available to change LC material from transmittance to a dispersion state. For this reason, shape of curve has nearly a 90 degree shift between Zone I and Zone II regions; in other words, it goes ballistic and nearly straight up ! This property allows LC material to shift from white to black at its threshold voltage (VT) without being concerned with partial transmission (gray scale). Furthermore, 90 Degree curve shape means that gray scale is not available from LC material itself and driving circuits must provide necessary fixes for levels of gray. STN displays inherently have a yellow on blue appearance (anyone remember old Zenith Laptops ?). Because many individuals found yellow and blue appearance undesirable, a number of techniques were developed to convert STN image to a black on white scheme. DSTN, developed by Sharp Corporation, was first commercial black and white conversion of STN display and refers to Double Super Twisted Nematic. DSTN displays are actually two distinct STN filled glass cells glued together. The first is a LCD display as described previously, second is a glass cell without electrodes or polarizers filled with LC material for use as a compensator which increases contrast and gives black on white appearance. The drawbacks are a heavier module, a more expensive manufacturing process, and a more powerful backlighting system. FCSTN is Film Compensated STN and is now most commonly used STN display technology on market. FSTN, monochrome STN, and Polymer film STN are all standard STN displays with a polymer film applied to glass as a compensation layer instead of second cell as in case of DSTN. This simpler and more importantly cost effective method provides preferred black on white image for this display technology. However, once again, this design lowers transmittance of light and requires a more powerful back lighting system. COMMONLY ASKED QUESTION NOTE: Why are STN displays slow ? Due to method used to address passive matrix (STN/DSTN) displays and high density of pixels required for standard VGA displays, liquid crystal material must respond to an extremely small change in voltage. In developing these materials for this voltage characteristic, there was a reduction in switching speed. A slow display can best be illustrated by tendency of cursor to "submerge" or disappear when rapidly moved across screen. Another common example is blurring of images when they quickly move across display as in case of high speed games. A fast display is less than 40 milliseconds, most STN type displays are between 200 and 250 milliseconds. However, some new LC mixtures are reaching 150 millisecond speeds. COMMONLY ASKED QUESTION NOTE: What is Contrast ? Contrast is defined as ratio of black to white, more simply put, how black is black when next to a white or clear pixel. In terms of numbers, passive matrix LCDs are usually able to produce a contrast ratio of approximately 13 - 20:1; in real terms you get a set of different grays and blues but no true blacks. 100% ^ | Figure 6 T | R | ******** A | * N | * S |<-Zone I----> *<--Zone II----> M | * I | * T | * T | * A | * N | * C | * E | * | * | * | * | * | ************ 0% ----------------------------> Vt Threshold Voltage Applied Voltage 1.3 Liquid Crystal Display Assembly Once switching devices or electrodes have been fabricated on glass halves and polyimide film has been applied & rubbed, spacer balls (usually 4 to 8 micrometers [1 X 10 - 6 meters] in diameter) are sprayed on one half of display. Spacer balls are used to insure that glass plates remain a certain distance apart over entire area of display; this is also known as cell gap. If cell gap is not uniform, an image will appear different from one end of display to other. If spacer balls are not applied correctly, they will collect and user will be able see them as strange areas of non-uniform dust or distortion. (Single spacer balls are too small to see and they are not black dots.) If Display has a very large cell gap, when you apply slight pressure to display by touching it with your finger, you will see image change and LC material shift under glass. Doing this does not damage display, but take care when bringing any sharp objects, such as pen or pencils, near screen; it is very easy to damage polymer film and or polarizers on display. The two glass panels are then aligned and glued together with an epoxy. During panel assembly, if dirt is trapped between two glass plates, you most likely will see these as annoying spots on display. During application of glue, one corner is left open. In a vacuum chamber, liquid crystal material is drawn into display through open corner. Upon completion, remaining hole is filled with another epoxy. The LC material will align itself to grooves in polyimide and spread out around spacer balls. After final assembly, excess glass is cut and driver ICs are mounted. The finished display is mounted onto a backlight assembly (also known as an inverter assembly) and encased in metal. There are a number of methods for backlighting a LC display. STN displays usually have a side, top, or bottom lighting system. In simple terms, this is where fluorescent tube is mounted. For example, in a side-lit display one or two fluorescent tubes will be located at left and or right edges of display. A fluorescent tube normally 4 mm in diameter is used. This is dispersed by a plastic plate around entire area of display. A dispersion plate looks like a white sheet with small holes; each of holes provides a small point of light. On top of dispersion plate, a diffuser is placed. A diffuser takes numerous points of light and uniformly spreads it out over entire area of display. The net effect is providing a backlighting source around 4 or 5 mm thick ! An Active matrix display, especially color modules, transmit much less of incident light and require more elaborate backlighting systems. An active matrix TFT display has a matrix fabricated on one piece of glass; metal lines and transistor elements are not transparent and block a significant percentage of light. In order to obtain higher contrast, newer displays incorporate what is called a black matrix. This is a black film that surrounds pixel elements (this can be on matrix, but is usually around color filters); although this yields higher contrast, it also reduces brightness. Further complicating this, polarizers and color filters reduce output to less than 50f incident light. As a result, most backlighting systems designed for active matrix based displays usually consist of 4 or 5 four mm tubes placed directly behind display with a diffuser plate to insure uniform irradiation. Therefore, they are called backlit. This method of lighting makes display slightly larger, heavier, and greatly increases power consumption. The final metal encased display is called a display module or sub-assembly and this is what end user or notebook manufacturer written permission from author. Every effort was made to insure validity of this document. LCD FAQ Part II: Addressing and Color Technology (Passive and Active Matrix Displays) 2.0 General Overview: Addressing describes method employed to transfer charge (data or display image) from outside world to display. Unlike a CRT, which is just a surface of phosphors scanned with a beam of electrons in a vacuum, a liquid crystal display is an array of conductors with metal (or metal like) lines running in both horizontal and vertical directions. For case of a CRT, electrons travel through a resistance free medium (vacuum) and deliver a clear consistent signal. The charge traveling through metal lines of a LCD matrix is affected by properties of metal. As a result, magnitude and waveform of applied charge can vary from one end of display to other. This variation imposes limitations on display quality and capabilities. 2.1 Addressing: Passive and Active Matrix Displays There are distinct differences between active and passive matrix displays, but two factors make greatest impact on potential customers. Active matrix displays can cost twice as much as an equivalent passive matrix display and add more than $1000 to cost of a notebook form-factor computer. However, active matrix displays produce a stunning and bright image without ghosting or artifacts that rivals quality of CRTs. Furthermore, even with price differential, manufacturers are able to sell every active matrix color notebook they can produce. 2.11 Liquid Crystal Cell Charging In general terms (regardless of display type), in order to protect liquid crystal material from deteriorating, cells are addressed by alternating current (AC), not direct current (DC). There is no resultant charge in LC material following two addressing cycles; build up of charge in LC material will permanently damage it. In other words, a positive and then an equal but opposite negative charge is applied to LC material every other frame. By applying dual polarity addressing, LC material changes twist direction every other cycle and net charge is zero. Furthermore, since liquid crystal material is changing twist directions every other cycle, screen savers or screen inverters are not required and in reality do absolutely nothing. Passive matrix displays utilize DIRECT ADDRESSING; charge is applied directly from drivers to pixel element. Active matrix displays utilize INDIRECT ADDRESSING; charge is "filtered" through a switch before reaching pixel element. 2.12 Driving Methods: Passive Matrix Displays Passive matrix displays have rows of electrodes on one half of display glass and columns of electrodes on other. The electrodes are usually fabricated out of Indium Tin Oxide (ITO), which is a semi-transparent metal oxide. When two pieces of glass are assembled into a display, intersection of a row and column form a pixel element. Furthermore, if a pulse is sent down one row and a specific column is grounded, established electric field can change state of liquid crystal(from white to black). By repeating this process (display scanning) an image can be formed on display. Problems arise as number of rows and columns increase. With higher pixel density, electrode size must be reduced and amount of voltage necessary to drive display rapidly increases. Furthermore, higher driving voltage creates a secondary problem; charging effects. Even though only one row and column are selected, liquid crystal material near row and column being charged are affected by pulse. The net result is pixel selected is active (dark), but areas surrounding addressed point are also partially active (grays). The partially active pixels reduce display contrast and degrade image quality. A final problem is speed of STN material, a display must be able to react in less than 40 milliseconds for performance similar to a CRT. Most STN materials are between 150 and 250 milliseconds and can not switch from black to a white image that quickly. This problem results in disappearing cursors and blurred images when high speed graphics are utilized COMMONLY ASKED QUESTION NOTE: What is STN Gray Scale ? As discussed in part I section 1.24, STN curve does not possess an intrinsic gray scale capability like TN curve, therefore driving methods have been developed to create illusion of true gray scale. Gray scale can be derived from frame-rate control and dithering/space modulation. Frame-rate control quickly switches on and off a pixel, eye perceives this as gray. Dithering or space modulation is accomplished by alternately keeping some pixels black and some white in a checkerboard layout; when using this method, layout is in a random order. If dithering is in a regular (repeating) pattern, it is detectable by human eye. In real world applications, combinations of both technologies are applied to commercial displays. The result, however, is sometimes wavy or moving grays. ( The image appears to be moving in waves or a solid color appears to be in motion when a large area is set to a specific gray level.) RECENT TECHNOLOGY NOTE: What are Dual Scan STN Displays ? This is simply taking currently made color STN displays and applying some previously developed technology. Back in late 1970's and early 1980's liquid crystal chemistry was not as advanced and in order to build high data content displays, manufacturers were forced to build two displays on one glass plate. A dual scan display utilizes similar technology. Instead of running columns down entire display, they are terminated in center of display, a small gap is left, and line is continued to bottom of display. In reality, you now have two 640 X 240 displays on one glass plate. Therefore, if IC drivers are mounted on top and bottom of display, charge must only travel half distance of a normal display. As a result, effects of contrast limitations discussed in section 2.12 can be reduced. The end result appears to be a brighter display, but in reality it is only improved contrast (the blacks appear darker). The dual scan STN display still suffers from ghosting and artifact problems inherent in all slow STN displays. 2.13 Driving Methods: Active Matrix Displays In order to eliminate problems of STN/passive matrix display, active matrix display was developed. Active matrix displays have a thin film Transistor or diode on glass substrate that indirectly addresses each pixel element. Depending on display type, application, or manufacturer, TFT may be comprised of amorphous silicon (a-Si) or polycrystalline silicon (p-Si), The TFT completely isolates one pixel element from others in display and eliminates problem of partially active pixels. Simply put, TFT acts as a switch ! When a row of TFTs are addressed gate lines are active-- switch is turned on, this allows charge to flow from columns into pixels and set image for frame cycle. Once a row has been addressed, gate line is reversed biased (the switch is turned off) to insure that no charge can pass from columns into pixel element. Thus, pixel is now completely isolated as rest of display is addressed. The LC material acts as a capacitor and stores charge. After a charge is placed on a liquid crystal cell ( defined pixel area), it begins draining similar to a discharging capacitor (an exponential function). As a result, unless display can be written quickly (all 480 rows scanned and return to top of display to rescan starting from row 1 for a VGA display) image will not be uniform from top to bottom of display as LC material starts to untwist. In order to insure charge storage for one frame and carefully control charge on a pixel element, TFT displays incorporate a second capacitor in parallel with LC material. The combined capacitance gives active matrix displays essential capability to accurately maintain amount of charge applied; thus reliable partial charges can be utilized and gray scale or full color displays are possible. With proper drivers and high quality TFTs, 256 gray scales have been obtained with quality that surpasses that of CRTs. TN material can also switch much faster than STN, thus 40 millisecond TFT displays are common yielding CRT like speed. COMMONLY ASKED QUESTION NOTE: What is cross Talk ? Cross talk is best described as effect when a dark image (box or widow shape) is placed in middle of a white background. Faint vertical and horizontal lines will be seen from edge of window proceeding to edges of screen. This can be caused by poorly designed drivers or poorly made TFTs. The off selected TFTs are not completely off and some charge, very strong at edge of a window, has leaked into pixels creating effect. This is a common problem and is being addressed by manufacturers of TFT based displays. 2.2 Color Display Pixel Layout and Yields In order to build a fully functional color VGA display, a TFT LCD must have 480 X 640 X 3 pixel elements. The 640 X 480 is well understood VGA pixel layout (640 X 400 for Apple Powerbook series and 640 X 480 for 180c), except 640 red, 640 green, and 640 blue columns or stripes of color pixels are now required. This is a total of 921,600 TFTs that must work in order to build a perfect display. Using a semiconductor analogy, it is similar to building a 1 Megabit DRAM on a 10 inch glass plate; not an easy task considering a particle smaller than diameter of human hair can destroy a single TFT. Achieving a 100 % yield or a perfect display is virtually impossible, thus even though manufacturer yields are starting to reach 60% (for sellable devices) prices are very high. Furthermore, if 4 defective pixels are found on a color VGA screen, this already represents a 99.99 % pixel yield-- not bad for any process. For most part, most common pixel defect is caused by some form of contamination damaging a TFT and preventing it from turning pixel off (seen as bright spots on a dark background). There are two layouts for pixels on TFT displays. The most common for computer applications is STRIPE layout. A stripe layout has repeating stripes of red, green, and blue columns across display. For multimedia and high density arrays (projection LCD modules), a triad pixel layout is used. A triad layout has three color sub-pixels in a triangle shape. Figure 7 illustrates difference between two layouts. Figure 7: Color Pixel Layouts Stripe Layout Triad Layout RGB RGB RGB R R R RGB RGB RGB G B G B G B Note: What is a color filter ? A color filter works by absorbing specific wavelengths of light and only passing light of a certain wavelengths (In other words, a red filter will remove all wavelengths of light except for red -- thus it looks red !). White light is made up of a spectrum of wavelengths, so it can yield red, green, and blue for displays. However, when filtering out unwanted wavelengths, overall brightness is reduced. COMMONLY ASKED QUESTION NOTE: What is a pixel ? Unfortunately, in most of literature and magazines, there is not a clear definition as to what a pixel is. In its most basic form, a pixel is described as one element on a display screen. For a monochrome screen this is an adequate description. However a color pixel is actually made up of three subpixels: a red, green, and blue pixel. This is sometimes called a pixel triad. Therefore care must be taken in describing pixels. In terms of this document, a pixel is entire element consisting of red, green, and blue sub- elements. A subpixel consists of individual red, green, or blue elements. Gray scales for LC displays are always calculated as a function of subpixels. 2.3 Color Displays: Gray Scales and Bits Due to overall poor performance of Passive Matrix color displays, only active matrix displays will be specifically discussed. However, major points are applicable to both display addressing technologies. Unlike an analog CRT, in a digital color TFT active matrix display, you literally get what you buy...forever. Even if you upgrade to a new video driver or display card, you will still have same number of colors and gray scales. The number of colors is a direct result of number of gray scales a display can reproduce. The standard VGA format is rated to display 256 colors, however it can select from a 18 BIT CLUT (color look up table) which means choice of 262,144 colors(this calculation is based on a bit calculation for a pixel triad -- 2 ^ 18 --- see later section on calculations). Intrinsic gray scale reproducibility for TFT displays is a result of two factors: quality of driver ICs used on display and resistance of gate metal(the rows of display). The gate metal must carry a clear and undeformed pulse from one end of display to other( 640 X 3 = 19200 lines). If pulse is not maintained TN curve will not charge to desired level and correct color can not be displayed. Therefore, more gray scales required, greater control that must be exerted over gate lines. For example, most displays sold today can display 256 colors out of 4092 or 512. The 256 colors is based on VGA video controller, 4092 is a display limitation. 4092 possible colors indicates that a display can reproduce 16 gray scales. This is derived from 16 (red) X 16 (blue) X 16 (green) = 4092 possible colors. Once again, dithering can be used to extend this, but there are displays in limited production that can reproduce 256 gray scales or more than 16 million colors ! Most current TFT color displays feature 3 bit drivers (where 2 raised to third power yields 8); these drivers can produce a total of 512 colors. This is more than adequate unless later on you decide you wish to pursue some multimedia functions which require more than 32 levels of gray scale. Although controller and computer may be fast enough to handle functions, 8 or 16 gray scales will be inadequate-- your image will not be what you expect (It will look like a collection of color shadows). Sharp has recently demonstrated 10 inch 640 X 480 displays running on Apple Macintoshes displaying 64 gray scales. These 6 bit drivers are supposedly entering production and will enter commercial market shortly. The color reproduction of these displays is excellent. 2.4 Understanding Digital Color Pixels 4 Bits, 8 Bits, 16 Bits, 24 Bits Just how many colors can they actually generate ? Digital video divides number of colors or gray scales into a distinct number of points. Based on these "POINTS" system can generate a fixed number of colors or gray scales. Manufacturers tend to play games with numbers, so sometimes it is very difficult to understand "BIT" color talk. First of all, bit system is based on binary system so: 1 Bit color, which is 2 raised to first power is 2. In other words a black & white display where pixel has a state of being either on or off. This can currently be extended to 24 bits which (at 2 to 24 power) yields more than 16 million gray scales. OK, now that we understand how gray scales are calculated, lets convert this to a color display: Once again, manufacturers play a game with numbers and here we introduce bpp or BITS PER PIXEL. Now depending on manufacturer, a pixel can be made up of 1 subpixel (the individual Red, Green, and Blue pixels) or can be a composite of all three colors. If we examine a 16bpp system following calculations are applicable: 2 raised to 16th power is: 65536. So if system is 16 bpp for combined primary colors, system can produce a total of 65536 colors. If system produces 16 bpp for all three colors then 65536 X 65536 X 65536 = 2.8 X 10^14 colors. The small table below summarizes Bits confusion. Triad Pixel refers to a combination of RED/GREEN/BLUE pixels. Subpixel refers to individual red, green, or blue pixel. The number of gray scales for a monochrome display is always same as a triad calculation bpp display. The numbers listed down columns refer to how many gray scales or colors that a system configuration can produce. CRT based color is usually calculated as triad pixel calculation. Bits/Colors Mono or Triad Pixel Subpixel (R/G/B) 1 2 8 4 16 4096 8 256 1.68X10^7 16 65536 2.81X10^14 24 1.68X10^7 4.72X10^21 COMMONLY ASKED QUESTION NOTE: What is analog video ? Unlike digital or bit based video analog video is based on a continuous flow of data. The wave form can be thought of as a continuous wave of points with distance between points so small that it is impossible to differentiate between them. In other words, it can theoretically provide an infinite number of gray scales. VGA is an analog system and VGA CRTs are analog displays. The advantage of an analog display is that when you upgrade your video card and drivers to handle more colors, your existing monitor should be able to operate with extended color ranges. NEC makes an analog XVGA TFT LCD, but due to power handling requirements, it is not suitable for battery based portable computers. COMMONLY ASKED QUESTION NOTE: Why are TFT Color Displays Expensive ? There are numerous reasons for this. As discussed above, displays with large numbers of defective pixels can not be sold and as a result, yield is usually thought to be major problem. In reality, one should be aware that largest cost of TFT displays are materials utilized for production. Since Japanese manufacturers have not standardized size of displays yet, each manufacturer has specific material needs (glass, holders for machines, robots, and etc.). This fact alone keeps display prices extremely high since material and machine suppliers can not make standard parts for an entire industry at this time. COMMONLY ASKED QUESTION NOTE: How many gray scales are required for multimedia operations ? Usually 64 gray scales or more are required for true multimedia operations. TFT LCDs with 64 gray scales will probably be available in volume within a year. COMMENT: Why are some color TFT displays much brighter than others ? The brightness of a screen is determined by two related factors; size of screen and aperture ratio of pixels. On surface of an active matrix array there are both pixels and electronics, as a result of opaque electronics, some of area that light could pass through is blocked. The ratio of light passing through pixel to entire area of pixel and associated electronics is called aperture ratio. The larger ratio is, more light that can pass through pixel and brighter image on display will be. Furthermore, if display itself is bigger, there is more room for pixels and result is more light passing through individual pixels. For this reason, DTIs 10 inch display found in IBM Thinkpad 700c is much brighter than some of smaller 8.4 and 9.5 inch TFT displays. 2.4 Basic Principles of TFT Operation For all intensive purposes, a TFT can simply be considered a switch; when selected (on) it allows charge to flow through it and when off it acts as an barrier preventing or at least restricting flow of charge. As mentioned earlier, a TFT is a MOS FET device or a Metal Oxide Semiconductor Field Effect Transistor. The gate line can be considered "switch" of transistor, with this you turn it on, partially on, or off. The Source and drain are entrance and exit, respectively, for charge you want to pass through switch. In case of a display, this is charge that you want to appear on pixel. Looking at Figure 7, source and drain metal electrodes are separated by an amorphous silicon (a-Si) semiconductor layer; with absence of charge a-Si layer acts as an insulator or resistor and prevents flow of charge from source to drain; thus isolating pixel from rest of display. SiNx or silicon Nitride is gate insulator and forms gate dielectric; electrons do not pass from gate line into transistor, but are used to influence charge distribution in semiconductor layer. A MOS FET that fits this description (you turn it on)is called an enhancement device. When a positive charge is placed on gate line, electrons (or negatively charged particles) will begin to collect in area above gate, on other side of Silicon Nitride (SiNx) in a-Si. When charge on gate is increased to a certain point, called VT or threshold voltage, enough electrons will have collected in a-Si to change it from an insulator to a conductor. In other words, you build up a channel of electrons, so if there are electrons at source (high) and nothing at drain (low), electrons will begin to move through electron filled channel until charge is same at both sides or you turn transistor off. The result is a charging of pixel and a change in state of liquid crystals. The unique aspect of this device is nonlinear characteristics after TFT passes through Vt. It exponentially moves to a conduction state (usually 6 to 8 orders of magnitude) and makes it very easy to turn a TFT on or off around Vt value. For more information on MOS FET device operation, pick up a book on Semiconductor Physics or Solid State Physics. The above is only meant as a basic simplified description of MOS device operation. 2.5 TFT Connections The gate line of TFT determines whether or not TFT will pass a charge into pixel. These are controlled by row bus-lines. On a standard VGA display, gate lines would be 480 horizontal lines. The source lines of TFT are connected to column or data bus-lines. These lines provide charge for pixel or contain data for image. The drain lines of TFT are directly attached to ITO pixel, this transfers charge from semiconductor region into pixel. Figure 7: Thin Film Transistor Cross Section ^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^ Source Metal ^ ^ Drain Metal ^ ^ $$$$$$$$$$$$$$ $ SiNx $ *********************************** * a-Si Semiconductor * *** * ** ~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~ SiNx ~ ~ __________________ ~ ~~~~~~~~~ | | ~~~~~~ | gate | ________________________________________ glass ________________________________________ 0.0 LCD FAQ introduction Flynn. Some definitions: LCD: liquid crystal display Supertwist: a design that improves contrast Backlit: lamp behind display instead of reflected Electroluminescent: lamp that gives off a cool glow, powered by hi-voltage AC, typically powered with a converter LED: light emitting diode, other popular backlight Extended temperature: modules designed for (surprise) temperatures outside standard range. Standard range is 0 to 50 C, extended range is -20 to 70 C. Extended range units need -7VDC as well as +5VDC. Standard range units need only +5VDC. Driver: LCD modules of all sorts have column and row, or x and y, drivers. To control one of these modules directly from a CPU would take a great deal of time and software overhead, because each bit (dot) has to be written separately, usually 4 dots at a time. Controller: chip, such as an HD44780 or SED1130, which acts as an interface between your CPU and row and column drivers. It may have on-chip ram, or may need external ram. The controller takes care of generating characters, refreshing display, and so on. The modules discussed in this FAQ have integral controllers. 1.0 sources for modules These are sourses I've seen that are most accessible to 2.0 specs 2.1 Pin description These numbers are same no matter physical arrangement of pins, (for instance, a row along top on Optrex units, or a 2x7 .100" center array on side of Sharp units). Pin SymbolLevelFunction 1 Vss GND Ground 2 Vcc +5V Module power 3 Vee note1 Liquid crystal drive 4 RS H/L Register select,H=data,L=instruction 5 R/W H/L Rd/Wt,H=read(mod->CPU),L=write(CPU->mod) 6 E H/L Enable 7 DB0 note2 Data bit 0 (least significant bit) 8 DB1 " 9 DB2 " 10 DB3 " 11 DB4 " 12 DB5 " 13 DB6 " 14 DB7 " Backlight drive: If module has a back light, it will be driven by a pair of pads separate from interface pads. Check your datasheet for power requirements. Electroluminescent strips usually need 100VAC from a DC-AC converter driven off 5V power supply. note1: On standard modules Vee is between GND and 5V, on temperature extended modules it is between GND and -7V. The potentiometer is contrast adjustment. Standard: +5V ------------*----------- Vcc | / 10k to \<---------- Vee 20k pot / \ | GND ------------*----------- Vss Temperature extended model: +5V ------------------------ Vcc GND ------------*----------- Vss | / 10k to \<---------- Vee 20k pot / \ | -7V ------------' Extended temperature types may also employ fancy temperature correction circuitry to provide automatic contrast adjustment. note2: For an 8-bit interface, DB7-DB0 are driven by your CPU on a write but must be switch to hi-Z (or pulled up with pullup resistors only) so module can drive them on a read. For a 4-bit interface, only DB7-DB4 are used. The most significant nybble is written first (bit7-bit4), then least significant nybble is written (bit3-bit0) on next Enable cycle. DB3 to DB0 are left unconnected. The DB, RS, and R/W pins have internal pullups, so open collector drivers may be used with them. Power consumption: Modules use between 10 and 25mW (2 to 5 mA), not counting backlight, roughly proportional to number of rows and columns. Be careful when hooking power to module. Reversing +5V and GND will destroy unit. (Personal experience, eh.) Carefully examine your datasheet to correctly identify Pin 1. 2.2 Character set None of standard ASCII control codes [chr(1)-chr(31), chr(127)] are implemented. Standard ASCII is used for chr(32) through chr(125) '}' <00100000b, 01111101b> The lower case characters do not have decenders. This is because some LCD's (those with 5x7 dots or 5x8 dots) would chop off bottom. Lower case characters with decenders appear near top of character table. You can access them readily by adding 128 <10000000b> if you want decenders and have a 5x11 dot unit that will properly display them. Eight user-defined characters are displayed by chr(0) through chr(7), and redundantly with chr(8) through chr(15). <00000000b, 00000111b, 00001000b, 00001111b> chr(16) to chr(31) are undefined. <00010000b, 00011111b> chr(126) is a right arrow -> <01111110b> chr(127) is a left arrow <- <01111111b> chr(128) through chr(159) are undefined. <10000000b, 10011111b> chr(160) through chr(223) are katakana (Japanese) characters. <10100000b, 11011111b> You might find chr(223) useful, it looks like degree symbol (it's a 3x3 box in upper lefthand corner), and chr(165) <10100101b>, a dot in center of character. chr(224) through chr(255) are Greek and other symbols. <11100000b, 11111111b> 11100000b LC alpha 11100001b LC 'a' with two dots over it 11100010b LC beta 11100011b LC epsilon 11100100b LC mu 11100101b LC sigma 11100110b LC rho 11100111b decending LC 'g' 11101000b radical (square root sign) 11101001b katakana character 11101010b decending LC 'j' 11101011b tiny 3 by 3 'x' in upper left corner 11101100b cent sign 11101101b UC 'L' with 2 horizontal bars(pounds Sterling?) 11101110b LC 'n' with a bar over it 11101111b LC 'o' with two dots over it 11110000b LC 'p' with decender 11110001b LC 'q' with decender 11110010b LC theta 11110011b infinity symbol 11110100b LC omega 11110101b LC 'u' with two dots over it 11110110b UC sigma 11110111b LC pi 11111000b LC 'x' with a bar over it 11111001b decending LC 'y' 11111010b katakana character 11111011b " " 11111100b " " 11111101b division symbol (dash with a dot above and below it) 11111110b blank 11111111b solid black cursor 2.3 Instruction Set Binary data from bit7 to bit 0. If using a 4-bit interface, bit7-bit4 of data are sent first, then bit3-bit0, on sucessive enable cycles. The following instructions are sent with RS (register select) and R/W (read/write) both low. Clear display 00000001 clears display and returns cursor to home position (address 0) Home cursor 0000001x returns cursor to home position, returns a shifted display to original position. Display data ram is unaffected. x=don't care Entry mode 000001ab sets cursor move direction and specifies whether or not to shift display a=1: increment, a=0: decrement, b=1: with display shift. decrement is for languages that write from right to left. On/off control 00001abc turn display on or off, turn cursor on or off, blink character at cursor on or off a=1: display on, b=1: cursor on, c=1: blink character at cursor position Cursor/shift 0001abxx move cursor without changing display data ram, shift display without changing display data ram a=1: shift display, a=0: move cursor, b=1: to right, b=0: to left x= don't care Function set 001abcxx set interface data length, mode, font a=1: 8-bit, a=0: 4-bit, b=1: 1/16 duty, b=0: 1/8 or 1/11 duty, c=1: 5x10 dots, c=0: 5x7 dots Character ram 01aaaaaa aaaaaa=lower 6 bits of ram address to point to, address set i.e. to read or write custom characters. MSB's are always 01, so character generator ram resides from 64 (40h) to 127 (7Fh). Display ram 1aaaaaaa aaaaaaa=7 lower bits of ram address to point to, address set i.e. reposition cursor. MSB is always 1, so display ram is from 128 (80h) to 255 (FFh). Data write operation: RS=1, R/W=0, data on bit7 to bit0 data is written to current cursor position and cursor is incremented Data read operation: RS=1, R/W=1, bit7 to bit0 hi-z (inputs to CPU) data is read from current cursor position Read Busy Flag: RS=0, R/W=1, bit7 to bit0 hi-z (inputs to CPU), bit7=1: busy, bit7=0: OK to send, bit6-bit0 returns current address counter. 2.4 Timing Execution times: Clear display and home cursor 1.64ms, all others 40us, except read busy flag which is complete in a single enable cycle (or two cycles, in 4-bit mode). These execution times mean that after an operation, CPU must do Busy Flag checks until BF (bit 7) is 0, or else wait more than execution time before next operation when connection to module from CPU is write-only. Enable cycle time (TcycE) 1000ns min Operation cycle time cannot be less than 1 microsecond Enable pulse width, high (PWEH) 450ns min Enable pulse must be at least 450 nanoseconds long, no maximum length Enable rise and decay time 25ns max Enable line must change state (L->H or H->L) in less than 25ns Address setup time (tAS) 140ns min Register Select and R/W lines must be valid 140ns before enable pulse arrives Address hold time (tAH) 10ns min RS and R/W must be valid at least 10 ns after enable goes low Data delay time (tDDR) 320ns max When doing a read, return data will be valid within 320ns of enable going high Data hold time, read (tDHR) 20ns min When doing a read, return data will be valid at least 20ns after enable goes low Data setup time (tDSW) 195ns min When doing a write, data on lines bit7-bit0 (or bit7-bit4 in 4-bit mode) must be valid at least 195 ns before enable goes low Data hold time, write (tH) 10ns min When doing a write, data on lines must be valid for at least 10ns after enable goes low Generally, there are no max time requirements on user except Enable rise time. An LCD module can be driven with just toggle switches for data, RS, and R/W, and a debounced pushbutton on enable line. WRITE: ____ ___________________________ ______ RS ____X________valid_RS_level_____X_____ | | | | |<-tAS-->| tAH->| |<-- ____| | | |_______ R/W ____\________|___R/W_low___|____/________ | | |<----PWEH--->| | | |<------------|TcycE-->| |_____________| |____ E _____________/ \________/ | |<--tDSW-->| | -->| |<--tH ________________|_____________|______ dat ________________X__valid_data_X_______ READ ___ _________________________ ______ RS ____X_______valid_RS_level____X_____ | | | | |<-tAS-->| tAH-->| |<-- ____|________|_ _ ___|___|__ R/W ____/ | R/W high | \___ | | |<---PWEH--->| | | |<-----------|TcycE->| |____________| |____ E _____________/ \_______/ | | tDDr-->| |<-- | | -->| |<--tDHR ________________|_____________|____ dat ________________X_valid_data__X____ 2.5 Memory map 2.5.1 Custom characters: I haven't figured out yet how to write custom characters. Please help! Character generator ram appears to reside at 40h to 7Fh in memory. There is room for 8 characters of 8 rows each. 2.5.2 Addressing display ram: 16x1 module is arranged as two 8-character lines side by side. "Line 1" addresses are 80h to 87h "Line 2" addresses are C0h to C7h So, as you write characters to module, cursor will automatically increment until you get to 9th character--you have to move cursor to address C0h before writing 9th character on 1x16 module. 16x2 module is two lines by 16 chars Line 1 addresses are 80h to 8Fh Line 2 addresses are C0h to CFh 20x1 module Line 1 addresses are 80h to 93h 20x2 module Line 1 addresses are 80h to 93h Line 2 addresses are C0h to D3h 20x4 module Line 1 addresses are 80h to 93h Line 2 addresses are C0h to D3h Line 3 addresses are 94h to A7h Line 4 addresses are D4h to E7h 40x2 module Line 1 addresses are 80h to A7h Line 2 addresses are C0h to E7h 2.6 Initialization Modules with Hitachi controllers will properly self-initialize if Vcc rises from 0 to 4.5v in a period between .1mS and 10mS. I suppose an RC circuit would be needed to keep powerup rise time as slow as .1ms, so manual initialization will be required in most applications. If you do use auto initialization, it will come up in this mode: 8-bit interface, 1/8 duty cycle (1 line mode), 5x7 font, cursor increment right, no shift. On most displays, you want to switch to 1/16 duty cycle (2 line mode) because for all but 20x1, there are two logical lines as controller sees it. If you have an 8x11 dot matrix module, you'll want to switch to 5x10 font as well (the 11th line is cursor). 2.6.1 Initialization for 8-bit operation: POWER ON Wait 15ms RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 1 1 x x x x x=don't care Wait 4.1ms RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 1 1 x x x x Wait 100us RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 1 1 x x x x Wait 4.1ms RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 1 1 1 F x x 8-bit operation 1/16 duty cycle F=font, 1 for 5x11 dot matrix 0 for 5x8 dot matrix Wait 40us RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 0 0 1 0 0 0 Display off, cursor off, blink off Wait 40us RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 0 0 0 0 0 1 Clear screen, cursor home Wait 1.64ms RS R/W DB7 DB6 DB5 DB4 DB3 DB2 DB1 DB0 0 0 0 0 0 0 0 1 1 0 Increment cursor to right when writing, don't shift screen Wait 40us INITIALIZATION COMPLETE 2.6.2 Initialization for 4-bit operation: First four instructions are nybbles, then following are bytes, sent on consecutive enable cycles (no delay necessary between nybbles), most significant first, followed by least significant nybble. POWER ON Wait 15ms RS R/W DB7 DB6 DB5 DB4 0 0 0 0 1 1 Wait 4.1ms RS R/W DB7 DB6 DB5 DB4 0 0 0 0 1 1 Wait 100us RS R/W DB7 DB6 DB5 DB4 0 0 0 0 1 1 Wait 4.1ms RS R/W DB7 DB6 DB5 DB4 0 0 0 0 1 0 4-bit operation Wait 40us RS R/W DB7 DB6 DB5 DB4 0 0 0 0 1 0 0 0 1 F x x 4-bit operation 1/16 duty cycle F=font, 1 for 5x11 dot matrix 0 for 5x8 dot matrix x=don't care Wait 40us RS R/W DB7 DB6 DB5 DB4 0 0 0 0 0 0 0 0 1 0 0 0 Display off, cursor off, blink off Wait 40us RS R/W DB7 DB6 DB5 DB4 0 0 0 0 0 0 0 0 0 0 0 1 Clear screen, cursor home Wait 1.64ms RS R/W DB7 DB6 DB5 DB4 0 0 0 0 0 0 0 0 0 1 1 0 Increment cursor to right when writing, don't shift screen Wait 40us INITIALIZATION COMPLETE 3.0 Interfacing Manual test circuit ______ | | GND---*---------| 1 | < | | 10K ><--. | | < | | | +5V---*---------| 2 | | | | `---| 3 | RS | | ,----switch-----------------------| 4 | | | | | +5V----. GND------| 5 | | | | | | # 3.3k pullup resis. | | | Enable | |\ | | *-pushbutton-*--| >o-------------| 6 | | | |/ 74LS04 inverter| | | --- | | | --- 1uF cap | | | | | | | DB0 GND | | *--switch-------------------------| 7 | | | | | DB1 | | *--switch-------------------------| 8 | | | | ~~~ ~~~ ~~~ ~~~ ~~~ ~~~ | DB7 | | *--switch-------------------------| 14 | | |______| | GND 3.2 Interfacing to a CPU bus These modules use an interface like those in Motorola and Zilog systems. R/W on module can come from R/W line on CPU, which is set up about same time as address, while RS can be selected by low order address line A0. Select Enable by ANDing output from your address decoding and E clock. Address decoding can be done with a magnitude comparator like 74LS688, or if you have address space to spare, with just high order address line (hogging a big chunk of addressing space). Because Intel CPUs use separate read and write lines, and they are not set up ahead of time, R/W as well as RS must be set via address lines. Then Enable signal may be generated from (NOT(/RD AND /WR)) AND PSEN). Using data bus method limits CPU clock speed because of tDDR read delay and TcycE and PWEH requirements of Hitachi controller. 3.3 Interfacing to a CPU port Another interface is use of 7 I/O bits from a port. In this case, driving module is very simple in 4-bit mode. Example: | .-------- | | PA0 |-----| DB0 PA1 |-----| DB1 PA2 |-----| DB2 PA3 |-----| DB3 PA4 |-----| R/W PA5 |-----| RS PA6 |-----| E PA7 |-- | | `-------- First, put most significant nybble on PA3-PA0, and appropriate R/W and RS signals on PA5 and PA4, and PA6 low. Then toggle PA6 high. Then toggle PA6 low. Then put least significant nybble on PA3-PA0, then toggle PA6 high, toggle PA6 low. 40 microseconds later, you can send next character. There's no need to worry about cycle timings at all with any but most blinding speed CPUs. Just be sure to notice that toggling PA6 is done by itself, while other pins are held constant. If you have port pins to spare, then an 8-bit interface can be done with 11 port pins (10 for write only). If you don't expect to ever be reading back from LCD, you can conserve resources by grounding R/W, saving a pin, and thus using digital outputs instead of bidirectional ports. Sample in-line assembly code, by Jordan Nicol, for implementation under Dunfield's Micro-C for Miniboard can be found on ftp cher.media.mit.edu It uses port pins PA7-PA3 for 4-bit data and PC6 and 7 for RS and E. 3.4 Interfacing to Parallax BASIC STAMP Sample program with physical hookup described in commented code can be found on ftp site wpi.wpi.edu in /stamp directory. It's a CPU port hookup as described above. The application also involves use of a radio control servo. 3.5 Serial interface A way to save even more I/O space is to use a serial interface, requiring just 3 digital output port pins. It uses a shift register with serial-in, parallel-out, and output latch. ________________ __________ _____ | 74LS595 | | | | | QA|-------|DB4 | |--------|>ser. clock QB|-------|DB5 | CPU | | QC|-------|DB6 | OUT |--------|>latch QD|-------|DB7 | PORT | | QE|-------|RS | |--------|serial data QF|-------|E | _____| | QG|-- | | 10Kohm | QH|-- | | pullup | | | | +5V--^^^---|\Reset | .---|R/W | .-----|\OE | | |__________| | |________________| GND GND This method (which could be adapted to turn any 8-bit digital output into up to 64 8-bit digital outputs, albeit at a slow speed) uses just three pins. It does take more processor time to implement, but it is "care free" because it will take more than 40us for most controllers to send a byte this way, so a whole screen rewrite could be done without worrying about timing. The Amateur Robitics column in June '94 Nuts & Volts demonstrated how to use this technique with 68hc11's SPI port, using MOSI, SCK, and /SS. This would be especially handy with a non-networked Miniboard, which has MOSI, MISO, SCK, and /SS conveniently routed to top left corner where resistor pack 2 goes. The experimenter could put contrast potentiometer and latch on a daughterboard mounted underneath LCD module. 4.0 Notice /* * Trademarks and service marks appearing in this FAQ are property * of their respective owners. Copyright 1994 by Christopher Burian and * other contributors. All rights reserved. Permission is granted to * distribute and reproduce this FAQ freely as long as this notice and * all attributions remain intact. Please email cburian@uiuc.edu with * any suggestions, additions, or corrections. Have a groovy day.