`
`INNOLUX CORP. V. PATENT OF SEMICONDUCTOR ENERGY
`
`LABORATORY CO., LTD.
`
`|PR2013-OOO66
`
`
`
`E _
`,
`k SOCIETY FOR INFORMATION DISPLAY
`
`
`
`SEMINAR lECTURE NOTES ,
`
`VOLUME I: MAY18 -
`
`ISSN 0887—915X
`
`
`
`SEMINAR M-l:
`
`ELECTRONIC INFORMATION
`
`DISPLAY PERSPECTIVE
`
`(Addressing, Addressing, Addressing)
`
`Terence J. Nelson
`
`Panasonic Technologies, Inc.
`Princeton, NJ
`
`Summary
`
`This seminar is intended to serve as an introduction to the more detailed CRT and flat-panel topics
`which follow in the other seminars, application seminars, applications sessions, and symposium. It
`is intended to provide both the novice and serious display technologist a snapshot in time as to the
`relative status, features,“ and limitations of each of the major direct-View display technologies. The
`technologies covered include conventional CRTs, electroluminescene, plasma, flat CRTs, and LCDs.
`Current status, features, fundamental strengths and weaknesses, and an overview of the most recent
`developments'in each technology will be presented. Technology trends and prospects for the fitture
`will also be reviewed.
`
` SI
`
`SOCIETY FOR INFORMATION DISPLAY
`
`ISSN0887-915X/98/0000-M—1—$1.00 + .00 © 1998 SID
`
`
`
`NOTES
`
`
`
`ELECTRONIC INFORMATION DISPLAY PERSPECTIVE
`
`Terence J. Nelson
`
`Panasonic Technologies, Inc.
`Princeton, NJ
`
`Introduction
`
`Display devicesl play a very important role in today’s information-dominated societies. This is
`because designers of electronic information systems almost always implement the human interface
`visually. Figure 1 depicts the range of common electronic display technologies in a two-dimensional
`display phase space, where the degrees of freedom are the display diagonal and the pixel count, with
`some other non-electronicimage formats.2 Most printers have higher resolution than all displays, so it
`seems clear that display technologyIS still a workin progress.
`In fact, many displays are still1n use that
`- have less than one million pixels. Direct-view displays have also rarely been produced much larger than
`40-inches1n diagonal.
`
`
`
`PRODUCT SEGMENT
`CRT Television
`
`
`Desktop PC
`g09'. 5 ,'U0
`Electronic Camera
`
`
`
`Handheld PC
`
`
`
`
`
`
`
`
`Electronic Projector
`Word Processor
`2.0
`
`
`NA
`
`3.0
`
`0.4
`
`
`1990 (MILLION) 1997 (MILLION) CAGR (%)
`93.0
`167.0
`87
`
`59.
`1
`Ace
`20.
`4.1
`2NA
`
`
`20.4
`3. 6
`8. 6
`
`65. 5
`13. 4
`11.4
`
`
`
`
`
`
`Table 1. World Production of Display-Enabled Products (multiple published Sources)
`
`Table 1 gives recent unit volumes for electronic display products broken into seven market segments.
`Of course, cathode-ray tubes (CRTs) have the lion’s share of the two biggest market segments, direct-
`view television and desktop personal computers. According to the Consumer Electronics Manufacturer’s
`Association (CEMA)3, sales of direct—view televisions in the US dropped 5% between 1996 and 1997.
`However, sales of large-screen (30-inch and up) models climbed 4%. CEMA also estimates that sales of
`unbundled personal computer monitors to dealers rose 9.8% in 1997. At the same time, the average
`selling price increased 1.7%, owing to the transition from 14-inch to 15-inch screens.
`
`Matsushita
`
`
`
`
`
`Company
`Date
`New investment Construction
`Production
`Diagonal
`Reported
`(1000/montl1)
`(inches)
`(billion Yen)
`
`
`2/23/98
`2/18/98
`120
`Fujitsu
`
`
`1 1/20/97
`Pioneer
`11/6/97
`2 (as ofA5797)
`
`
` Mitsubishi
`1 1/6/97
`5
`
`Table 2. Recent Level of Investment in Production of Large-Area Plasma Displays
`
`40
`
`1999
`
`1N999
`
`M—1/3
`
`
`
`The CRT market came under attack in 1997 on both fronts, television and computer monitors, but it
`suffered negligible casualties. On the television front, widescreen 42-inch and NTSC aspect-ratio 40-inch
`color plasma displays with 480 rows of pixels were in limited production at Fujitsu and several other
`companies. Although production to date has been small, large investments continue to be made, as shown
`in Table 2.
`It is not yet clear who will buy flat-panel televisions based on these displays, which cost
`$10,000 or more.
`
`100
`
`1°
`
`—l
`
`P a
`
`0.01
`
`
`
`PbtelCount(Nbixels)
`
`.
`
`10
`
`100
`
`Display Diagonal (inches)
`
`Figure 1. Select Examples of Common Direct-View Displays and
`Image Formats (after Alt [1]).
`
`On the computer front, liquid-crystal displays (LCDs) are attacking the desktop market from their
`established base in notebook computers.
`In the notebook computer market, suppliers are said to be
`moving away from less costly dual-scan supertwisted nematic (DSTN) towards better thin—film transistor
`(TFT) liquid-crystal displays. Dual-scan displays, which use two sets of column drivers, are cheaper
`because it is more expensive to process a thin-film transistor in each pixel, thus creating what is known as
`a matrix of active elements. It is dangerous to declare a winner between passive and active-matrix LCDs,
`however, as the reversal (for TFT-LCDs) that occurred in 1995 shows (see Figure 2.4)
`
`M—1/4
`
` 1
`
`£81] I11 quer
`(8.5% 1")
`
`lanoanapa
`(at-Mi“)
`
`-
`
`.(Mdmasdole not resolvable)
`
`
`
`Figure 3 shows the demand expected for large-area TFT-LCD displays, which depends partly on the
`demand that may develop for desktop personal computerss. Currently 12.1-inch TFT-LCDs are widely
`used on notebook personal computers, and they provide the same viewing area as traditional 14—inch CRT
`monitors. There is already a lot of discussion going on about the price multiplier at which users will
`choose an LCD over a CRT monitor. By March 1998, one vendor was selling a 16.1—inch AMLCD
`monitor for about $3500, which was about 7x the price of an equivalent “flat—square” 17-inch CRT
`monitor.6 A multiplier of 7x is probably unacceptable except in niche markets like financial services,
`where one worker often uses 3 or more monitors. However, the same flat-panel monitor vendor priced a
`14.5-inch AMLCD monitor at about $1400. When this price falls to $999, which is projected to occur by
`year-end 1998, a large market could develop for corporate users who don’t use demanding applications
`like desktop publishing and could use the extra space on their desks.
`
`LCD Demand (thousands)
`
`35'000
`
`IColor STN
`EColor TFT
`
`30,000
`
`25,000
`
`20,000
`
`15,000
`
`10,000
`
`.”Wm-v.1.
`
`5,000
`
`'I'
`
`ul-
`
`'1i.
`1+3
`
`.31"
`6"."
`F‘-
`
`I.
`
`O
`
`1993
`
`1995
`
`1997
`
`1999
`
`1994
`
`1996
`
`1998 2000
`
`Year
`
`Figure 2. Projected Demand for Color STN and TFT
`Liquid-Crystal Displays (after O'Mara [4]).
`
`At the same time that TFT-LCDs are starting to invade the market for desktop monitors, their base in
`the notebook computer market may be in some jeopardy. Notebook computers are wonderfully compact
`for what they cando, often rivaling the power of the user’s desktop machine. They are not cheap,
`however, and they may be overkill for the corporate user who has a personal computer in the office and
`another one at home. This user may be able to get by with a handheld PC on the road for checking his or
`her email, writing first drafts of reports, and searching the web. At least one new handheld PC7 can also
`
`M—1/5
`
`
`
`show slides through a video port and an external display system, which hopefully will be provided at the
`meeting site. Long battery operation is critical in this application, and this requirement may lead to
`further improvements in passive-matrix LCDs, particularly those with color capability.
`
`{Thousands}
`
`(Units)
`
`14,599 .
`
`12.1«inehss
`I} I .34an
`
`
`[Z]. I 0.44an-
`
`
`
`
`.,,..,..,
`
`Eliaxéimhes
`
`
`
`
`1995
`
`1997
`
`1998
`
`1999 Flscal
`
`Figure 3. Large—Size TFT—LCD Demand by Size (this
`graphic is provided by SHARP CORPORATION [5]).
`
`Camcorders and digital cameras need smaller high-quality displays than are used on desktop,
`notebook or even handheld PCs. These displays allow a user to preview an electronic image before
`capturing it in a handheld device or to review images previously stored in memory. However,
`it is
`awkward to compose an image on a direct—view display because the user must focus at the distance of the
`handheld device instead of at the depth of the actual scene. To remedy this problem, electronic
`viewfinders are sometimes used to provide a virtual
`image that the user’s eyes can focus on more
`naturally. A viewfinder display also provides a kind of virtual reality in that the scene changes naturally
`as the user’s head pivots once the user learns how to pan without thinking. Viewfinder displays are also
`not affected by ambient light, which helps maximize battery-powered operation. Active-matrix LCDs are
`generally needed to achieve the quality needed in these handheld applications. AMLCDs made with
`polysilicon thin-film transistors or on single-crystal silicon are sometimes provided with integrated
`drivers to avoid the very- high density of interconnections that would otherwise result, especially for
`viewfinder displays.
`Integrated drivers can also be used effectively in small light valves for portable
`projection systems.
`
`Checklist of Display Attributes
`
`
`In most displays, reflected ambient light reduces contrast and, therefore, the grey-scale range, which
`is the distinguishable number of luminance levels.
`(As a handy rule of thumb, most people will notice
`.luminance changes of 1%.) There are several ways to maintain the contrast ratio under ambient
`illumination, but they always involve trading off some other attribute, usually luminance. CRT screens,
`for example, tend to diffusely reflect light falling on them, somewhat like front-projection screens, in the
`absence of contrast-enhancing measures.
`
`M—1/6
`
`
`
`
`Display technologies generate color in various ways, but almost all of them still depend ultimately on
`powder phosphor materials. Fortunately, the human visual system can be satisfied by mixtures of only
`three primary colors.
`(White light consists of equal optical powers of red, green and blue light having
`luminance ratios of about 2.5:6.5:1.) Because CRTs have dominated the display field for a long time,
`many cathodoluminescent phosphor material systems have been investigated, and rapid technical progress
`is considered unlikely.
`
`The vertical resolution of CRTs is known empirically to be about equal to the number of scan lines
`per unit of vertical height divided by 21/2. This “Kell Factor” is better than the corresponding Nyquist
`factor of 2 probably because the beam profile in a CRT is Gaussian. The reader of display literature is
`also supposed to know that when someone uses a term like “VGA resolution,” he or she is referring to a
`display that has 480 rows of 640 pixels each.
`(It might be preferable to remove the conflict by speaking
`of “VGA multiplication” instead.) The number of rows that can be successfiilly addressed in a given
`display technology without making a dedicated connection for each pixel is known as the multiplexibility
`of that technology. This is very important because dedicated connections are usually not feasible except
`for very small arrays of pixels.
`
`
`It often turns out that a display’s response time is too long to accurately/convey the intended motion
`in the reference image. This can be either because the pixels don’t respond quickly enough after being
`addressed or because the address time per row is long so that the frame time,
`in which all rows are
`addressed once, becomes excessive.
`
`The physical m and weight of a display technology and its power consumption determine whether
`it can be used in portable, battery-powered applications. Notebook computer users, for example, want
`machines that are easy to carry but will continue working without recharging on long trips. Lithium-ion
`batteries can deliver about 10 watts for 5 hours per pound of battery, for example. One state-of the art.
`notebook computer, which has a l4.l-inch XGA (1024 x 768) TFT liquid-crystal display, weighs
`(slightly) less than 10 lbs. in a configuration that runs for about 3.5 hours.8
`
`The efficacy with which electrical power in Watts is converted to luminous flux in lumens is of great
`importance to display technology. Strictly speaking, one has to integrate the luminous intensity over 4El
`sterndians to get the luminous flux and then divide by the power inputP to get this.
`In practice, the
`luminous intensity per unit of projected area of the emitting surface (the definition of luminance) is often
`measured on axis. When the distribution is Lambertian, which means that the luminous intensity goes as
`cos 67in other directions, the luminance turns out to be independent of direction. Then, when the on-axis
`luminance B is given in foot-Lamberts and the areaA is given in fiz, the efficacy will be given by 77 =
`BA/P.
`
`The term “brightness” is often used (loosely) for luminance, and luminance is fiequently represented
`by B in formulas. This is probably because there are so many closely related quantities that could be
`represented by L. The result of the efficacy calculation from the on—axis brightness can be referred to as
`the “efficiency” of the display. Efficacy is troublesome to measure and not really as important as
`efficiency. When the distribution is not Lambertian, the viewing-angle ranges are usually given as the
`horizontal and vertical limits at which the brightness does not fall below 50% of its on—axis value.
`
`Depth and E together often determine whether a technology is suitable for family entertainment
`television and some other group applications.
`In the'US, most people watch television from a viewing
`distance of 8 or 9 feet. The human visual system peaks at about 3 cycles per degree, and the smallest
`feature that can be seen is about one arc-minute, so a pixel size of less than about 1 mm won’t be
`resolvable. When a video source with about 480 lines of resolution is being used, the display doesn’t
`
`M—1 [7
`
`
`
`have to be bigger than about 18 inches high or 30 inches in diagonal (assuming an aspect ratio of 4:3).
`The depth of 31 and 32-inch television receivers in the US market in 1997 averaged about 15% larger
`than their picture heights.9 This seems to be acceptable in view of the fact that many fumiture items are
`at least 18—inehes deep, even including typical subwoofer speakers for home theater installations.
`
`The traditional rule of thumb for the price of a display is 1/3 glass, 1/3 electronics, and 1/3
`distribution and profit. We refer to the bare display device cost as ggg and consider that the cost of the
`electronics depends on the voltage levels that drive the display device. Another rule of thumb says that
`the OEM cost of the display will usually be one-fifth of the price the end user pays for the box. If the
`display cost is a larger fraction, it will seem expensive, and the system designer will usually add more
`functions or choose a cheaper display. The bare display device cost also depends on the sim lici
`, or
`lack thereof, of the precessing steps required to make it. One important component of this cost is the
`interest and principal per display of the processing equipment.
`
`
`Display Attribute
`Brief Explanation
`
`
`ambient light
`Luminance of dark pixels in a lighted environment limits contrast
`tolerance
`
`
`
`
`
`Quality of red, green and blue primary colors
`
`Luminance ratio without ambient light
`
`Bulkiness of display device
`
`Conversion of electrical power
`Lumens per Watt.
`
`into luminous flux, expressed in
`
`color range
`
`contrast
`
`depth
`
`efficacy
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`glass
`
`Cost of materials in display device
`
`gray-scale range
`
`Controllable luminance range in a dark environment
`
`life (operating)
`
`10,000 hours for 3-year replacement cycle
`
`luminance
`
`Often referred to as brightness, and
`
`0.2919 foot-Lamberts = 1 candela/m2
`
`motion (response time) < 30 to 50 msec needed for video,
`
`viewing-angle range
`
`Limited by variation as seen from different positions, vertically and
`horizontally
`
`voltage
`
`weight
`
`Cost of electronic drivers
`
`»
`
`Portability of display device
`
`'
`
`Table 3. Checklist for Evaluating a Display Technology or Application.
`
`M—1/8
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`CRT vs Flat-Panel Shoot-Out
`
`If CRTs are going to have serious flat-panel competition in the next few years, it will probably come
`from plasma displays for home theaters and AMLCDs for desktop computer monitors. Nobody knows
`yet what premium the affluent consumer or the corporate personal computer user will be willing to pay
`for home theaters that look larger and feel
`less cluttered or for valuable desktop space that would
`reappear. But the premium that will be necessary will erode over time provided new customers step
`forward in correspondingly larger numbers.
`In the long run, the better scalability of plasma displays to
`larger area and AMLCDs to higher resolution could be the determining factors.
`
`Table 4 gives some of the specifications of comparable large-area CRT and plasma monitors that are
`suitable for television use. Widescreen (16x9) 40-inch plasma displays are actually more common, but it
`is hard to find a CRT monitor to compare them against. The CRT is much heavier than the plasma
`panels. The plasma panels are themselves heavy for conventional picture hangers, so a secure wall
`mounting might not be totally flush. The inherently digital way brightness is controlled in plasma
`displays causes a contouring effect at low brightness that can be overcome by spatial and temporal
`dithering.
`
`
`
`(MegaView)
`4(Leonardo)
`4xA3
`
`
`Mitsubishi
`CRT
`
`Mitsubishi
`PDP
`
`Pioneer PDP-
`V401
`
`Aspect Ratio
`Brightness
`(ed/m2)
`Color Range
`
`210
`
`400
`
`RGB analog
`
`18-bit digital
`
`24- bit digital
`
`Contrast (0 lux)
`
`Contrast (200
`lux)
`
`Depth (inches)
`Life (hours of
`operation)
`Power (W)
`
`Stripe Pitch
`
`NA
`
`27
`
`15AO:1
`
`44
`
`1NSO:1
`
`200:1
`
`380
`
`0.96 to 1.1 mm 1:1.26
`
`1.26
`
`300
`
`1.32 mm
`
`Plasmaco
`PDP
`
`
`
`
`(prototype)
`4x3
`300
`
`
`18-bit digital
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`(mm)
`Viewing Angle
`(degrees)
`Viewing Area
`(diagonal
`inches)
`Weight (lbs.)
`
`
`
`
`
`
`
`
`
`
`
`
`
`Lambertian
`
`160
`
`40
`
`'
`
`40
`
`160
`
`40
`
`160
`
`42"
`
`Table 4. Specifications of Large-Area CRT and Plasma Displays for Television
`
`Figure 4 shows an audio-technologist’5 view of the important components of a home-theater
`installation. The display shown here seems to be about the same depth as the sub woofer, which could
`be about 18 inches. A depth of 18 inches15 possible with a 27-inch CRT, but a DVD player can provide a
`a video signal with about 500 lines of resolution that looks stunning on a 40-inch screen.”
`
`Table 5 gives some of the specifications of middle-range CRT'2 and AMLCD monitors for personal
`computer use. While CRT monitor depth has been reduced”, the depth and weight of the packaged
`AMLCD monitors14 are not as small as one might think. The excess is presumably mostly1n the base to
`
`M—1/9
`
`
`
`> Mitzi 71.6212“ Radnanc
`fithfiiMiinfiQr’f V
`V
`
`'
`..
`
`
`
`7, 12
`30mm
`
`iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiNA
`w.
`' 23
`’
`figambaftiah '
`~
`~
`.
`
`
`v
`
`(fig!
`
`
`r353“ Spfiéificaimnsef I‘lmch CR”? sadfiMLQ Galfiwmrfionimrs,
`
`'5‘
`
`
`
`M—mo
`
`
`
`Shadow-Mask Color CRT
`
`Neat Tricks
`
`The basic CRT structure15 as shown in Figure 5 comprises an electron gun that emits a beam of
`electrons that comes to a focus at a screen. Along the way, magnetic deflection coils steer the beam
`towards the desired spot on the screen. Finally, a phosphor coating at the screen emits light when the
`electrons land on it. The beam current is modulated in the gun and causes a spatially varying brightness,
`or an image, to appear on the screen. The problem of sapplementing these basic components to produce a
`color system was famously solved when the head of RCA Labs promised all the resources of the company
`and demanded a working prototype within six months”. Most television tubes now use one of the two
`variations of the shadow mask approach shownin Figure 6. The neat trick is that the phosphor stripes are
`photodefined using the same shadow mask that will be installed in the tube when it is shipped. This way,
`good registration between the openings and the phosphor stripes can be maintained over a large area.
`
`In the Trinitron,® a vertical aperture grill is used that is not self supporting, but it also does not
`deform when the electron beam heats the metal stripes. Heating is one of the problems with shadow—
`mask CRTs because only about 20% of the beam current actually passes through the openings to strike
`the phosphor layer. Thus about a factor of S in brightness is traded for color capability. The Trinitron
`design is also used for computer monitors. However an approximately hexagonal array of circular holes
`is used in a shadow mask of a second type of computer monitor that also has an inline gun.
`
`Tension band .
`
`-
`..
`.wShadow mask
`Magnetic shield
`
`Anode contact
`.
`
`Electron beam
`
`Deflection yoke
`
`Electron un-.
`Magnet assembya :
`
`Base ‘.
`~
`“5
`
`,. Phosphor
`
`.Panel
`
`Figure 5. Major Components of Color CRT (after Hitachi [15]).
`
`The phosphor dot arrangementsl7 shown in Figure 7 have a shorter horizontal pitch than a stripe
`pattern, which is better for showing text. The nearest neighbors of the conventional pattern are displaced
`along lines running at 30 degrees to the horizontal, which is also an advantage. Actually, the two
`
`® Trinitron is a trademark of Sony Corporation.
`
`M—1/11
`
`
`
`advantages may be different aspects of the same phenomena if the hyperacuity of the human visual
`system to horizontal and vertical pixel structure is learned from reading. The modified dot structure
`shown on the right-hand side of Figure 7 has an even shorter horizontal pitch. This was accomplished by
`increasing the vertical pitch, which'IS still shorter than the horizontal pitch. With these adjustments, 2M
`pixels were achievedin a tube that15 only 19 inches long”.
`
`It should be pointed out that, in CRT displays, a unit cell of the phosphor dot array is not a pixel. A
`variety of methods can be developed for multiplexing pixels, which means that pixels are selected and
`updated by two or more mechanisms working together.
`Some simple displays can use a separate
`connection for each pixel.
`In high-infomation—content displays, however, the number of connections
`must be much smaller than MV where M is the number of pixels in each row and N is the numbers of
`rows of pixels in the display. The CRT’s long dominance is probably due to the simplicity of addressing
`a pixel by means of currents applied to horizontal and vertical deflection coils and data signals applied to
`the three guns. Thus the number of connections doesn’t grow at all.
`
`electron
`
`p/ guns \
`\{1
`direction of
`
`face plate
`scanning lines phosphors
`
`n glass
`
`in-Iine gun trinitron“
`
`precision in-line system
`
`Figure 6. Trinitron Linear Grill and Vertically—Slotted Shadow-Mask Systems (afier Herold [16]).
`
`Sequential addressing is a mixed blessing, however, as the temperature rise of the horizontal
`deflection coil becomes a problem when images are displayed in high-resolution formats at high frame
`rates. Figure 8 shows the reduction of this temperature rise that was recently reported13 where the surface
`area of the coils was increased by a factor of 2.9. Of course, even though the temperature rise has been
`mitigated with by this design, the power dissipated in the horizontal deflection coil is still high and
`adversely affects the efficiency comparison of the CRT with comparable LCDs. It is probably fair to note
`that the technology is limited by the addressing mechanism temperature rise.
`In this case, as in most
`others, the temperature rise is directly related to the addressing mechanism.
`
`M-1/12
`
`
`
`Conventional 0-28 mm
`
`New EDP CRT
`
`dot pitch CRT
`
`
`
`30 4050 60
`
`70
`
`80
`
`90100110
`
`Hor. deflection frequency (kHz)
`
`Figure 8. Deflection Yoke Temperature Rise (afier Kato [13]).
`
`M—1/13
`
`
`
`Matrix Addressing
`
`In most high-information content displays other than CRTs, an unmanageably large number of drivers
`and connections is avoided using time multiplexing, where pixels are addressed, individually or in small
`groups, in a specific sequence in time. In particular, matrix addressing allows pixels to be addressed with
`N row drivers and M column drivers, usually by selecting one row at a time to receive data from the
`columns.
`In many cases, an array of pixels is formed quite simply by bringing two substrates bearing
`linear arrays of conductors together at right angles with an unpattemed display medium in between.
`Matrix addressing uses all the connections that are made accessible when the display fabric is formed in
`this self-aligned way.
`’
`
`or;
`
`«0
`
`\l«\
`
`0‘
`
`0
`
`0
`
`O
`
`\1
`
`‘6
`
`Qt
`
`V
`
`m
`
`6'
`
`(v flithvoi
`th
`m
`GND
`
`+03
`(U
`t
`
`c O0
`
`L
`0
`
`GND
`
`E'
`
`9
`
`Vth vo lta ge
`
`Figure 9. Matrix Addressing with Nonlinear Display Medium.
`
`As shown in Figure 9, matrix addressing requires a nonlinear display medium to suppress the
`response of pixels in a given row to data intended for any of the other rows. This is a severe requirement
`for high information content displays where N is at least about 240 and may be over 1000. Some kinds of
`displays degrade when the drive waveform has a dc component. Fortunately,
`it is usually possible to
`alternate the polarity of the waveform periodically. The two choices indicated in Figure 9 allow dc-
`cancellation with unipolar data drivers. This is usually an advantage because the data drivers are more
`complex because they have to output many voltage levels in order to generate gray levels.
`
`The row and column waveforms are shown in more detail in Figure 10, which also indicates that the
`rows are floated when they are not being addressed. This helps when charging the pixel capacitance to
`the modulation voltage puts an excessive capacitive load on the data drivers. As will be discussed below,
`this capacitive loading is pattem-dependent.
`
`M—.-1/14
`
`
`
`Vm+Vm
`
`row n
`
`gnd
`
`floating
`______________
`
`floating
`
`gnd ________________
`floating
`
`row n+1
`
`"on" column
`
`"off" column
`
`
`
`
`
`write time
`
`write time
`
`row n
`
`row n+1
`
`Figure 10. Waveforms for Matrix Addressing Using Unipolar Data Drivers (after
`King [34]).
`
`M—1/15
`
`
`
`Active-Matrix Liguid-Cflstal Displays
`
`An active matrix19 comprises a transistor, or a network of diodes, in series with each pixel. These .
`elements are incorporated primarily to create a sharp threshold to improve the multiplexibility of the
`display.
`The active matrix also prevents charge from leaking away from pixels until their rows are
`selected again, making it possible to display a gray-scale range by varying the voltage at each pixel.
`Active elements unfortunately require complex processing steps that raise the cost and lower the yield of
`acceptable panels.
`
`data
`lines
`
`liquid
`crystal
`
`common
`electrode
`
`l/
`
`pixel
`electrode
`
`—).
`
`—>
`
`—p
`
`back ->
`light "
`_>
`
`—>
`
`—>
`
`
`
`gate
`lines
`
`glass
`substrates
`
`Figure 11. Basic structure of thin-film transistor liquid—crystal display
`(after Morozumi [19]).
`
`Figure 11 shows the basic structure of a thin-film transistor (TFT) LCD. The row electrodes are used
`to turn the transistors on, and the column electrodes are used to charge the pixel capacitances in the
`selected row. The row and column electrodes are formed on the same substrate and therefore must be
`aligned accurately.
`
`Figure 12 gives some idea of the sophistication that goes into the processing of thin-film transistors
`for high-performance displays.20 This design used compound layers for the gate oxide and the source-
`
`M—1/16
`
`
`
`drain contact to improve their performance and to compensate for defects in that might occur in any one
`layer. Moreover, leaving the ITO under the source—drain lines improves yield because one layer can
`continue to carry current when the other is compromised by defects. Note that the gate metal was also a
`composite thin-film structure.
`
`uC-Si(n+)
`
`tic—sum)
`
`pafijg’gfib"
`
`bus #2
`
`pixel
`electrode
`
`
`
`source
`bus #1
`
`(ITO)
`
`_
`Insulator #1
`
`‘
`gate
`(TaN/Ta/TaN)
`
`(T8205)
`
`gate
`
`insggfitrgr) #2
`x
`
`Figure 12. Cross section of TFT structure used to achieve 17-inch workstation
`AMLCD (after Kawai, et al. [20]).
`
`Figure 13 is a schematic diagram for a design21 that uses diodes instead of transistors. Normally,
`diodes are not as effective as transistors because the forward voltage of the diode affects the voltage
`transferred to the pixel, and variations in the forward voltage become visible. In this case, however, an
`em conductor is provided at each column. These extra column conductors are held at a reference
`voltage through a single extra connection.
`
`The waveform shown in Figure 14 is applied to the row electrode. When the reset pulse is applied to
`the row electrode, current flows through the pixel capacitance and diode D2 to the extra column
`conductor. The row voltage then drops, and some of the charge that was stored on the pixel bleeds off
`through diode D1. The voltage on the pixel capacitance stabilizes at a value that is determined by the row
`
`and column voltage, but the forward voltage of the two diodes cancel. Finally, the row voltage rises to a
`value that cuts off conduction while other rows are addressed. Because a dc bias would degrade the liquid
`crystal, the waveform is inverted in alternate frames.
`
`Many interesting active-matrix LCDs have been developed in the last two years. The best contributed
`paper award for SID 96 went to Handschy and co—workers at Displaytech for a virtual-image display”. In
`this case, the active matrix was made on real silicon, and the light valve was made with a ferroelectric
`
`M—1/17
`
`
`
`
`
`column 1
`(data)
`
`column 2
`
`Figure 13. Double-Diode and Reset Circuit (after Kuijk [21]).
`
`Vreset
`
`<—— Tframez20msec—->
`
`Vs2
`
`Vn32
`
`---:iVns1
`
`<-— Taddz64usec
`
`—>
`
`Vs1
`
`Figure 14. Waveform Applied to Row Electrode of Double-
`Diode and Reset Circuit (after Kuijk [21]).
`
`liquid crystal. The fast switching response of this type of liquid crystal allowed 15 bits of color and gray
`scale to be generated in the time domain. Ferroelectric liquid-crystal cells are unusually thin, which is
`probably a disadvantage for direct—view displays, but it could be an advantage for high-resolution virtual
`
`M—1/18
`
`
`
`margedisplays {11this 121156? 1113c1111"g11p-;1~11$11111313111311; '15 @111,
`
`and W11 diffemm3’3611’ 2561 111111111113
`
`
`V
`'
`'
`'
`'
`H
`i
`V
`"
`'
`‘
`'
`“13181111” redgzaen 2111111 131111;: LEDz
`
`was; 115131111111liummate1113-1515111
`)1;
`
`
`
`
`
`
`
`
`
`
`Multiplexed Liguid—Cgstal Displays
`
`As indicated in Figure 17, multiplexed LCDs are basically simple devices containing linear arrays of
`transparent electrodes on opposing sheets of glass. The configuration of the molecules is partly controlled
`by alignment layers, which are thin polymer coatings that have been applied to the two substrates and
`then rubbed in specific directions. Various liquid-crystal configurations are created in this way. The
`molecular configuration also depends on the voltage difference between the two substrates. The
`molecular configuration,
`in turn, causes the polarization of the light passing through the cell to be
`transformed. In the conventional twisted-nematic configuration, the polarizers and the rubbing directions
`are aligned at each surface. The polarization follows the 90-degree twist in the liquid-crystal orientation
`in the off state. In the on state, the liquid-crystal partially aligns with the electric field along the normal to
`the substrates. The light then substantially retains its initial polarization on passing through the liquid
`crystal and is absorbed by the exit polarizer.
`
`alignment
`
`
`
`pol.
`
`comp.
`film
`
`comp.
`film
`
`pol.
`
`Figure 17. Structure of Typical Passive-Matrix Liquid-Crystal Display.
`
`The potential difficulty with multiplexed LCDs is that pixels in a given row experience the correct
`data voltage once in a frame time and incorrect data voltages for N-l times. The only difference is that a
`select voltage is also applied to the row conductor for that one critical time. What is the best that one can
`do in this situation? In the beginning, it was known that the polarity of the applied voltage doesn’t matter
`except that the dc component had better be canceled.
`It was thought that the data voltages should be
`chosen to have opposite polarity, say iVo so that the interfering portion of the waveform would always
`
`M—1/20
`
`
`
`have the same magnitude, and the asymmetry that occurs when the select voltage is present would
`determine the response of the liquid crystal. This “3-to-l” addressing scheme works to some degree, but
`it is not optimum.
`
`nom‘al polarity i reversed polarity i
`
`normal polarity
`& offset by D
`
`reversed polarity
`g 8. offset by 3
`
`3+0
`
`_..-—-----—____-
`_--——-.nu-a---_—-
`
`....._.._..——-v—a--
`
`
`
`
`S
`
`
`
`
`
`
`
`--u-———--——-——_---.
`
`
`
`S+D
`
`-(S+D)
`
`
`
`
`.-—..-__—u-——.u-..----—.
`
`
`
`
`
`—(S+Dl
`
`Figure 18. Cancellation of dc—component with Reduced Voltage Range for Row Drivers
`(after Scheffer and Nehring [241).
`
`Figure 18 shows the type of row and column voltage waveforms that are usually applied to liquid-crystal
`displays“. Complete dc cancellation is necessary, so bipolar column drivers normally be used. However,
`this doubles the volt