I have been thinking of ways the increase projectors contrast ratio way over what we have right now. My reflexion brought me to using LCD chip on the light path of a LCD (4chips) or DLP (hybrid) projector. But I think now that it wont be that effective on a DLP because light is not polarized and doing so could reduce lumens too much. A 4th chip in a LCD projector doesn't seem that beneficial because no one have done it yet.

But what about 2 DMDs? I was looking at this diagram from Sharp showing the optical unit of the XVZ-20000 and notice that the light is reflected on a mirror not long before hitting the DMD chip. Could a second DMD chip be introduced in the design and replace that mirror? We would get contrast on : off in the millions with a desgin like that wouldn't we?

Why is this not working? To expensive? To complicated to implement. Need two DMDs of different sizes?

Thanks!

4-panel lcos is sold by SEOS. They quote from 250000:1 upto 900000:1
NHK and JVC has a 4-panel lcos research projector, 1000000:1

Brightside technologies now bought by Dolby had reference designs for many kimbinations of lcd, lcos and dlp but no 2DLP solution.
1 2x3LCD
2 3LCD +1LCD
3 3LCD + 1LCOS
4 3LCOS +1LCOS
5 IDLP +1LCD or LCOS

Perhaps it is difficult to sync a 2DLP system?

"Why is this not working?"

I might not have thought it through sufficiently, but I think the problem is that there needs to be one-to-one correspondence between the pixels in each DMD.

This means precise physical alignment between them and more space and optics to give two focal planes.

"Why is this not working?"

I might not have thought it through sufficiently, but I think the problem is that there needs to be one-to-one correspondence between the pixels in each DMD.

This means precise physical alignment between them and more space and optics to give two focal planes.

I think you may have a valid point but if you look at how the brightlight TV work you would find out that the back panel (led) actually has less resolution than the front LCD panel. In this case the first DMD chip could act as the back panel et the onther one as the front panel.

I will think it will be easy to make both of them work as the need the exact same signal compare to a 4 LCD design were the 4th LCD chip need to have a composite signal of the 3 main LCD.

well i havent a clue, but some 17 year old kid cracked the new i-phone by Apple. Why not just type up your theory and send it to Texas Instuments.

Joey

Why is this not working?

Who says it's not working? The Vidikron Helios had 2 chips many years ago. From what I know, it was the only 2-chip DLP projector ever made.

I had a demo of it, I think that was back in 98 or 99. Besides of the much higher brightness in comparison to 1-chip projectors, there was no advantage at all. Back then, CRT was still the ultimate choice.

Isn't the main problem is that DLP uses Pulse Width Modulation. Adding another PWM panel doesn't really get you much more than an alignment headache, more complicated optics and no real additional control of the light modulation. You could possibly do a green panel and a red/blue colour wheel panel. That might give you something.

"The Vidikron Helios had 2 chips many years ago."

The DMD's were in parallel, like 3-chips, not in series as is being discussed here.

Stephen the Vidikron was not using the same principal. Both chips were used in parallel instead of series.
The HELIOS uses two DMDs, one for red and one for green and blue, totaling 1,017,600 micromirrors.

Roger that is where we don't agree. I don't beleive that the first DMD need to be perfectly align with the second one. Actually it need to be align yes but not perfectly focus.
If you read this review or just go to how it work you will understand what I talking about.
http://www.bit-tech.net/hardware/2005/10/04/brightside_hdr_edr/1


Here is a picture from the review on bit-tech that resume the principal quite well.
The first DMD would do the equivalent of the job done by the led in the brightside display. Noctice that they don't cover perfectly each pixel. In fact there is only 1400 LED for more than 2 millions pixels in this case.

http://www.bit-tech.net/content_images/2005/10/brightside_hdr_edr/imled_array.jpg

The second DLP chip would work with less light from specific location to produce the image.

http://www.bit-tech.net/content_images/2005/10/brightside_hdr_edr/imled_fear.jpg

The result would be an image with better contrast and dynamique range.

http://www.bit-tech.net/content_images/2005/10/brightside_hdr_edr/matrix0.jpg

Remember that the brightside display used 1400 LED to enanced a 1920x1080 display. having 1920x1080 led would be even better I think.

In this design both chips are drive by the same signal. Both are sycn with the colorwheel perfectly.

Most of you are probably too young to remember Lirpa Labs, but they were rumored to be working on a two-color primary system. I heard rumors that they eventually built a 2-chip DLP version but it was never announced and the company apparently has not survived. The last story I saw that even mentioned their two-color primary system was from CyberTheater in 1996.


Lirpa Labs - A Giant
Disappears from Audio History

CyberTheater was saddened to learn last week about the apparent demise of one of the pioneering companies in high-end audio. Lirpa Labs, which annually announces its latest breakthrough technology in the April issue of Audio magazine, apparently, is no more.

Lirpa has always operated on a shoestring advertising budget and has chosen to promote its products only once a year in Audio. Although we have no direct confirmation that Lirpa has closed its doors, we find the omision of a new product announcement in the annual issue, to be a clear indicator that the company has fallen on hard times. We tried to contact Lirpa directly by phone but operators were unable to find any current listing for the once proud and innovative company. None of our other industry contacts were able to shed any light on where the Lirpa headquarters might now be found.

Although Lirpa was always a technical innovator far beside its time, it apparently wasn't able to convert its research to video quickly enough to keep up with the stunning market revolution to Home Theater. There were rumors in the industry that Lirpa was working on a breakthrough video monitor that replaced the conventional red, blue and green CRT primary colors with only mauve and burnt orange. It now seems that time ran out for this new two color system before it could be brought to market. CyberTheater can only mourn the many potential video innovations that will never see the dark of night, if Lirpa Labs fails to re-surface.

It has always been rumored that Audio magazine made annual investments to ensure Lirpa's corporate survival. Sales for Lirpa's leading edge products were known to be minimal due to constant manufacturing difficulties. We can only hope that Audio will step up once again, and save this giant of diminutive audiophile companies.

Copyright © 1996 CyberTheater

:)

I've thought about the same thing, bgosselin, and I agree, the difference between a two chip / two color DLP system and a "contrast multiplying" 2 DLP system (1 color, 1 luminance only) is large.
There would be no contrast benefits in the former, and no color benefits in the latter. I think it's a fairly obvious solution and part of the reason it probably hasn't been implemented is price / complexity. Each DLP would need it's own "light sink" or whatever they call it (been a while since I read the white paper) and would probably be a fairly complex optical path - less than 3 DLP but more than 1 DLP. You still need the color wheel too. Another thing is I bet there would be quite a hit in brightness in such a setup compared to a single chip DLP system, unless an iris is no longer needed because of the contrast boost.

I don't think an Iris would be needed. After all the two DMD should boost contrast by a huge factor. Lower brightness is a possibility.

Any body how less reflective a DMD chip is compare to a regular mirror?

I don't know if it's that complex. That is what I will like to get confirmation on by an expert on that field. If you look at my first diagram coming from Sharp there is already a mirror on the light path. Just switching by a DMD chip should do the trick. I do think that the mirror is probably bigger that the TI chip. So that is probably the problem. You would need then to use more optic to refocus the light.

But outside of it it should be simple. Both DMD chip would need the exact same signal to work, so no need to build a complex circuitry.

The color wheel would stay but I don't think it's that bad. I will think that the majority of the population don't have problem with rainbow!

I would like to get Dan Miller to comment on that idea. Greg Roger must have an opinion as well. He as so much knowledge on projector.

There were rumors in the industry that Lirpa was working on a breakthrough video monitor that replaced the conventional red, blue and green CRT primary colors with only mauve and burnt orange.

Y'know, that was always the problem with Lirpa Labs products. Most of the specs would be great but then one or two would be deal breakers. See, I could have gone with a chartreuse and tangerine two-color system, but mauve and burnt orange? No way.

Hi,

2 DMD systems have been implemented in the past. The main reason was to compensate for low UHP lamp spectral content in the red wavelength. I believe one color wheel was used with green and blue segments (probaly unequal) for one DMD and the other DMD was always illuminated with red light and always got written with red channel data. It worked (more brightness, better white points, and more time to write LSB data), but of course, DMDs are so expensive that these products died for cost reasons. UHP lamps have improved a little (though still deficient in red).

Shame they didn't last beyond '96 Greg
Maybe you should start the company up again - ready for the spring :)
I can think of many new products the AV world is crying out for.
Wall paint which doesn't reflect projected light.
Using your screen as a giant NXT speaker.
Video processor which can create new frames in-between existing ones ;)

Shame they didn't last beyond '96 Greg
Maybe you should start the company up again - ready for the spring :)
I can think of many new products the AV world is crying out for.
Wall paint which doesn't reflect projected light.Something better than flat black?
Using your screen as a giant NXT speaker.Like the ASS-60AK? (Gotta love that model number.)
Video processor which can create new frames in-between existing ones ;)Such technology has been available for a while now. Were you trying to be funny?

Were you trying to be funny?
:(

:(Sorry, I must have got out of bed on the wrong side this morning!

I will like to experiment on this. Anyone has an old Viewsonic PJL830 for me? I open it yesterday and think I could do some testing of this theorie. But need two of them. I have one just need another one.

Thanks!

If you check the archives you will find that years ago when dynamical irises came up for the first time in production units, I mentioned that you could use a DMD chip as an ultra-fast lamp-iris, that would have some serious advantages and disadvantages but has never been done so far, at least not that I know of...

There was a quite successfull two-chip DLP projector... the Vidikron Helios see here:

http://www.projectorcentral.com/Vidikron-Helios.htm

some people still use these today, I remember that when I sold my MetaVision one of the guys interested in it bought a used Helios instead because it was his Dream-projector for years...

This is not what I want to test. I want 2 chips in series. The Vidikron is using on chip for one color and the other for the other two. Both chip in the design I have been talking about will handle the primaries and the secondaries.

It will probably won't work but if I can fin an old Viewsonic I could get this idea out of my mind. :)

Two Infocus sp4805 could work also but I don't want to spend much.

What I like about the Viewsonic is that I can play with the gamma curve easily.

If I have two sp4805 It could work also because I have a Lumagen scaler that I could use to set gamma.

With the two chips in serie I would need to reduce gamma of each projector to achieve an output gamma closer to 2.2

I understood that the Vidikron is not what you want, the link was more in reference to the two-chip DLP design gregr mentioned above.

You would probably need additional optics to focus the first DMD on the second and it may turn out to be quite difficult to operate the first DMD sans the rest of the projector. Definitely a bigger project... certainly not to be solved over the weekend.

I understood that the Vidikron is not what you want, the link was more in reference to the two-chip DLP design gregr mentioned above.

You would probably need additional optics to focus the first DMD on the second and it may turn out to be quite difficult to operate the first DMD sans the rest of the projector. Definitely a bigger project... certainly not to be solved over the weekend.


It doesn't need to be perfectly focus. But it need the light beam from the first chip need to be close to the second chip when it hit.

I want to send both projector the same exact signal. I want to make sure both projector have the exact same firmware and colors calibration before starting. But even if there is a difference I should still be able to check the impact on static image. Movment is tricky because both chip need to be in perfect sync. That will probably be impossible to achieve.

Since nobody has explained this I thought that I (longtime lurker) would take a stab at it as my first post.

Why a Dual DMD sequential modulated projector is difficult to make work from a practical standpoint.

For the sake of argument, lets assume that no matter the technology, that the light modulators are linear (i.e. no gamma). Lets also assume that we are going to use dual modulators at 50% to end up with 25% output.

Example 1. Both modulators are either LCD or LCOS.

For the sake of simplicity, lets assume that the polarization and everything else in our system is set up properly.

Light passing through the first modulator is reduced from 100% of what that modulator could let through to 50%. This 50% passes through the second modulator and is reduced from 50% to 25%. In the end, the system as a whole passes 25% of the maximum amount of light that could pass through the system.

Needless to say, such a system would have losses in total light output but ignoring that, we still have a system that works (albeit a dimmer system). This system works because each light modulator reduces the _intensity_ of the light. The end result is a system where the end contrast is roughly multiplicative and there are enough intensity values to easily specify light output values along the entire range of the system.

Example 2. Both modulators are DMD.

For the sake of argument and simplicity, lets assume that the devices are equivalent resolution and lined up perfectly. Lets also look at the devices in a more "theoretical" capacity and assume that they turn on and off "instantly". Lastly, when we refer to a "DMD", we are referring to a single pixel of that DMD.

Like example (1), we set both DMDs to pass 50% light. Now here is the whole crux of the problem: since DMDs are PWM devices, they are only ever on or off. To pass 50% light, a DMD is on 1/2 the time and off 1/2 the time. For this example, lets make a simplifying assumption that the DMDs are clocked and that to produce 50%, they reflect light on the first clock and absorb light on the next and so on. In other words, we have a square wave with a 50% duty cycle. Lets also assume that both DMDs are the same "speed" and run off the same clock. This is a buttload of assumptions but I think it allows us to present the problem in a way that is easy to mentally visualize.

If we assume that the both of the DMDs are operated in-phase, then when one DMD is "on" the other will also be "on" and the same for "off". In other words, in this scenario the 2nd DMD is really only acting as a simple mirror for the first. This system as a whole passes 50%, NOT 25%.

If we assume that the signals driving the DMDs are operated 180 degress out of phase with each other then when one DMD is "on" the other will be "off" and vice-versa. In this scenario, the 2nd DMD essentially absorbs the light from the first. This system as a whole passes roughly 0%, NOT 25%.

Now in your mind, change the DMD modulation coding from simple PCM to grey coding or something else. Change the clocks and phases, and think of what you end up with...

The problem here is that a DMD does not modulate a light's intensity. It only switches light on or off very quickly and our eyes perceive that as an intensity. An LCD/LCOS changes a light's intensity, a DMD "flickers" light in time.

If you take an existing projector (like the Sharp diagrammed in the opening post) and slap on a second DMD utilizing a similar formatter chip, you end up with nothing more than an artifact engine.

For this scenario to work, you would need a way for a 2nd DMD to properly modulate the output of the 1st DMD. This would work (theoretically) if you took the the output of the first DMD and passed it though an optical low-pass filter, i.e. something that converts a stream of pulses to an intensity. While this is trivial to do for audio, for video...

Without the "low-cost" optical low-pass filter, what we have as a system can be characterized (at the pixel level if we are overly simplistic) as two binary-state waveforms passed through an AND gate. At this point it becomes a coding/formatter problem.

Now if you had two DMDs of the exact same resolution and you could line them up absolutely _perfectly_ and you operated them off the same formatter, you'd have a system that had higher contrast but no additional grey-level resolution. One device is NOT modulating the output of the second. BTW, good luck lining them up that precisely.

If you have a lower resolution DMD feeding a higher resolution DMD (as in the Brighthand example) you have nothing more than an expensive artifact engine. What you are doing in this instance is "AND" modulating dissimilar resolution planar bit fields of an image together (please recall that a DMD cannot display a gray-level, it can only display a binary image so we must convert a gray-level image into binary planar bit fields and multiplex them through time).

Of course, there is no reason why you can't combine a DMD and a LCD-or-LCOS device (if I recall correctly, the Brighthand folks first used a DLP projector shining on the back of a LCD monitor panel).

Thanks for your first contribution. I'm sure I can get a second one from you. ;)

My focus is for two DMDs. I don't think adding a chip to a LCD or LCOS designs will be worth it.

I agree with what you say but don't see that as being a problem for the manufacturer engineer. If they use the same signal for both DMD they won't be out of phase but always in sync. The big gain from the two DMD system will be gain from the second chip working with the first chip blacks.

When TI is claiming success for contrast ratio up by 30% I think a 2DMD system that can increase contrast between 200x to 1000x would be still a big achievement. Getting more grey level out of the system is not what I was getting after. But I'm sure if the can wobulate a dmd chip the can find a way to create a 2 DMD system with more grey level.

The real problem for me is to get the 2 colors wheel of the two independents system to work in sync. I don't know if it can be done. Can someone tells me if I take two identical projectors and feed them an RGB signal via a RGB powered splitter. Will the colors wheel be likely in synch or not at all? Any way to make sure both signal coming out are in phase? If they are a almost in synch I may be able to get some result by changing brightness on the first chip. Having the chips more often on would help your 180 degrees out of sync mirror example.

The second problem like you said is to align the chip. I don't believe the need to be focus. But the light beam coming from the first chip need to be the same size as the second chip.

I can't seem to find another Viewsonic anyway. So I won't give this a try. This project being likely to fail. I don't want to invest big money in it. :)

Bruno