The Evolution of Projection Displays. Part II: From Mechanical Scanners to Microdisplays
The development of projection-display systems has encompassed numerous technologies through the decades. CRTs have always played a crucial role, but other critical technologies such as the Eidophor oil-film light valve, Hughes light amplifier, liquid-crystal devices in various forms, diffraction gratings, mechanical scanners, and digital micromirrors have all played an important role in the evolution of this technology. The second installment of this two-part series explores the innovations of the modern era of projection technology, from the 1990s to the present day, exploring how the development of LCOS, DLP, and LCD technologies threatened the dominance of light valves and CRTs. Part I of this series was published in the May 2008 issue of Information Display, which can be accessed at www.information display.org.
by Matthew S. Brennesholtz
FOR MUCH of the 20th century, the key technologies enabling projection-display systems were cathode-ray-tube (CRT) systems and light valves. However, beginning in the early 1990s, myriad new display technologies began to mature, and several of these – such as microelectromechanical systems (MEMS), liquid-crystal on silicon (LCOS), and liquid-crystal displays (LCDs) have had a tremendous impact on projection displays.
Microelectromechanical systems (MEMS) devices have been proposed for television applications for many years. In 1970, RCA Laboratories27 developed and demonstrated a MEMS system that used a dark-field Schlieren optical system and electron-beam addressing similar to the Eidophor or Talaria. An array of stretched-metal membranes were substituted for the oil film.
The MEMS efforts at Texas Instruments (TI) date back to 1977. Early efforts with light control, by analog motion of the mirrors,28were replaced by the digital off-on mirror effect used by TI today. By 1987, TI had demonstrated its first digital mirrors, and their development into a display product was aided by a $10.8 million ARPA grant in 1989.29 The first Digital Micromirror Device (DMD) product was released in 1990, an airline ticket printer with an 840-dots x 1-row DMD.
The first full demonstration of a color-sequential static image on a two-dimensional DMD microdisplay was in May 1992, and an improved 768-dot x 576-row DMD was demonstrated in 1993. By 1994, DMD generated enough interest for the SID Symposium in San Jose to dedicate an entire session (Session 30 with three papers) to it. In the same time frame, TI began to work with consumer-electronics companies30 to develop commercial products based on the DMD.
The first commercial projection DMD product was released in 1996,31 a portable front projector with a single VGA DMD. At this time, the system was given its current name, "Digital Light Processing" (DLP), to cover the entire system, not just the DMD chip itself. This single-chip system was followed by two-chip consumer-home-theater front projectors and three-chip professional front projectors.
Although most previous projection systems, from John Logie Baird through CRTs to oil-film and LCLVs, used analog modulation of the light in the image to produce gray scale, the DLP was only capable of producing black and white, with no gray scale in between. DLP systems produce gray scale by using time-division binary multiplexing where the fraction of the time each mirror is in the "white" state compared to the "black" state determines the gray level perceived by the viewer. This switching occurs fast enough so the human eye would perceive this as a continuous gray scale. In single-chip systems, it was necessary to produce gray scale for each of the three colors sufficiently fast enough so the eye would also merge the colors into a full-color image. The DLP, especially the early models, were only marginally fast enough to produce good gray scale in color-sequential systems. TI has developed increasingly sophisticated dithering algorithms and DSP chips to augment the native ability of the DMD to produce these gray scales.32
Figure 13 shows electron-microscope images of a DMD imager in 2004. Since then, the square holes in the center of each pixel have been reduced in size to improve both contrast and efficiency.
In three-DMD projectors, this gray-scale issue disappears, and the DLP projector is capable of producing very high quality images. In 1999, a DLP projector similar to the one shown in Fig. 14 was used to demonstrate electronic cinema by showing "Star Wars Episode 1" in theaters with paying customers. Most of the electronic-cinema projectors sold today are based on DLP technology, which is now also used in a wide range of projection systems from pocket projectors to electronic cinema – DLP projectors account for about one-half of all microdisplay-based projection-system sales.
Liquid-Crystal Projection Systems
After an abortive start in the 1930s and 1940s, serious work on thin-film transistors began in the early 1960s. There was considerable research into active-matrix systems for direct-view and projection applications from the 1960s through the 1980s. By 1976, for example, Westinghouse had achieved a 6 x 6-in. 20-line/in. LCD, and Hughes had achieved a 1-in.-square 100-line/in. active-matrix LCD on crystalline silicon.33 According to Brody, "The materials used in forming the thin-film circuits are normally metals, such as Al, Au, Cr, Mo, and In; insulators, such as Al2O3 and SiO2; and semiconductors, such as CdSe, CdS, Te (Westinghouse), and PbS (Aero-jet). No obviously superior combination has at yet emerged." At this point, Brody does not even mention silicon as an option! The basic circuit used in some of these early LCDs,34 shown in Fig. 15, was the design that continues into modern 3LCD panels or large-screen LCD-TV displays. Silicon-on-sapphire (SOS), the predecessor of modern high-temperature polysilicon (HTPS) used in most current 3LCD panels, was developed in this same time frame.35
The first LCD projector that can be considered "modern" was introduced by Seiko-Epson in 1986.36 This projector had several design features that would be recognized by a modern 3LCD projection engineer, including a 32-mm-diagonal (1.3-in.) polysilicon-TFT transmissive microdisplay with drivers integrated on the same substrate, a twisted-nematic LC effect, a three-panel unequal path architecture, and a crossed-dichroic color-combining cube similar to the modern one shown in Fig. 16. It had a resolution of 320 x 220 pixels, a 70-lm output from a 300-W halogen lamp, and a contrast of 20:1 – poor performance by modern standards, meaning it certainly represented no threat to the established oil-film projector vendors, or even the CRT vendors. Nevertheless, it set the stage for improved projectors to come.
Fig. 13: Electron microscope image of pixels in a DMD device.
Fig. 14: DLP projector used in the 1999 digital-cinema tests mounted on a Christie xenon-lamp lighthouse originally designed for a film projector.
The uncertainty over the optimum active-matrix technology continued. For example, at the SID Symposium in 1991, papers covering 10 technologies suitable for use as transmissive LC light valves in projection systems were presented:
• a-Si TFTs (Mitsubishi, paper 4.2; Sanyo Electric, papers 20.1 and 20.2; NEC, paper 20.3).
• Ferroelectric matrix (Seiko Epson, paper 4.3).
• Metal-insulator-metal (MIM) diodes (Citizen Watch Co., paper 12.2).
• SiN thin-film diodes (NEC, 12.4).
• SiN bilateral diode matrix (Seiko Instruments, 12.5).
• a-Si double-diode plus reset (Philips Components, paper 12.8).
• CdS TFTs (University of Stuttgart, paper 20.5).
• Passive-matrix STN-LCDs (Citizen Watch, paper 20.6).
• Polysilicon TFTs (Xerox Palo Alto, paper 27.2. This paper discusses both HTPS and LTPS).
• Unified-structure field-induction drain poly-Si TFT (NTT, paper 27.3).
That year, Session 13 was dedicated to "Projection Light Valves," with H. Wayne Olmstead as chair and Akihiro Mochizuki as co-chair. The session introduction stated: "Market demand for large-screen projection displays continues to drive a huge investment in basic R&D for various types of light-valve projection systems. The old adage that 'necessity is the mother of invention' is exemplified in the field of light-valve research. A large number of unique techniques and very creative solutions to light-valve problems have been reported over the years here at SID. This year, there were a record number of papers covering this broad topic. The papers on light-valve projectors were grouped into two sessions. Systems that use 'direct active-matrix' were in Session 20: 'LC Projection Television,' and all other novel approaches were covered in this session."
Barco Blows Oil-Film Projectors Away
The year 1992 was a key year in projection systems, especially in terms of professional applications. It marked the introduction of the BarcoData 5000 projector,37 popularly known as the "Light Cannon." This was the first LCD projector with performance comparable to a light-valve projector and doomed the Talaria and Eidophor. General Electric was so concerned about the effect of the Barco projector on its Talaria business that it bused about one-third of its employees to Infocomm in Philadelphia for the day so they could see the new competition. The JVC ILA also was threatened, but the underlying technology still survives in the D-ILA and other LCoS designs. The basic specifications for the BarcoData 5000 are listed in Table 1.
At $47,500, the BarcoData 5000 was expensive, but not as expensive as a Talaria projector with equivalent lumens. A Talaria projector with about 1250 lm would cost at least $79,980. In addition to the lower price, the BarcoData projector was significantly easier to use. For example, it required 2 minutes to warm-up compared to the 30 minutes to an hour required by oil-film projectors.
Fig. 15: 1974 circuit configuration used to achieve frame storage in liquid-crystal matrix displays.
Fig. 16: Modern crossed dichroic 3LCD prism with LCD micro-displays attached. The projection lens would attach to the fourth face.
Following the Barco Light Cannon, there was a flood of data-projector introductions featuring varying technologies, including primarily LCD, DMD, and LCoS technologies. Extreme cost sensitivity in the consumer market meant that most non-CRT projection systems first penetrated the professional market. "Microdisplay" cannot always be used to describe some of these professional projectors. The BarcoData 5000 and other products in the same series had three 5.8-in. panels. Even larger panels were used, such as the panel in the InFocus LitePro 550 shown in Fig. 17, which had a single 8.4-in.-diagonal panel.
There were three key developments in the years following the introduction of the Light Cannon, allowing the growth of the consumer and professional projector markets:
• Development of the DMD by TI, as already discussed.
• Introduction of the UHP lamp by Philips.
• Introduction of the flat integrator/PCS combination by Epson.
The Philips UHP lamp,38 first publicly discussed in 1995, revolutionized lamp technology for projection systems. All previous lamp technologies had serious flaws: incandescent lamps had very low efficiency, poor colorimetry, and very poor life; xenon lamps had low efficiency and short life; and metal-halide lamps normally had long arcs.
Metal-halide lamps actually represent an entire family of technologies. By adjusting the composition of the fill material, the color of the lamp could be changed to nearly any target color and spectrum. In general, however, the lamps had long arcs, color uniformity problems, and lifetime problems. They were, or could be, very efficient, with efficacies of 100 lm/W or more, and output spectra suitable for projection applications. The arc length and uniformity problems of these lamps were minimized when they were coupled to relatively large, high-étendue LC panels, such as the a-Si panels then available.
The fill material in a UHP lamp is nearly pure mercury, which gives the lamp a characteristic mercury emission spectrum. Unfortunately, this mercury emission spectrum is short of red, so most UHP-based projectors have a relatively poor red primary color. This is considered by most projector manufacturers to be a minor problem compared to the short arc, long life, high efficiency, and other advantages provided by the UHP lamp.
In general, high-performance projectors such as the Eidophor and Talaria used xenon lamps before the UHP was available. The Barco Light Cannon was introduced with a metal-halide lamp. Incandescent lamps, because of their very poor performance, were only used in certain low-performance low-cost projectors, at least after the introduction of the UHP.
Epson introduced the integrator and flat polarization system39 combination in 1997. These were the final design features that, when added to the features Seiko-Epson introduced in 1986, define the modern 3LCD projector engine. This design was used in projector model ELP-3500, which also used 1.3-in. TFT-LCD VGA panels and a 100-W UHP lamp to produce 650 lm. Although this does not seem like much in 2007, in 1997 it competed against projectors like the InFocus LitePro that produced 130 lm from a 400-W lamp. The use of the polarization conversion system largely eliminated the light throughput advantage of systems such as the DLP system that needed only unpolarized light compared to systems such as the LCD and LCoS systems where polarized light was required.
Future of Projection Systems
Sales growth of consumer rear-projection systems has currently slowed to a standstill, due to price competition from LCD and plasma systems. In the larger sizes (52 in. and above), rear projection continues to hold its own, although most marketing forecasts show even the larger-sized rear-projection TV sales declining by 2011. Front projection for consumer home theater continues to grow, as do professional projection applications.
Table 1: Specifications for the BarcoData 5000 projector introduced in 1992.
Fig. 17: The InFocus LitePro 550LS from 1998. This VGA projector used a 400-W halogen bulb, produced 130–155 lum, and was intended to replace an overhead projector.
Two new technologies are on the horizon that threaten the dominance of the UHP and similar lamps in projection systems. First, high-brightness LEDs are beginning to replace UHP-type lamps in certain low-brightness applications, including consumer rear projection and a new category commonly called the "Pocket Projector." The two main advantages of LEDs over UHP lamps are colorimetry and lifetime. LEDs also have very simple drive circuits and can be used in applications where even a 50-W UHP-type lamp produces too much light. On the other hand, the brightness of most LEDs in lm/mm2-sr40 is much lower than a UHP lamp. This generally prevents the use of LEDs in higher-output projection systems. Recently, Luminous Technologies has developed very-high-light-output R, G, and B LED devices using proprietary light-collecting and thermal-management techniques. This product, trademarked "Phlatlight," has seen initial commercial adoption as a light source in some large-screen DLP TVs.
The second technology is really a new category of projectors, often called "Pico-Projectors," which can be designed with LED illumination, although laser illumination is also common. These projectors are currently in the R&D phase and commercial models are expected in late 2008 and early 2009. These tiny projectors are designed either to be integrated inside a cell phone or be stand-alone projectors about the size of a cell phone. It is not currently certain exactly what these projectors will be used for when they are introduced but they will certainly have a major impact on the marketplace.
Laser Projection Systems
Lasers do not have the same étendue limitations as lamps or LEDs and can have luminance output values 105 times or more higher than UHP-type lamps. This would enable very high brightness systems to be built with very small microdisplays, or no microdisplay at all.
Lasers have been proposed for projection systems almost since they were first developed in 1960. For example, a 1969 patent41from Texas Instruments related to laser projectors contained a reference to a still earlier 1966 internal TI memo on the subject.42 To anyone who has examined Scophony drawings from 25 years earlier, this design shown in Fig. 18 will seem remarkably familiar. The main problem with lasers has been, and continues to be, their cost. Several companies, most noticeably Novalux,43 are working on low-cost high-powered lasers for projection systems. For a full-color system, of course, it is necessary to have at least three lasers: red, green, and blue.
In general, laser projector designs fall into one of three types:
1. Flying spot scanners where a single mirror, or a pair of mirrors, does the scanning – Fig. 18 is one example. A modern single mirror MEMs device from Microvision is shown in Fig. 19. For scale, the mirror in this system is 1-mm square.
2. Scanned one-dimensional linear arrays, such as the system shown in Fig. 20.
3. Microdisplay-based systems with a two-dimensional pixel array where the laser is used mostly as a light source in a conventional projection system.
Fig. 18: Laser projector scanner design patented by Texas Instruments in 1969.
Fig. 19: Microvision scanner with a single mirror mounted on horizontal and vertical hinges.
In flying-spot scanner systems such as the ones shown in Figs. 18 and 19, the laser beam is scanned in a raster pattern, or a variation on a raster such as a Lissajou figure. As the beam is scanned, it is modulated at the video rate in a technique familiar to anyone who knows CRT systems. This is repeated at a high enough speed (60-Hz frame rate or higher) so the eye merges the images into full-motion video. For a full-color system, each laser must be modulated separately and the three modulated beams are normally combined into a single white beam. The resolution of this type of system, like the Scophony, is limited by either the horizontal scan rate or the maximum rate at which the light source can be modulated. With modern electronics, the mechanical scanning of the high-speed mirror is more commonly the limit.
In a linear array system, the high-speed scanning is done by a linear array,44 such as the one shown in Fig. 20. The low-speed scanning at the field rate is still done by a moving mirror. Systems of this type can have extremely high resolution. For example, the Evans & Sutherland laser projector (ESLP) has a resolution of 4K x 5K, the highest-resolution image the author is aware of made by a non-tiled projection system.45 The NHK 8K projector,46 with a resolution of 8192 x 4320 (34 Mpixels), has nominally higher resolution, but this is achieved by tiling displays inside the projector and subpixel rendering. It has, in fact, fewer real pixels than the ESLP.
The linear array works much like the oil-film light valves used by the Talaria or Eidophor. The array, which is a MEMS system like the DLP projector or the Microvision scanner, has alternating fixed and moving microscopic ribbons. The array is typically used in a schlieren optical system similar to the ones shown in Figs 21 and 22. When the moving ribbons are in the "up" position, the GLV presents a flat mirror to the laser beam. Specular reflection takes the light back to the schlieren mirror and the source. When the moving ribbons are in the "down" position, the alternate high and low ribbons form a diffraction grating, changing the angle of the light. The light now misses the schlieren mirror, enters the projection lens, and ultimately winds up at the screen. Since they are based on diffraction, which is wavelength-dependent, most linear-array systems use three linear arrays, one each for red, green, and blue.
Laser illumination of microdisplay projection systems provide several major advantages over illumination by UHP or other types of lamps. These include:
1. The very low étendue of the laser beam allows use of smaller lower-cost microdisplays.
2. The low divergence angle of the laser allows the use of low-cost high-f/# projection lenses.
3. Lasers can be modulated at the field rate, eliminating the need for a color wheel in color-sequential systems and increasing energy efficiency.
4. Lasers provide very large color gamuts, larger even than LEDs.
Fig. 20: GLV linear array from Evans & Sutherland. Sony calls a similar array the GxL. Kodak has independently developed another linear array, called GEMS.
Fig. 21: Layout of Scophony mechanical scanning projection system. Note the use of cylindrical lenses because spherical lenses of sufficient size would have been too expensive.
Due to these advantages, laser advocates believe the future of projection systems is tied closely to lasers. But currently, lasers have two major problems that must be overcome:
1. Cost: Current commercially available lasers as of this writing are far too expensive for use in mass-market projectors. Development of laser designs specifically for display applications should end this problem soon.
2. Safety: Lasers are both a perceived and actual safety problem. The issues are much more serious for front-projection systems with 200 lm or more than they are for RPTV, so lasers may be used in rear projection before front projection.
The projection business continues to be an industry full of vitality, with a variety of new projection systems ranging from ones capable of resting in the palm of your hand to complex, multi-projector systems with total pixel counts up to 100 Mpixels. A wide variety of components needed for these advanced systems are also under development, including new microdisplays and scanners, new light sources such as lamps, LEDs and lasers, and the more mundane but essential components such as lenses, polarizers, and filters. This development is spread across the world, with new technologies coming from Europe, North America, and Asia.
As Charles Bensinger said about CRT projectors47:
"Big-screen TV is here to stay, and home VCRs give it a real reason to exist. The next step forward will be solid-state liquid-crystal or CCD-type one-piece wall screens that will be bright and sharp and not need the bulky tubes and optics used by conventional TV projection systems. We should see these devices in about 5 years, but in the meantime, we can have a lot of fun with the big-screen TVs."
Unfortunately, for Bensinger's power of prognostication, he wrote that in 1979. Today, 29 years later, big-screen projection systems are still with us, even for the consumer. On the other hand, he was talking about what would be considered today small systems, ones up to about 48 in. on the diagonal, and in this market category, LCD and plasma TVs have largely replaced rear-projection TV. As was true in 1936, in 1979, and in 1991, the consumer and other end users in 2008 seem to have an insatiable demand for larger and larger television screens, so even with the pressure from LCD and plasma displays, I do not expect to see consumer projection systems vanish in my lifetime.
27J. R. van Raalte, "A new Schlieren light valve for television projection," Appl. Opt. 9, No. 10, 2225–2230 (October 1970).
28L. J. Hornbeck, "128 x 128 deformable mirror device," IEEE Trans Electron Dev ED-30, 539–545 (1983).
29W. W. Gibbs, "Mirror, mirror," Scientific American 110–111 (1994).
30Including Philips Electronics, where the author worked on DMD systems starting in 1993. The Philips developments culminated in demonstration units, such as the one described in D. A. Stanton, J. A. Shimizu, and J. E. Dean, "Three-Lamp Single-Light-Valve Projector," SID Symposium Digest Tech Papers 27, 839–842 (1996).
Fig. 22: Principle of operation of an oil-film dark-field schlieren optical system.
31L. J. Hornbeck (TI), "Digital light processing for projection displays: A progress report," Proc 16th IDRC, 67–71 (1996).
32R. J. Grove, "The MVP: A single-chip multiprocessor for image and video applications," SID Symposium Digest Tech Papers 25, 637–640 (1994).
33T. P. Brody, "Large scale integration for display screens," Proc SID 17, No. 1, 39–55 (1976).
34T. P. Brody, F. C. Luo, D. H. Davies, and E. W. Greenreich, "Operational characteristics of a 6 x 6-in. TFT matrix array liquid-crystal display," SID Symposium Digest Tech Papers 5, 166 (1974).
35L. T. Lipton, M. A. Meyer, and D. O. Massetti, "A liquid crystal television display using silicon-on sapphire switching array," SID Symposium Digest Tech Papers 6, 78–79 (1975).
36S. Morozumi, T. Sonehara, H. Kamakura, T. Ono, and S. Aruga, "LCD full-color video projector," SID Symposium Digest Tech Papers 17, 375–378 (1986); J. Soc. Info. Display 15/10, 773 (2007).
37P. Candry, K. Henry, B. Verniest, and W. Schorpion, "A high-light-output active-matrix TN-LCD projector for video and data-graphics applications," SID Symposium Digest Tech Papers 24, 291–294 (1993).
38E. Schnedler and H. V. Wijngaarde, "Ultrahigh intensity short arc long life lamp system," SID Symposium Digest Tech Papers 26, 131–134 (1995).
39Y. Itoh, J-I. Nakamura, K. Yoneno, H. Kamakura, and N. Okamoto, "Ultra-high-efficiency LC projector using a polarized light illuminating system," SID Symposium Digest Tech Papers 28, 993–996 (1997).
40For a complete explanation of the importance of this factor, called étendue, see M. S. Brennesholtz and E. H. Stupp, Projection Displays 2nd edition (John Wiley and Sons, Chichester, U.K., 2008).
41G. Derderian, S. Seaford, and R. J. Klaiber, "Laser beam deflector," U.S. Patent No. 3,436,546, issued April 1, 1969.
42Texas Instrument Bulletin No. DLA 1324, "Experimental Laser Display for Large Screen Presentation, January, 1966.
43G. Niven and A. Mooradian, "Low cost lasers and laser arrays for projection displays," SID Symposium Digest Tech Papers 37, 1904–1907 (2006). In late 2007, Novalux was bought by Arasor, and the Novalux name is no longer used.
44R. B. Apte, F. S. A. Sandejas, W. C. Banyai, and D. M. Bloom, "Deformable grating light valves for high resolution displays," SID Symposium Digest Tech Papers 24 (1993).
45D. M. Bloom and A. H. Tanner, "Twenty Megapixel MEMS-based laser projector," SID Symposium Digest Tech Papers 38, 8–11 (2007).
46K. Hamada, M. Kanazawa, I. Kondoh, F. Okano, Y. Haino, M. Sato, and K. Doi, "A wide-screen projector of 4k x 8k pixels," SID Symposium Digest Tech Papers 33, 1254–1257 (2002).
47C. Bensinger, The Home Video Handbook, second edition, revised (Video-Info Publications, Santa Barbara, CA, 1979). •