Projection TV using DMD & GLV Chips - Seminar Report

Projection TV using DMD & GLV Chips


If you're thinking about assembling a home theatre system, you may be looking at large screen televisions as the heart of your system. Projection TV could give you the size that you want -- CRT screens generally top out at 40" (101 cm) or so, and at that size, they are huge and heavy. Plasma screens can be bigger than that and still manageable, but they can be extremely expensive. Projection TV technology can create large screen sizes at a reasonable price. Or maybe you need to equip a room, like a classroom or conference room, for multimedia presentations with a large audience. A projection TV gives you a lot of flexibility and is usually much better than the standard combination of a 35mm slide projector, overhead projector and TV/VCR.
Projection systems are mainly divided into Transmissive and Reflective projection TVs. In transmissive the Picture is produced when the light source shines through an image. While in reflective projection TVS, the light source illuminates the image formed, and this is reflected onto the screen
Presentations have moved from still pictures to animated, thus relying on the digital media. Projectors of more picture quality have been a requirement. Also with the concepts of ‘Home theatres’ imply for more picture quality than what CRT and LCD projection systems provide.
In the field of reflective projection TVs the recent innovations are Digital Mirror device and Grating Valve technologies. They have been able to produce lager pictures at much higher resolution than the existing CRT and LCD projection systems. Under constant research and designing, these technologies are sure to replace the CRT tube forever.

Digital Micromirror Device (DMD)

Digital Micromirror Device (DMD) developed by Texas Instruments (TI) is a new MEMS-based Digital Light Processor (DLP). The DMD microchip is a fast, reflective digital light switch. It uses standard 5-volt addressing and is fabricated with a monolithic, CMOS-compatible process. It can be combined with image processing, memory, a light source, and optics to form a DLP system capable of projecting large, bright, seamless, high contrast colour images with better colour fidelity and consistency than current displays. DLP systems can also be configured to project images for the production of continuous tone, near photographic quality printing.

DMD Architecture:

 A DMD consists of numerous (10,000 to 2million) micromirrors. The configuration of the array is flexible, depending on the application. Each micromirror is 16 µm square. The array places each micromirror on a 17 µm pitch, leaving a gap of less than 1µm between the micromirrors. This results in a >90% fill factor and is one significant advantage of the DMD.
A single micromirror (pixel unit) can be distinguished to be made of four layers.
1) CMOS Layer
It is the bottom most layer of the DMD. It consists of SRAM cells, one for each mirror. Thus each mirror can be individually addressed.
2) Metal-3 Layer
This layer is just above the CMOS layer. The layer consists of the Yoke address electrode and the Bias reset bus.
3) Yoke and hinge layer
The Yoke and the Mirror address electrodes constitute this layer. The mirror is connected to an underlying yoke which in turn is suspended by two thin torsion hinges to support posts. It is allowed to swing through ±10o from the normal flat position. It is limited with a spring tip, as a mechanical stop.
4) Mirror
The mirror is connected to the Yoke at the centre such that it covers the whole structure. The mirror is made of aluminium, selected as
The micromirror superstructure is fabricated through a series of aluminium metal depositions, oxide masks, metal etches, and organic spacers. The CMOS layer protected with a protective layer, excluding the contact sites. Then the metal layer is deposited over protective layer. A sacrificial layer covers this layer to a height for which the yoke and hinge layer can be deposited. Later the organic spacers are subsequently ashed away to leave the micromirror structure free to move.

Digital Nature of DMD:

A micromirror is said to be ‘ON’ or ‘OFF’ depending to which direction the light is reflected. The optical switching function is the rapid directing of light into and out of the pupil of the projection lens.
The yoke is electrostatically attracted to the underlying yoke address electrodes. The mirror is electrostatically attracted to mirror address electrodes. The direction of rotation is selected by a pair of address electrodes on either side of the rotation axis. The torsion beam rotates until its “landing” tip touches a landing electrode pad that is at the same potential as the beam. Complementary voltage waveforms (Ф1 & Ф2 address) are applied to these electrodes by an underlying memory cell. A bias voltage applied to the beam makes the beam energetically bistable. The result is lower address voltages, permitting larger deflection angles.  The mirror and yoke are connected to a bias/reset bus. The address electrodes are connected to the underlying CMOS memory through via contacts. Movement of the mirror is accomplished by storing a 1 or a 0 in the memory cell (one address electrode at ground and the other address electrode at VDD) and applying a bias voltage to the mirror/yoke structure. When this occurs, the mirror is attracted to the side with the largest electrostatic field differential, as shown in figure. To release the mirror, a short reset pulse is applied to the mirror that excites the resonant mode of the structure and the bias voltage is removed. The combination of these two occurrences results in the mirror leaving the landing site. The mirror lands again when the bias voltage is reapplied.

DMD in Projection TV:

DMD Optical switching principle:

In projection display technology DMD entered as “picture on chip”. In this procedure a single chip projection was used. A bright light source was made incident to the DMD chip, such that in the ‘ON’ position the light would be reflected into the focusing lens. In ‘OFF” position of the mirror the light would reflect outside, onto an absorbing field. Thus on ‘ON’ position the pixel corresponding to the screen would be bright; and ‘OFF’ as dark.


Gray scaie was achieved using a technique called binary-weighted pulse width light modulation. Because the DMD is a digital light switch, its only capability is to turn light on or off. But because of the high switching speed,(order of µsec) it was possible (during each video frame time) to produce a burst of digital light pulses of varying durations that led to the sensation of grey scale as perceived by the viewer.
In the case of colour projection, the same unique feature of speed was utilized, but with Red, Blue and Green colours and more chips. There came three types of projectors, based on economic to high end clarity.

Address Sequence:

The address sequence to be performed once each bit time can be summarized as follows:
1. Reset all mirrors in the array.
A voltage pulse or reset pulse is applied to the mirror and yoke, causing the mirror and yoke to flex. Because this is done at the resonant frequency of the mirror/yoke structure and this frequency is well above the resonant frequency of the hinges, the hinges flex very little during reset.
2. Turn off bias to allow mirrors to begin to rotate to flat state.
During this period the SRAM loads the yoke address electrode. But the mirror doesn’t deflect as bias is absent.
3. Turn bias on to enable mirrors to rotate to addressed states (+10/-10 degrees).
4. Keep bias on to latch mirrors (they will not respond to new address states).
The mirror is at a stable state, as long as the bias is present.
5. Address SRAM array under the mirrors, one line at a time.
6. Repeat sequence beginning at step 1.

Colour Fidelity:

Current DMD architectures have a mechanical switching time of ~15 µs and an optical switching time of ~2 µs. Based on these times, 24-bit colour (8 bits or 256 grey levels per primary colour) is supported in a single-chip projector while 30-bit colour (10 bits or 1024 grey levels per primary colour) is supported in a three-chip projector. Twenty-four-bit colour depth yields 16.7 million colour combinations while 30-bit colour depth yields more than 1 billion colour combinations. Even higher bit depths can be achieved by multiplexing techniques.

Projection Systems:

Single Chip Projector:
The single-chip projector has a colour disc that alternately passes R, G, and B to the DMD chip. Although the singe-chip diagram in figure includes an integrator rod and TIR prism, these may be omitted in lower cost designs. Without a TIR prism, the projection and illuminating lens will mechanically interfere unless the projection lens is offset from the centre of the DMD. The single-chip projector is self-converged, lower in cost and permits the very lightest portable designs.

Two-Chip Projector:
The two-chip projector has a spinning colour disc that alternately passes yellow light (R+G) and magenta light (R+B). The dichroic colour-splitting prisms direct R continuously to one chip and G and B alternately to the second chip. The colour which goes exclusively to one chip is determined by the spectral content of the lamp. Metal-halide lamps have a high colour temperature that produces higher intensities for GB compared to R. Therefore, for that type of lamp, the red is directed exclusively to one chip. This makes up for the deficiency in R and provides the correct colour balance for the projected images.  The two-chip projector provides greater light efficiency and is well suited in applications requiring the very longest lifetime lamps that may be spectrally deficient in the red.

Three-Chip Projector:
The three-chip projector has one chip for each of the primary colours, red (R), green (G), and blue (B). Light from an arc lamp is focussed onto an integrator rod, which acts to homogenize the light beam and change its cross-sectional area to match the shape of the DMD. The white light (W) then passes through a total internal reflection (TIR) prism. The prism adjusts the incidence angle of the light beam onto the DMD so the beam can be properly switched into and out of the pupil of the projection lens by the rotating action of the DMD mirrors. A set of dichroic colour-splitting prisms splits the light by reflection into the primary colours and directs them to the appropriate DMD. The modulated light from each DMD traverses back through the prisms that now act as a combiner for the primary colours. The combined light (R, G, B) passes through the TIR prism and into the projection lens. It is not reflected at the TIR prism because the angle of incidence has been reduced below the critical angle for total internal reflection. The three-chip projector has the highest optical efficiency and is required in the brightest large-venue applications such as trade shows and public information displays.
The light source is usually metal halide because of its greater luminous efficiency (lumens delivered per electrical watt dissipated). A condenser lens collects the light, which is imaged onto the surface of a transmissive colour wheel. A second lens collects the light that passes through the colour wheel and evenly illuminates the surface of the DMD. Depending on the rotational state of the mirror (+10 or -10 degrees), the light is directed either into the pupil of the projection lens (on) or away from the pupil of the projection lens (off). The projection lens has two functions: (1) to collect the light from each on-state mirror, and (2) to project an enlarged image of the mirror surface to a projection screen.


The Grating Light Valve technology is a means for manufacturing high-performance spatial light modulators on the surface of a silicon chip. The technology is based on simple optical principles that leverage the wavelike behavior of light by varying interference to control the intensity of light diffracted from each GLV pixel. A GLV array is fabricated using conventional CMOS materials and equipment, adopting techniques of Micro- Electromechanical Systems (MEMS).

GLV Architecture:

The GLV chip consists of tiny reflective ribbons mounted over a silicon chip. The ribbons are suspended parallel over the chip with a small air gap in between it and the substrate. This constitutes the 1080 pixels arranged linearly. The linear GLV array's 1,088 pixels are at a pitch of 25.5 µm, thus giving a total active area of 25µm by 27.7mm. The linear GLV array is surrounded by four custom driver chips (each with 272 output stages) and assembled into a multi-chip module. The primary function of the driver chips is to provide the digital-to-analog conversion needed for analog grayscale control. A linear GLV array can be used to modulate a single column of image data, while a mechanical scan mirror is used to sweep that column across the field of view
A GLV pixel is an addressable diffraction grating created from moving parts on the surface of a silicon chip. A typical GLV pixel about 25 microns square in area and include six (even numbered) ribbons, each about 3 µm wide, 100 µm long, but only about 125 nm thick. These ribbons are suspended above a thin air gap (typically about 650 nm).

The ribbons are made of flexible silicon nitride, a ceramic material chosen for its high tensile strength and durability. The ribbons are over coated with a thin layer of aluminum that functions as both an optical reflector and an electrical conductor. Integrated-circuit-like package with a clear, optically flat, hermetically sealed glass lid. .

Working of GLV:

These ribbons are suspended above a thin air gap allowing them to move vertically relative to the plane of the surface. The ribbons are held in tension, such that in their unaddressed state, the surfaces of the ribbons collectively function as a mirror. When a GLV pixel is addressed by applying an electrostatic potential between the top of the ribbons and the substrate, alternate ribbons are deflected. Viewed in cross-section (as in figure), the up/down pattern of reflective surfaces creates a square-well diffraction grating. By varying the drive voltage applied—and thus the grating depth—at each pixel, we can achieve analog control over the proportion of light that is reflected or diffracted.
Precise control of the vertical displacement of the ribbon can be achieved by balancing this electrostatic attraction against the ribbon restoring force; more drive voltage produces more ribbon deflection.
Because the electrostatic attraction is inversely proportional to the square of the distance between the conductors, and also because the distances involved are quite small, very strong attractive forces and accelerations can be achieved. These are counter-balanced by a very strong tensile restoring force designed into the ribbons. The net result is a robust, highly uniform and repeatable mechanical system. The combination of low ribbon mass, small excursions (about 1/800 of the ribbon length), and large attracting and restoring forces produces extremely fast switching speeds. GLV pixel switching times have been measured down to 20nsec—three orders of magnitude faster than any other spatial light modulator we have seen reported.

GLV in Projection TV:

In the Scanned GLV Architecture, a linear array of GLV pixels is used to project a single column of image data. This column is optically scanned at a high rate across a projection screen. As the scan moves horizontally, GLV pixels change states to represent successive columns of video data, forming one complete image per scan. The high inherent switching speed of GLV devices makes a scanned linear architecture, and its many benefits, possible. For example, to create a 1,920 x 1,080-HDTV image with a 100 Hz refresh rate, each column of video data is displayed in stasis for about 4.2 µs (assuming a 20% flyback time); this requires a pixel switching time significantly less than 4.2 µs.

High speed operation facts:
The on/off switching speed (or the time required to switch between any other two arbitrary intermediate values) of the GLV device can be several orders of magnitude faster than competing technologies. Specific GLV devices capable of switching speeds as fast as 20 nanoseconds have been fabricated.
The fundamental switching time of the GLV element is related to the resonant mechanical frequency of the ribbon design, determined by such factors as ribbon length, ribbon width, ribbon tension, ribbon mass, composition of the surrounding atmosphere, etc. Because the GLV ribbon is a mechanical element, it can be subject to resonance effects that manifest themselves as a “ringing” characteristic following a step excitation. These dynamic effects can be mitigated through the proper design of electronic drive circuitry and by "tuning" the GLV device and its ambient atmosphere so that it is critically damped at its natural frequency.

Optical working: Analog and Digital
When a pixel is not addressed, the undeflected ribbon surfaces collectively form a flat mirror that reflects incident light directly back to the source, as shown to the left of figure below. When a GLV pixel is addressed, alternate ribbons deflect downward creating a square-well diffraction grating, as shown to the right in the same figure. Varying the applied drive voltage—and thus the grating depth—at each pixel controls the proportion of light that is either reflected back directly to the source or diffracted.
A Schlieren optical system is used to discriminate between reflected and diffracted light. By blocking reflected light and collecting diffracted light, very high contrast ratios can be achieved. We have measured the contrast of our GLV device at up to 1,000:1 (the sensitivity of our instruments). Thus the GLV pixel can be said to be in an ‘ON’ state when diffraction occurs and ‘OFF’ when it is reflected out of the system. For analog grayscale operation, the 1 µsec switching times shown is more than sufficient to create a 1,920 x 1,080 HDTV display at a 96 Hz refresh rate.
Digital operation capitalizes on the GLV technology’s tremendous switching speed to achieve shades of gray by alternately switching pixels fully “ON” and fully “OFF” faster than the human eye can perceive. Very accurate grayscale levels are obtained by controlling the proportion of time pixels are on and off. In analog mode, video drivers precisely control the amount of GLV ribbon deflection; pixels are fully “off” when not deflected, and fully “on” when deflected downward exactly one-quarter the wavelength of the incident light. Deflecting GLV ribbons between these two positions creates variable grayscale intensity.

Optical efficiency:
The optical efficiency of the GLV device depends on three main factors: 1) the diffraction efficiency, 2) the aperture ratio (the ratio of ribbon width to ribbon gap) and 3) the reflectivity of the top layer material chosen. In an ideal square-well diffraction grating, 81% of the diffracted light energy is directed into the +/- 1st orders. Aluminum alloys typically used in semiconductor processes allow cost-effective manufacture and are greater than 90% reflective over most of the wavelengths used for optical communications and imaging applications. Device efficiency, then, is the product of diffraction efficiency (81%); fill factor efficiency (typically >95%), and aluminum reflectivity (typically >91%). Overall, the device efficiency is about 70%, corresponding to an insertion loss of about 1.5dB.

Optical precision:
When a voltage is applied to alternate ribbons, the GLV device is set to a diffraction state. The source light is then diffracted at set angles. These diffraction angles are fixed with photolithographic accuracy when the GLV device is manufactured. Therefore, very precise light placement is achieved without the need for complex control electronics. This feature of the GLV device allows for significantly smaller and less expensive packaging and lower power requirements for optical components and subsystems.

GLV Driver Chips:
The custom GLV driver chips are very similar in function to standard LCD column driver chips – they receive and present data to the modulator at the line rate. The GLV drivers are designed for line times as short as 4 µs (corresponding to a pixel rate of 250 kHz per drive channel), which is adequate to support a 1,920 x 1,080 HDTV display at a 96 Hz refresh rate. Each driver output is programmable to 256 levels. The shape of the driver response curve is programmable, such that the effective grayscale resolution of the drive circuitry very closely matches the inherent electro-optic response of the GLV device, thus preserving effective grayscale resolution and eliminating banding or contouring at low light levels. A module operating all 1,080 channels at 8 bits at a line rate of 250 kHz is capable of processing video data at well over 2 Gbits/sec!

Laser and lens system:
A specific example of illumination optics for a high power laser bar is as shown below. The red laser bar illustrated consists of 24 emitters (each 1 µm high by 40 µm in length) spaced along their long axis at a pitch of ~400µm. A single cylindrical lens is used along the length of the bar for the fast axis collimation, while a perpendicularly oriented cylindrical lens array achieves collimation along the width of the bar. In this system, each of the 24 emitters is imaged to completely illuminate the entire array. Such an illumination design gives good uniformity (essentially the average of all 24 emitters) and also offers protection against potential failure of any given emitter (one emitter failure would result in about a 4% power loss, distributed evenly across all pixels.) Even with this relatively complex optical source, an illumination efficiency of >70% is achievable.
            Although a mechanical scanning component is not common to other high-resolution displays, the scanner requirements of the Scanned Linear GLV Architecture do not pose a significant system challenge, as the system needs only scan at the refresh rate, not at the line rate.

Projection Systems:
GLV elements can be operated in either a digital mode (with alternate ribbons either not deflected or deflected to precisely λ/4) or a continuously variable analog mode (with alternate ribbons deflecting to positions between zero and λ/4). Results with actual projection display systems yield unparalleled on-screen performance, having uniformity greater than 99% corner-to-corner, high contrast, 10-bits of grayscale per color, and no visible pixel boundaries. A linear GLV array can be used to modulate a single column of image data, while a mechanical scan mirror is used to sweep that column across the field of view
Single chip refractive method:
One way of reproducing color images is by using different ribbon pitch to create a red-green-blue pixel "triad" instead of the monochrome pixel described earlier (see figure  below). In such a system, white light is introduced at an angle slightly out-off--axis of the GLV device. In essence, the red area, having the widest pitch, refracts red light normal to the GLV plane while green and blue light is refracted at other angles.
         The green and blue areas, having narrower pitch, do the same for green and blue light, respectively. Color is produced by reducing the slit width to allow only a limited bandwidth about each of the primary colors to be selected.

Single chip method:
In a frame-sequential projection system (figure below) a white light source is filtered sequentially (by a spinning red-green-blue filter disk, for instance). By synchronizing the image data stream’s red, green and blue pixel data with the appropriate filtered source light, combinations of red, green and blue diffracted light is directed to the projector lens. In this system, as shown, a turning mirror is used both to direct light onto the GLV device, and as an optical stop blocking reflected light.

Single chip RBG method:
An even simpler, handheld, color display device uses three LED sources (red, green and blue). A single GLV device diffracts the appropriate incident primary -colour light to reproduce the color pixel information sent to the controller board.

Three-Chip projection method:
A more elaborate and accurate color projection system can be build using three GLV devices. By passing the source’s white light through dichroic filters, red, blue and green light are incident on three separate GLV devices. Diffracted light is collected and directed through the optical system to a viewing screen. This represents a much smaller and lower-cost solution, say, to the three-tube projection systems now used for large screen projection of PC images and videos. 


The pixel was operated at 2 MHz – accelerated approximately 8 times over its normal 250 kHz switching rate – and 20o C, for approximately 20 days. The GLV pixel product design life of between 1013 and 1014 switching cycles. For comparison, operating at a 100 Hz frame rate with 1,920 lines for 10,000 hours requires approximately 7 x 1012 cycles.
The ribbon natural frequencies decreased by ~ 2.5% as the temperature changed from 18 to 100o C because the ribbon's positive temperature coefficient resulted in less ribbon tension.
A higher incident power, orders of 30W, causes the GLV ribbon to heat and linearly expand, thus reducing its tension and its natural frequency. The same heating causes the device fill gas to become more viscous, thus increasing the damping time. But again, after an initial burn-in cycle and a ~0.5% change, the resonant frequency and damping factor are stable over time at both low and high operating powers.

Video Processing:

 For GLV projectors, the system receives 1080p video data at 24 or 30 fps via a standard SMPTE 292M serial digital interface. The electronics architecture supports the following system performance:
• 1920 x 1080 resolution
• Up to 120 Hz refresh, progressive scan
• 10 bits/channel R, G, B

The SMPTE 292M serial digital input contains luma (lightness) for all pixels and chroma (red and blue color difference) for odd pixels. The even pixel chroma values are generated by FIR filtering the red and blue chroma input. The luma and chroma are decoded into red, green, and blue with gamma using multipliers and adders. The use of 16-bit table entries results in maintaining a human-perceived signal quality while RGB is expressed in linear intensity.
This step maps RGB intensity to the GLV intensity voltage characteristic. Conventional spatial light modulators that create grayscale values through digital pulse width modulation have an inherently linear optical response. However, the inherent GLV electro optic response creates a natural, continuous grayscale with wide dynamic range that is well matched to the human visual system (Figure 5). Due to this mechanical simplicity, the GLV response is highly predictable and can be mathematically calculated from relatively simple models. If only a few data points near the peak intensity and maximum slope of the I/V response curve are collected, the rest of the curve can be calculated with a high degree of accuracy. Since the linear GLV array uses only a small number of physical pixels, each pixel can be exercised and the data necessary to fully calibrate the complete image can be collected using a simple optical integrator and single point detector. This simplicity enables a calibration technique that can efficiently measure all sources of variation within a system (particularly non-uniformities introduced by the system optics) and adjust the response of each pixel to show the highest quality image at all times.
The SMPTE 292M input is row-centric, meaning the video data is presented sequentially by row. Since the scanned linear GLV system as currently implemented scans left to right by column, a frame buffer is used to store data by rows and transpose it into column data for display. Since higher refresh rates produce better image quality, the frame buffer accepts progressive data at the source rate and sends it out at a faster rate for display. The frame buffer in the current system typically reads data in at 24 or 30 fps and refreshes the display up to four times the input rate.
By refreshing the display 3 or 4 times per frame, we can achieve 1.6 or 2 additional effective bits of grayscale through dithering. Through temporal dithering, the system exploits the GLV device’s inherent speed and the novel scanned line approach to achieve 10-bit grayscale using simpler and lower cost 8-bit drivers. For example, suppose the display needs to show the 10-bit grayscale value of 201.75. Using 8-bit drivers and temporal dithering, the system would display the refresh sequence below. Because we dither only the least significant bit(s), no flicker is perceived.

Benefits of horizontal scan:
First, it requires a smaller and less expensive linear GLV array (1080 pixels vs. 1920 pixels, a 44% pixel count reduction). Second, this smaller modulator allows additional system cost savings, such as smaller recombination and projection optics, smaller look-up tables, etc. Lastly, a horizontal scan also enables electronic support for variable aspect ratios (Figure 5). For example, a horizontal scan system can easily change from 4:3 to 16:9 for HDTV or from flat (1.85) to cinemascope (2.35) for electronic cinema, without requiring anamorphic lenses or complex scaling algorithms that tend to degrade image quality.

Comparison of DMD and GLV technology

Even if GLV technology involved column scanning to produce a complete picture, its architecture over rules the possibility of DMD being better. GLV had significant aadvantages over DMD as given below.
1. Significantly faster operating speeds.
At the order of 2µsec which is much higher than of DMD.
2. High optical efficiency (low insertion loss)
As GLV chip has high fill factor (of 95%) and continuous nature of the pixels.
3. Continuously variable attenuation that is highly accurate and repeatable
GLV pixels can be varied dynamically, compared with DMD only digitally.
4. Optical angular repeatability that is permanently set with photolithographic precision
Any slight change in the DMD structure can result the light reflected at some other angle.
5. No contact surfaces — high reliability and stability.
The DMD had to place springs and anti-stick layers (Teflon) so that the mirrors didn’t stick to either sides of operation.
6. Scalability to very large numbers of separately addressed channels
7. Ease of manufacturing
The number of steps for manufacturing GLV chips is much lower than that for DMD.
8. Ease of integration with CMOS logic.
The GLV chip, having a linear structure, has its CMOS logic on either side. But for DMD they are under the mirror and have to be fabricated before the mirror level is.


With these technologies projection TVs have become much more meaningful. GLV and DMD Projection TVs have shown much higher quality pictures and videos than any other. In the era where everything is getting digitised, they will surely replace CRT technology. 
For those who wished to bring essence of theatres into their homes have now a dream come true.

1 comment:

leave your opinion