The development of projection-display systems has encompassed numerous technologies since its beginnings in the early 1930s. CRTs have always played a crucial role, but other critical technologies such as the Eidophor oil-film light valve, Hughes's 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 projection systems. The first installment of this two-part article explores the innovations from the 1930s until the early 1990s. In Part II, to be published in the August issue of ID, we'll continue the story to the present day.

by Matthew S. Brennesholtz

THE FIRST electromechanical television system was patented by German scientist Paul Gottlieb Nipkow in 1884, having 18 lines of resolution. It is not known if he built an actual working system. The first recorded working system came in 1902, where synchronized disks rotating at 30,000 rpm were used to produce an image with a 40-Hz frame rate. This would result in a 12.5-line image, except it was not line-scanned; the scan was a spiral. The projected image was so dim it could only be viewed in total darkness. The inventor planned to solve this problem by eliminating the screen and projecting directly on the retina.

While the Nipkow system, as later developed by John Logie Baird, was ultimately unsuccessful in competition against electronic image systems, it established the principle of image transmission by decomposing the image into lines and transmitting the lines sequentially.

Fig. 1: Layout of Scophony mechanical scanning projection system. Note the use of cylindrical lenses because spherical lenses of sufficient size would have been too expensive.

The years 1929–1939 were seminal years for the development of modern television. The question of mechanical vs. electronic TV continued into the 1930s¹ and, indeed, continues today. Certainly, electronic scanning at the camera was the clear long-term winner. To modern eyes, it might seem like the electronic answer would be the obvious one for displays as well. On the other hand, disks rotating at the field rate have never really vanished from the television scene. The field-sequential-color system from CBS in the early 1950s used a rotating color wheel, as did color pictures sent from the moon by NASA in the early 1970s. The rotating wheel continues to exist today in most consumer digital light-processing (DLP) systems from Texas Instruments (TI). In addition, scanning mirrors fundamentally similar to the vibrating mirrors described by Priess in 1936 and 1937 are used in modern handheld projector prototypes from companies such as Microvision. Priess could create a picture 3 ft. on a side at a projection distance of 6 ft. In the Priess system, the lamp was modulated, much like the modulation of the laser in a modern laser-scanning system. Priess wrote in 1936: "I do not believe – other things being equal – that the public will choose a small picture system when they have the opportunity of purchasing a large-picture device. They have been trained to theater and home movies."

The Scophony system,² shown in Fig. 1, was another mechanical projection system developed during the 1930s. Based on two rotating mirrored drums for horizontal and vertical scan, this system featured a light source that operated at a constant brightness and was not directly modulated. Rather, an acousto-optical modulator based on the diffraction of light by sound waves passing through a crystal or fluid was used as a modulator. A dark-field schlieren optical system blocked all undeflected light rays and allowed the diffracted rays to pass on to the scanner and projection screen. The system initially used Kerr cells for the light modulation, but in 1934 switched to the more efficient Jeffree cell. For a 441-line image, the motor driving the high-speed drum needed to rotate at 39,690 rpm. The lifetime of the synchronous motor driving the drum had a nominal lifetime of 1000 hours, although reportedly they actually lasted longer than that. Scophony projection systems intended for public presentations with screens up to 9 x 12 ft. were installed in several theaters. Consumer versions were designed but were never put into production because of the onset of World War II.

Table 1 shows the improvements in television displays from 1929 through 1937.³ In Table 1, the original ratings of candle power per Watt are given, with modern lumen per Watt values in parentheses. The first number is the estimated value if an f/4.5 projection lens (considered a normal lens in 1937) were used while the second value was calculated assuming a modern f/1.0 cathode-ray-tube (CRT) projection lens. While the 5.6 lm/W for a monochrome CRT projector is not competitive with a modern efficiency of about 10 lm/W for a full-color microdisplay projection system, it is still respectable.

Fig. 2: Dr. Law with a 3 x 4-ft. image produced by an RCA projector at the 25th IRE Meeting in 1937. (Photo courtesy of David Sarnoff Library, Princeton, NJ.)

The viewing public largely validated these quality ratings, which were based on the best available televisions in a research laboratory, not the ones used routinely in public. For example, the 1936 Berlin Olympics was televised, but was widely ignored by the German public since the poor image quality sometimesmade it impossible to even recognize the images being broadcast. By 1937, these 441-line systems were referred to as "High Fidelity" and even "High Definition" tele-vision.

In 1937, at the 25th annual meeting of the Institute of Radio Engineers (IRE), predecessor to the Institute of Electrical and Electronics Engineers (IEEE), a CRT projector was demonstrated.⁴ Twelve-hundred IRE members at the Hotel Pennsylvania in New York saw a demonstration of a 10-ft.-diagonal projected image. The monochrome CRT system was built by Dr. Harold Law of RCA Laboratories in Harrison, New Jersey, as shown in Fig. 2. The tube had magnetic deflection, an anode voltage of 10 kV, an image of 1.5 x 2.25 in., a flat faceplate, and a f/4.5 projection lens. The system was normally used to create 3 x 4-ft. images, but for the IRE demonstration it was used to project a 10-ft. image, presumably, with all the room lights off.

The CRT used in this demonstration had a flat faceplate, compared to the convex faceplates of previous projection systems that were needed to provide the glass envelope with suf-ficient strength. The flat faceplate coupled withwhat today seems to be a high-f/# projection lens provided good focus across the full screen. For previous convex faceplates, it had been necessary to compromise center-to-edge focus quality.

By 1937, surprisingly modern CRT projectors were appearing in the literature. For example, M. Wolf⁵ published a paper on a rear-projection system that would be recognizable as a modern CRT projector except it was monochrome. The system used a concave faceplate instead of a convex or flat one. According to Wolf, to obtain satisfactory focus quality on a flat faceplate, the central area no larger than 48-mm diameter could be used. With the concave faceplate designed to match the curvature of the image plane of the f/1.9 projection lens used, Wolf could focus the 48 x 55-mm 405-line image from the CRT faceplate onto a rear-projection screen that could be as large as 100 x 120 cm, although a 40 x 50-cm screen was used more commonly. The tube used 20–25-kV anode voltage and magnetic focus and deflection. With the anode at 25 kV and the grid at 0 V, 400–800 μA could be produced with a spot size of 0.1 mm. The system produced 4–8 lux.⁶ This was considered inadequate by Wolf and Philips, so they used a screen with an estimated gain of 2.5 and provided an estimated viewing angle of ±25°.

Fig. 3: First prototype of the Eidophor large-screen projection system as demonstrated in 1943.

Oil-Film Light-Valve Projectors

During World War II, most television and CRT research was directed toward military purposes such as radar and remote guidance of weapons. However, research into television projection systems continued during that time in Switzerland, a neutral country. Professor Fritz Fischer, working at the Technical Physics Department of the Swiss Federal Institute of Technology in Zurich, built the first prototype of what would later grow into the Eidophor⁷ projector from 1940 through 1943. This prototype, shown in Fig. 3, was demonstrated on New Year's Eve in 1943.

Here is how the eidophor worked. An electron beam wrote a diffraction grating onto the surface of a thin layer of oil in a vacuum. The optical path of both systems was dark-field schlieren, as shown in Fig. 4. When the surface was flat there was no diffraction, the light passing through the oil film hit the schlieren stop and that portion of the image produced a dark area on the screen. When the electron beam, which ran at a constant current, was wobulated with an RF field, the charge distribution on the oil film was non-uniform and the surface deformed under the electrostatic forces. These surface deformations diffracted the light so it missed the Schlieren stop and was projected onto the screen. A Schlieren lens ensured all diffracted light missed the stop and all undiffracted light hit the stop. The projection lens worked with the schlieren lens to image the oil film on the projection screen. Note that the light was diffracted, not refracted or scattered. Regardless of the amplitude of the diffraction grating on the oil film, the diffraction angle was always the same. Low-amplitude diffraction gratings left most of the light in the zeroth order, producing a dark but not black spot. High amplitude produced maximum white.

While, obviously, there was nothing "Micro" about this system, it had all the aspects of a modern microdisplay projection system, including an external light source, a small image to modulate the light, and a projection lens to generate the image on the screen. The Scophony system and other mechanical scanning systems scanned the light and generated the image by modulating the light intensity. In the Eidophor and other light-valve, light-amplifier, and microdisplay systems, the light is not scanned. Instead, the light is steady, illuminates the entire image simultaneously, and the scanning is done electronically in the microdisplay.

The 1943 system used a carbon arc source, the brightest and most compact source at the time (Fig. 4). In 1957, when the first pre-production Eidophor system was under construction, it was switched from a carbon arc lamp to a 1600-W high-pressure xenon lamp introduced by Osram.

In 1953, a color field-sequential Eidophor was demonstrated at the Pilgrim Theater in New York City. 20th Century Fox was impressed and ordered two improved models. The contract for these projectors required the electronics be built by General Electric (GE), so they were shipped to Syracuse, New York, for this work in 1955.

A planned follow-on order for 1000 of these color sequential units was put on hold and ultimately canceled with the introduction of the RCA simultaneous color transmission system. Another factor in the cancellation was the development of Cinerama and Cinemascope, which drew viewers back into the theaters and reduced the perceived need for theater television systems. A final issue would be familiar to modern ears: the theater owners and studios could not agree on how the projectors would be paid for and what type of television content could be shown.

Fig. 4: Principle of operation of an oil-film, dark-field, schlieren optical system.

The prototype simultaneous color Eidophor system had four channels: red, green, blue, and white. Tests of this monster machine where all four channels were in a single vacuum chamber were disappointing: the system could only be used to make a 3 x 4-m image on a 196-in.-diagonal screen. While the system had been paid for by 20th Century Fox, it was never shipped to the U.S. and went into storage.

In 1958, Gretag decided to go into production with the Eidophor. Six pre-production Model ep.1 units were built; four were finished as color-sequential projectors with 500 lm and two as black and white systems with about 2000 lm. This was followed by full production of Model ep.2 with a planned production rate of 5 units/ per month. Gretag did not have the marketing or support networks needed for full production. Therefore, CIBA, Gretag's parent company at the time, set up a joint venture with Philips called Eidophor, Ltd., to market and support the projectors. The first production unit was sold as a color-sequential version to Redifon, Ltd., in England for use in a flight simulator.

Gretag returned to the multi-channel design in 1961, using a more-conventional three-channel system. Gretag designed the basic projector while Philips developed the electronics. This projector is shown in Fig. 5 and produced 2500 lm from the 1600-W xenon lamp, the brightest color-television image that had been shown to date. Note the prominence of the protective cover over the electron gun. Access to the electron gun was needed because the lifetime of a cathode in the Eidophor was at best 100 hours. The cathode could be replaced in approximately 2 minutes by the operator, minimizing disruption of the performance, since the other two color channels continued to operate. With a three-channel Eidophor used 12 hours a day, this failure and cathode change during a performance could occur several times a week.⁸

Gretag continued to bring out new models with improved light output and reduced service requirements. Eventually, they reached 4000 lm from a 2500-W xenon lamp. In 1965, Gretag signed an agreement with JVC to produce Eidophor projectors in Japan. The first projector built in Japan was finished in 1967. Unfortunately, JVC had tooled up to produce the single-channel version, which could be used either as a black-and-white projector or as a color-sequential one.

Demand for these types was dwindling while demand for the simultaneous three-color projectors was increasing. Changes in the oil material used that simplified the design of the projector and made the system more reliable also reduced the response speed of the oil film, so the Eidophor could no longer operate in a color-sequential mode. JVC ceased production after building nine Eidophors.

New models of Eidophor projectors continued to be developed and sold into a variety of professional markets. By 1989, there were about 600 projectors in use worldwide. With the introduction of first LCD and later DLP projectors with comparable brightness and image quality, demand for the Eidophor declined rapidly. The Gretag Display Systems Division closed at the end of 1997, after pro-duction of approximately 650 projectors. AmPro of Melbourne, Florida, acquired Gretag's Eidophor business in 1998. They are believed to have built a few projectors from existing parts acquired from Gretag. Support for all Gretag Eidophor projectors came to an end in June 2000. The last Eidophor was believed to have been removed from service in 2000.⁹

Talaria Projectors

The Talaria projector was developed under the technical leadership of Bill Glenn at GE in Syracuse, New York, and the first systems appeared in 1958.¹⁰ The initial Talaria system was introduced in monochrome and color versions. A 1977 Talaria model is shown in Fig. 6. The Talaria projector was similar to the Eidophor in some ways, and derived some of its technology from the Eidophor during the 1955–1958 time period when Gretag and GE were cooperating. But the Talaria was not just a copy of the Eidophor; it had several major differences, including:

• The light valve was transmissive rather than reflective.

• The light valve was sealed and no vacuum pump was designed into the projector.

• The initial color design used a single light valve to produce all three colors without color-sequential operation.

Fig. 5: Eidophor Model ep.6 simultaneous color projector capable of 2500 lum.

Each of these changes conferred a major cost advantage on the Talaria compared to the Eidophor. The transmissive optical path was simpler and more compact than the Eidophor's reflective path. The sealed light valve, designated the T1, not only eliminated the vacuum pump, reducing cost and making the system more compact, but it increased the lifetime of the system to several thousand hours. It also reduced the warm-up time from the 1 hour reported for the Eidophor to about 30 minutes, since no time was used to pump the system down to a vacuum on start-up. The remaining warm-up time was needed to bring the oil up to its operating temperature.

Since the initial color Talaria was a full-color single-light-valve design, no optical convergence was necessary. On the other hand, the Talaria had about 46 trim pots that needed to be set up correctly in order to get a proper image on the screen, a formidable task for even the most experienced engineer at GE. The single-light-valve design of the Talaria limited it to about 600–1000 lm, depending on the model and the lamp power. There were also serious color and gamma artifacts in the image. The physics of the Talaria was similar to the physics of an Eidophor.

The full-color Talaria operated by wobulating the electron beam with three different radio frequencies, producing diffraction gratings with three different spatial frequencies on the oil film. The color-segmented pupil and the Schlieren bars were designed to have each RF modulate only one wavelength band of light (red, green, or blue). This minimized but did not fully eliminate interactions between colors. Red and blue, in particular, interacted strongly, especially in the mid-brightness levels.

Operation of the transmissive Talaria system shown in Fig. 6 was very similar to the operation of the reflective Eidophor. The effect of the diffraction gratings was doubled because of the double pass, requiring lower modulation of the oil film to achieve maximum white.

To overcome the brightness and color problems, a version of the Talaria with two light valves called the MLV was introduced in 1987.¹¹ This unit had one monochrome light valve dedicated to green and one two-color valve split between blue and red. Each light valve had its own xenon lamp. The two images were then converged at the screen. This dramatically improved the colorimetry of the system because a Talaria light valve could satisfactorily modulate two colors with diffraction gratings parallel and perpendicular to the raster lines. The third color, however, was modulated by a second diffraction grating with a different spatial frequency also perpendicular to the scan lines. These two parallel gratings, used to modulate red and blue light in the single-light-valve version, interacted with each other to affect the color. In the MLV, the red/blue light valve used diffraction gratings perpendicular to each other to modulate the light, avoiding this problem.

The ultimate Talaria contained three monochrome light valves and three lamps. Again, all light valves were screen-converged to produce the full-color image. The 3LV (Fig.7) could produce about 7000 lm at the screen, if a handy 50-A 230-V outlet was available.

Cost was a serious problem with the Talaria projectors, as can be seen in the 1992 MSRP price list shown in Table 2. The light outputs in Table 2 are approximate for two reasons. First, output of Talaria varied from projector to projector. Second, 1992 was before ANSI lumens were commonly used, so the specified lumens for these projectors was not measured by a standard method. For example, the 10k in LV 10k stood for 10,000 "CRT equiva lent peak lumens." This measure was typically 3x–10x what would be measured by the ANSI lumen test method.

Fig. 6: 1977 Talaria PJ5050 Projector, mounted on optional accessory stand. Detachable control unit allows remote operation to 200 ft.

Fig. 7: 1991 Talaria 7000 Lumen 3LV Projector.