Data- and Video Projectors

Autor: Mag. Erhard Ipp, current research and development director of Science-Vision, the research division of In-Vision.
Published in the annual report of the HTBLuVA Mödling, 1999.

About 90% of the information from the environment are perceived by the human eye. Of course this has to be regarded in everyday teaching in classes, and so many teachers labor with blackboards or overhead projectors to visually provide study content for their students. Occasionaly, unfortunately much too rarely, there is a video that suits the study content. Here trouble starts: TV sets are broadly available, but what is the viewing angle for the student from the last row? At least he won't be able to recognize details. Let's therefore go into the cinema and turn on the video projector. This takes time and is not always possible, but no student is disadvantaged.

But how does such a video projector work, and why can not every classroom have its own projection device? The second quesiton is easy to be answered: The projectors are (still) too expensive and/or (still) provide poor intensity of light for a classroom that is not shaded. The question about the functionality is not as easy to be answered. In recent time various companies brought projectors to the market that are based on quite different functionalities. Up to a few years ago, video projectors were almost entirely CRT projectors, that are projectors where the TV picture is separated into the three basic colors, which are displayed on cathode ray tubes each and projected by three lenses on the screen. This technic has a couple of disadvantages. Starting with the difficulties to bring the three single images on top of each other when changing the position, up to the physically caused connection of the picture quality with the intensity of light: Because of the mutual repulsion of electrons, an increase in intensity of light, that is only to be achieved by a more intense electron beam,  inevitably leads to a bigger pixel and therefore to a lower resolution, that is the sharpness of the image. This system is neither suited for day light projection nor for high definition TV (HDTV).

Modern video projectors are so-called light-valve projectors. In general, light valves consist of a controllable substance that imposes the image information to the luminous flux from a light intense projection lamp according to electronically applied signals. One possibilty to influence the luminous flux by the image is the liquid crystal display (LCD), and various companies brought projectors based on this technology for special applications as well as for the consumer sector to the market. Most of these projectors are based on slide projection, that is the light of the condenser is directed through the LCD and influenced for the image information. The principle is the same as for a slide or film projector where the slide or the film strip is replaced by a LCD. Since LCDs only can operate gray tones, for the color projection usually a system of 3 LCDs is used. Here the white light from the condenser is first separated into the three basic colors (red, green and blue), and after passing the LCD is mirrored together by a prism and projected by a common lens. The most light intense of such projectors can reach up to 1000 ANSI-lumen on the screen and would be quite suited for the classroom. Further increase of the luminous flux is limited though: The functionality of the LCD is based on tilting the polarization plane - useless light (black parts of the image) is absorbed in the polarization foil. Since light is a form of energy, it causes heating of the polarization foil that generally is glued to the LCD. And LCDs only work up to a temperature of about 65°C.

LCD projectors that use color LCDs are older and therefore cheaper, but bigger and less intense in light.

Apart from these systems which are based on the principle of slide projectors, there are also methods based on epiprojection. The oldest projector of this kind, capable of large-screen projection, is the so-called Eidophor projector. It contains a shperical reflector with an oil film that is deformed according to a  charge pattern written by an electron beam. The light intensity is not reduced by reflection, but its direction is changed slightly (phase-modulated) on those spots with a deformed oil film. The conversion from phase modulation to intensity modulation (projection image) is accomplished by an optical streak dark field system. Simply spoken, those light beams that were not changed in direction are reflected back to the lamp, and those that changed direction due to the deformed oil film go through the projection lens and cause an according pixel on the screen. Since the light intensity only depends on the bulb used, large images with high resolution are possible. And since the light that is not for the image is reflected and not absorbed, there is no heating. Such projectors were used in the past few years for example during the "Wiener Festwochen" in front of the city hall in Vienna for the projection of operas and concerts.

Recently, there has been intense research in the field of light modulation that should eventually lead to a projection system similar to Eidophor projectors. One of these methods is downright the ideal combination of mechatronic and optical principles. Since this system is on the verge of bringing on to the market, I want to describe it here in a little bit more detail. The heart of this projection system is an arrangement of mirrors that consists of a large number of tiny single mirrors. This arrangement is called DMD (Digital Micromirror Device). Each single quadratic mirror has a length of 16µm (even more recently 13µm), and is supported by torsion springs. In figure 1 you can see the arrangement of two such mirrors. In the idle state, that is without activation, the mirror planes of the single mirrors (in figure 1 drawn transparently) are arranged parallel to the base. If they are controlled by electronics (CMOS matrix), they can be tilted into two depicted directions: In one end position they are tilted by -10° (recently -12°) (off state), and in the other end position they are tilted by +10°(+12°) (on state). The tilting angle is limited by the stop position. The mirror tiltings are caused by electrostatic attraction, caused by the voltage difference between mirror resp. moving electrode and the corresponding memory component on the CMOS matrix.

Starting point for the production of a DMD is the creation of a CMOS circuit on an according substrate. A relatively thick oxid layer is applied to the last metal layer of the CMOS and using the special chemical mechanical polishing method (CMP) it is exaclty planar polished. This is necessary in order for the single mirrors to have exactly the same conditions for the procuction.  The creation of the micro mirror structure is achieved using six photo masks with witch the electrodes, the torsion springs, the mirror as well as distant pieces are formed out of aluminum using thin film technical methods. To produce the necessary distances between the aluminum layers, additional layers made from organic material are also applied. The structure of the aluminum layers is formed by plasma etching using a SiO₂ - etching mask. After completion of the last layer (mirror layers) the organic layers are dissolved again by plasma etching. At last, a thin oil film is applied so that the stop positions of the moving electrodes will not stick to the support by adhesive forces during switching.

Figure 2 showes the structure of a micromirror retainer yet without the mirror as an image of a SEM (Scanning Electron Microscope). You can see clearly the torsion springs and the stop positions at the vertices of the moving electrode.

Optical Principle:

The DMD can be understood as a "light switch" (SLM, Spatial Light Modulator) which either reflects the light bundled from the condensing lens through the lens system or or aside from it (DLP, Digital Light Processing system). In fig. 3 this principle is shown for a single mirror. Since the change of light direction is twice as large as the rotation angle of the mirror, the condensing lens can be pivoted by about 20° from the projection axis. It is important that the so-called "0°-light" does not enter the projection lens, since a pixel error would cause a bright, disturbing image pixel.  In the practical construction of course this design is not possible, because the DMD area has to be illuminated uniformly and the light has to be concentrated in the entrance pupil of the projection lens to obtain a bright projection image. Therefore, further optical elements - prisms and/or lenses - have to be arranged.

Different gray tones can be optained by varying pivot durations of the micro mirrors, since the eye integrates repeating light pulses within short times (1/50 second). Since the minimal optical switch time is about 2 µs, there are enough time intervals at a refresh rate of 50 images per second to display 256 or even 1024 different gray tones.

But how can we add color to the system? One possibilty results from the short switch times of the DMD and the fact that the human eye integrates consecutively displayed colors and lets us recognize them as a mixed color. The image information provided by an RGB signal is switched consecutively by the electronics of the DMD and therefore consecutively projected. Between the condenser lens and the DMD there is a so-called color wheel, a tripartite filter wheel, that lets red, green and blue parts of the white light pass through its color segments. Of course an exact synchronisation between rotational speed and position of angle of the color wheel and the controlling of the DMD is necessary so that the DMD will for example exactly switch to the gray tone fraction of the blue image information when the blue segment of the color wheel enters the optical path of the condenser.

Another possibility to build a color projector is using three DMDs. Here the light first has to be split into its three basic colors by a prism system and after reflection by the respective DMD reflected back together. One possible design of such a projector is shown in fig. 4 in a simplified way: The light of a metal vapor or xenon lamp is first aligned parallel by a paraboloidal reflector and directed by a condensing lens to the DMD. In order to keep the distance between the projection lens and the DMD small, the reflection is accomplished by a so-called TIR-prism. "TIR" means "total internal reflection". The angels of the prism are devised in such a way that the light coming from the condensing lens is totally reflected, but the light coming from the DMD can pass unobstractedly. The two single prisms of the TIR prism are not glued to each other, but have a distance of about 10µm. The color separating prisms have to be designed so that the distance from each DMD to the projection lens is exactly the same. Before gluing the prisms of the color splitting system, dichroic coating is vapor deposited. These coatings reflect only a certain spectrum range (blue or red), and let light of different wavelength pass almost completely. For example, in fig. 4 the first color separating prism reflects blue and lets green and red pass through. The blue light reaches the DMD3, is reflected there according to the image information and comes after repeated reflection at the dichroic layer of the first color splitting prism into the projection lens.

The resolution of such a projector depends on the size of the DMD. Since the central distance of two micro mirrors is 17µm, a DMD with a diagonal of 0.7 in. (17.8mm) can achieve a resolution of 800x600 pixels (SVGA). For XGA resolution (1024x768), a diagonal of 0.9 in. (22.9mm), for SXGA resolution (1280x1024 pixels) a diagonal of 1.1 in. (27.9mm) is necessary. Recent DMDs with 13.68µm mirror distance can have SXGA resolution at a diagonal of 0.9 in. Also projectors with HDTV capable resolution (1920x1080 pixels, 16:9 aspect ratio) have been introduced at international fairs. The light intensity is by all means comparable to the abovementioned Eidophor projectors.