The image above was cropped for the 4:3 aspect ratio of the CF-LCOS and DLP (cropping off some of the left and right side of the picture) and then the common parts of the image were projected to the same size. The DLP projector that was used further cropped the image to 290 of 320 pixel rows on the 480x320 DLP microdisplay, thus cropping off the bottom of the picture.
The next image shows close up pictures of the same area of actual projected images from each technology so that individual pixels can be seen. Even though the DLP image is in focus as the individual pixel boundaries can be seen, the image appears blurry due to the lack of resolution. The CF-LCOS image shows a severe lack of color and while the 640x480 image has the same number of pixels as the 640x480 portion of Syndiant’s image, the effective resolution is clearly lower plus the image has a “dotty” appearance from the color filters.
Pico projectors are going to be used for sharing photographs and videos and the comparison above demonstrates Syndiant’s advantage at projecting large images.
Major uses for pico projectors will include viewing and sharing photographs and videos, browsing the Internet, looking at email attachments, and giving presentations to small groups. The following comparison shows a close up of the same text and graphics image on the three technologies. The resolution and color quality advantages of VueG8 are clearly evident.
Now we are going to take a brief look at the devices generating the images above. Below are the three devices that generated the images above. They are shown to scale with a microscope photograph of the individual pixels for each device. It should be noted that Syndiant’s device with 854x600 pixels has about 3.3 times the number pixels of the DLP device with only 480x320 pixels. The packaged Syndiant device is about 20% shorter than that of the DLP as well. The CF-LCOS device uses 3 color filtered sub-pixels to form one color pixel and each of these sub-pixels is bigger than Syndiant’s pixel. The result is a much larger device even though it has fewer pixels. One might also notice that the CF-LCOS device looks darker than Syndiant’s or DLP’s. This is because the color filters are absorbing over two-thirds (2/3rds) of the light.
Looking at the microscope picture above one can see the individual pixels. Both Syndiant’s VueG8 and DLP use field sequential color, a process where different colored light sources (LEDs or lasers) are turned on in rapid sequence to form a color image. There are no colors on the mirrors.
Compared to the DLP, Syndiant’s pixels are smaller than DLP’s (about 2x the pixels per unit area). What cannot be seen is that Syndiant’s pixels are much easier and less expensive to make as the Syndiant pixels are simply the top level of metal of a high volume planar CMOS process with no moving parts. The DLP pixels on the other hand are MEMs devices that physically tilt and require a complex “superstructure” with a number of complex process steps on top of a CMOS wafer which makes them more expensive to fabricate.
Looking at the CF-LCOS device above, it is obvious that the device is much larger due to the requirement to have 3 sub-pixels for each pixel. This device is too large to embed in a typical cell phone. The size of the device is also a cost problem as they get about 1/4 the number of devices per silicon wafer. The color filter process contributes to the processing complexity and cost. Adding it all up, the cost of manufacturing at maturity for the color filter device should be more than 4 times as much as Syndiant's VueG8 device.
With the CF-LCOS, the electrical fields controlling the sub-pixels spread in such a way that when controlling one color, it affects the intensity of the surrounding pixels which are different colors. This interaction of the colors causes the poor color saturation and color control evident in the CF-LCOS images shown earlier.
A manufacturing advantage not evident above is that Syndiant’s device is the only one that is fully electrically testable. Syndiant uses a digital SRAM bit under the mirror that can be read back under test. While the DLP can electrically test the SRAM bit that controls the mirror, the mirror tilting superstructure, the most complex part of making a DLP device, can only be tested by optical inspection. The CF-LCOS device has analog storage capacitors under each pixel that cannot be read back and thus can only be optically tested.
Another big issue for pico projector microdisplays is the packaging cost. The CF-LCOS and Syndiant devices have similar packaging, but the Syndiant device has cost advantages in being smaller and thus making many more devices per wafer of silicon. In the case of DLP, Syndiant has significant packaging cost advantages some of which can be seen side by side comparison below.
Note that even though the Syndiant device has 3.3 times the number of pixels because of the VueG8’s on-display processing, it requires less than half the number of bond pads (39 versus 96). The benefit of many fewer bond pads is fewer signals to route and less complexity in the packaging. The DLP device itself sits on an expensive ceramic package where Syndiant’s device sits on a very low cost 2 layer circuit board. The DLP has a chamber formed over the tilting mirrors that has a piece of glass with a hole in it (doughnut spacer) and a top piece of glass that must be affixed to each individual device. In the case of Syndiant LCOS a single wafer of glass is bonded to a wafer of silicon in a wafer level process to simultaneously form hundreds of devices.
Future Roadmaps: Syndiant, DLP, CF-LCOS going to higher resolution
Syndiant can already fit more than twice the pixels per unit area compared to DLP and CF-LCOS. DLP appears to be approaching optical and physical limits as to how small they can make a mechanically tilting pixel. With the CF-LCOS, the need for at least 3 subpixels and the problems with color bleeding between sub-pixels severely limits its scalability to smaller pixel size and higher resolution and they are already too big for most cell phones.
Syndiant’s technology can be readily scaled to smaller pixel sizes and we expect to have HD 720P and beyond resolutions small enough to embed in cell phones in the not too distant future.
A Brief Discussion of Ferroelectric LCOS (FLCOS) and Laser Beam Steering (LBS)
The discussion above has focused on DLP and CF-LCOS because they are the two most common competing technologies that are in the market. Ferroelectric LCOS (FLCOS) and Laser Beam steering are two other technologies that have been vying for the pico projector market but appear to be much less viable.
Ferroelectric LCOS (FLCOS) also uses field sequential color similar to Syndiant’s VueG8. FLCOS has a major drawback caused by the requirement to “DC-Balance” the LC, the light has to be turned off more than 50% of the time while the ferroelectric LC material is “balanced.” FLCOS has been used in many near eye applications where the 50% light-on time is not a major issue, but in the pico projector market, having such a short light-on time means lower brightness and/or needing more expensive LEDs and/or lasers. Another drawback to the FLCOS devices to date has been their large pixels which are about 4 times the area of Syndiant’s pixels.
Laser Beam Steering (LBS) uses one or two micromechanical mirrors to steer red, green, and blue laser beams in a raster scanning motion. On the surface it seems simple enough, but the reality is quite different. While small optical modules are shown, smaller optical modules can be made for Syndiant’s VueG8 device (3-6 cm3) while the electronics to drive the laser and mirror appear to be much larger and more complex than required for VueG8.
A compelling marketing point with LBS is that of “focus free” operation, but it should be noted that Syndiant’s VueG8 is also focus free when using laser illumination due to the optical properties of laser light. Syndiant’s design partners and customers have a choice of using either LEDs or lasers based on their cost and performance trade-offs.
A major cost problem with LBS systems is that they require very high speed (on the order of 100 megahertz) switching lasers because the light must turn on, off, or to an analog level in between in the time the beams sweep past one pixel. This type of laser has proven both difficult and expensive to produce, particularly for green. Syndiant can use low speed “continuous wave” (or CW) lasers which to date have proven easier to produce, less expensive and are more electrically efficient.
An image quality issue with lasers is that of “speckle” or a dotty/noisy appearance cause by the coherence of laser light. While there are known ways to reduce speckle for lasers that are spread out to illuminate microdisplays such as Syndiant’s VueG8, these techniques can’t be used on a tightly focused laser beams as required by LBS and so LBS systems to date have significant amounts of speckle in the images.
LBS systems claim a power savings in turning the lasers off when scanning across black/dark pixels, but offsetting this power savings is the power required by the high speed analog power drivers of the lasers and the high speed switching lasers themselves may be less energy efficient. In a ViewG8 system, the LEDs or lasers switch fully on or off about 100,000 times slower which takes less power than high speed analog switching.
LBS systems require complex and space consuming electronics to drive the mirror and the high speed switching laser. The mirror sweeps the laser light non-linearly both in speed and direction. A change in speed without correction of the three (RGB) lasers’ intensities would result in pixels that are either wider and dimmer or narrower and brighter at various parts of the image. Additionally, the scaling process to correct for the non-linearity of the beam sweeping hurts resolution.
Another major drawback to LBS is that due to the tightly focused beam, LBS is limited to about 15 lumens of total light output due to ANSI/FDA in the US and IEC international safety standards. With lasers illuminating microdisplays such as Syndiant’s VueG8, the light exiting the projector is spread out to illuminate the entire pixel array and thus can be made much brighter while still meeting safety requirements.
Laser beam steering also has a host of potential manufacturing issues that start with the custom MEMs mirror manufacturing process, whereas Syndiant is using high volume standard CMOS. It would seem that each MEMs device might need individual calibration as small differences in manufacturing would cause the beam to sweep differently. The laser positioning and aiming are critical in LBS systems to emit reasonably coincident RGB beams.
While some companies have heavily promoted LBS, the reality in the market place so far has been that it is far too expensive for the mass market and the image quality appears to be much lower than that which can be achieved with Syndiant’s VueG8 technology using either LEDs or lasers.