[noah katz]

Thumper,


As you may have read, my hopes for acceptable blacks from the new DLP's were dashed after seeing the Sharp 9000.


In a previous thread you described how you improved contrast on an LCD by adding polarizer plates: "Together they cut the brightness 40% but did yield better contrast."


How much better? Any guess what contrast and lumens you could end up with if you added polarizers to, say, the Sanyo 21/30/45? I'd be willing to give up more than half the light for 1500 or 2000:1 CR.


Does this have anything to do with the LCD's mechanism of stopping light transmission?


Back to DLP:


1) I was in the process of sketching up some light trap designs we've used in our instruments (I do optomechanical design) when I saw in a post yesterday that TI uses a cone for this purpose, which is a good approach. Light entering reflects back and forth between opposite walls, so if you have 10% reflectivity and 6 bounces before the ray exits, its intensity is reduced by a factor of one million. Depending on its design and implememntation, this might not be worth trying to improve. What's your opinion on this? If you think it could be improved, I'll send you what I have.


2) Someone else mentioned light scattering by the dimples in the center of the mirrors. This sounds like it could be the biggest problem and may be why a seemingly paltry increase from 10 deg tilt to 12 deg may be worthwhile, by steering more of the scattered light away from the lens. What do you think?


3) Then there's the basic quality of the mirrors themselves. If they're not very flat and smooth, they will scatter a percentage of any incident light in all directions.


Here's a thought on reconfiguring the light engine: Tilt the DMD to increase the reflectivity, which is what happens as the incident and reflected light rays move away from normal (even a piece of wood looks mirror smooth at a glancing angle).


Taking this further, if you tilt the DMD by 41.4 deg about the horizontal image axis, when you look through the lens, the DMD now has an AR of 16:9 (cos 41.4 = .75). This requires a radical mechanical redesign, because to keep the light exiting straight out through the lens, the incident light has to come almost directly from the side (81.8 deg from the exiting light).


I don't know if this would worsen or improve scattering from the dimples.


Thumper, do you know if the TI site has any actual pictures or mechanical dwgs of the light engine?


Thanks


------------------

Noah

 

[noah katz]

Is there more on the site than besides that cartoonish animation?


I hadn't thought abouy it before, but a basic limit on contrast ratio will be set by light diffracted/scattered off of the mirrors' edges. The only way to improve it would be to increase each mirror's size to inccrease the ratio of mirror surface to edge length.


Thanks


------------------

Noah

 

[indygreg]

ti's site has lots of pics. www.dlp.com.


the cone i referred to in my other post was the cone of light eminating from the dlp. i think you are talking about something else.


ti made a redesign of the chip a year or so back that changed the dimple and reduced the stray light dramatically. not sure what else can be done. light also reflects off the edges of the mirrors in the off position as i understand.


greg


------------------

-------------------------

Compaq MP2800 DLP, 1Ghz htpc w radeon aiw, 45 x 80" diy accoustically transparent screen, lexicon dc-1. ipaq touchscreen for control

 

[cutman0122]

I have a Davis 450 clone DLP PJ, and I've found along with darkening the DMD chamber (based on Thumper's mod), placing a hood over the optical lens has further improved black level. This hood basically reduces any stray light bouncing around the chamber from reaching the lens by blocking it. I realized that the lens in my PJ is just sitting there like a mouth wide open, soaking up any light it can. I painted the hood black inside and out to further reduce intensity of stray light that may still be reaching the lens. It seems to have further improved black level, I can now enjoy watching in complete darkness! One other area I think may help in black level, in my PJ anyway there is a clear protective cover lens in front of the DMD. I guess it's there to isolate the DMD from dust and what have you, but I think it's also reflecting a lot of the white (unprocessed) light to reach the lens. Maybe if it's angle relative to the lens is optimized (reflect away), we might see further improvements!


Now if I can just figure out the rest of Thumper's mods..

Thumper, can you please comment on the idea of optimizing the angle of the clear DMD lens cover, and whether you do something similar with placing a hood over the optical lens? Just wondering, with your experience I'm sure you can do a better job on these tweaks than I ever can! Thanks!

BTW, adding a hood over the lens will limit zoom based on the hood's dimensions.


Gerald


[This message has been edited by cutman0122 (edited 09-13-2001).]

 

[Thumper]

Ok Guys,


I'll bite (a little) on this one.


Noah,


It is difficult to say what improvement (if any) could be done with the 18/21/30/45 high contrast Sanyos (noticed I left out the PLV-60 in this group...). Briefly, this is due inpart by the *improved* polarizing plates used in those models and the particular Epson LCD panels employed. The tests I did last year were on much lower contrast units (Sony 400Q and a Mitsubishi LV-2000). In those units, the stock polarizers were less effective and allowed more non-"perfectly" polarized light thru (roughly 4% of off axis light made it thru). It was possible to get a 40-50% increase in overall contrast but at the expense of 40%(+-) of the brightness with the use of additional (in series) plates.


Tilting the DMD (particularly in relationship to the output lens assembly) simply would not work as the precise focus over the entire DMD active surface is critical. This would place parts of the surface out of focus unless some wild, unimaginable, non-spherical lens was employed. It would be much easier to install two miniture Cygnus-pannymorph style prisims between the DMD and the output lens http://www.avsforum.com/ubb/smile.gif


Cutman,


I do not place a "hood" over the lens assembly. Losses from small amounts of oblique angle light entering the lens do not raise the black level significantly (most lenses have flat black internal housings). Without giving out all my little secrets, I can say that I do use a "hood" of sorts, but it is on the DMD. It is collectively known as a "halo mask". These "masks" are standard by most manufacturers but they are usually woefully inadequate. A poor mask does raise the black level by a fair amount as is is more nearly perpendicular to the output lens and in the focal plane. In most projectors I modify, the outright addition (or major modification) of a mask is needed. To get an idea of what one looks like, take a hollow pyramid that has 3 x 4 side dimensions, lop off the top of the pyramid to make a "cone", place the opening where is is lopped off against the edge of the DMD and the open base of the pyramid is where the light is both supplied to the DMD and reflects out to the lens (or light trap). It does get more complicated as the approach angle of the supply beam and exit angles have to be considered as the tip of the mask ideally should be only a few pixels width away from the active edge of the DMD array. Add to the formula the precise angle needed of the pyramid's slope to achieve both supply/exit angles without shadowing (shading) and to achieve a combination of maxium reflection and absorbtion of the light outside of the active area (just remember: reflective panel technology is a great game of geometry). It does get better with the possiblities of the finish on the mask to achieve the right combination of reflection/absorbtion properties.


Properly implemented, a halo mask (or "DMD hood") does not affect zoom or focus and to hold halo leakage to a minium is installed as tight as possible to the active surface without causing any shadowing.


As to discarded light "traps", they are important. Depending on the positional relationship with the lens, it can be a simple flat black surface or in some cases multiple inward facing facets (usually flat black) that absorb and reabsorb or they may be corragated surfaces or heavily textured surfaces...


There are some great white papers by T.I. and others on the subject of various optical engine designs.


Ok, I'm going to jump back in the water now. What the heck, its catch and release anyway http://www.avsforum.com/ubb/smile.gif


Thump

 

[noah katz]

Thumper,


Thanks for the response.


So even with the full treatment, it sounds like LCD's still wouldn't cut it.


Oops, wasn't thinking about tilting the DMD's affect on focus.


It sounds like you're well-versed in light trap design. Do you think the stock light traps leave room for improvement, and do your mods address this aspect?


Regarding the DMD shrouds, since many current projectors use the same XGA DMD, do you think it likely that you could use the same one you're developing for the LT150 on other projectors, or is there a lot of variation in the landscape immediately surrounding the DMD?


Thanks


------------------

Noah