[noah katz]

Joe House got me thinking in this thread

http://www.avsforum.com/avs-vb/showt...19#post2157719


I thought it was interesting enough to post separately; hopefully those more knowledgeable chip making will comment.



Consider for example a 3840x2160 chip. Group pixels into 2x2 arrays, giving 1920x1080 addressable pixels (this also saves on processing power).


Now when a pixel is bad, it's only 1/4 of a visible pixel. Yields could skyrocket - hundreds or even thousands of bad pixels would not be noticeable, as long as they weren't adjacent to eachother (1000=0.012% of 3840x2160=8,294,400).


A lot depends on what typical yields are. I understand they're pretty low for LCOS.


I'd think that in order for this to be economically attractive, two things would need to be true:


1) Chip sizes would have to have reached a minimum size determined by their ability to deal with the heat of a concentrated light source (otherwise lower res would be smaller and chipper) and/or the the light cannot be efficiently focused on any smaller area


2) They could make the high-res chip the same size as one with half the resolution and bigger pixels, so that it didn't cost much more for the higher res (it's my understanding that chip size is the biggest determinant in cost)

 

[rmostad]

Quote:

Originally posted by noah katz Consider for example a 3840x2160 chip. Group pixels into 2x2 arrays, giving 1920x1080 addressable pixels (this also saves on processing power). Now when a pixel is bad, it's only 1/4 of a visible pixel. Yields could skyrocket - hundreds or even thousands of bad pixels would not be noticeable, as long as they weren't adjacent to eachother (1000=0.012% of 3840x2160=8,294,400).

I think the real problem with this idea is that 'badness' isn't quite so cooperative. The failure could just as likely be one of the trace lines that goes bad or the failure could be localized to a multi-pixel group. That would mean you would loose a whole line of pixels or a small but noticeable group. It might help to have redundant driver circuits, as long as it doesn't reduce the fill rate but it might not reduce bad chips enough to be worth it.

 

[JHouse]

I've been HIJACKED!

 

[JHouse]

The word is that the Sony and Sanyo HD pj's will both use the Sony chip. So if anyone knows anyone at Sony, they probably already know the answer to this question. This really could be a huge motivating factor to go higher resolution. Cool.

 

[noah katz]

Roger,


"The failure could just as likely be one of the trace lines that goes bad or the failure could be localized to a multi-pixel group."


For the vast majority of people who have complained about bad pixels, it's been isolated pixels.

 

[rmostad]

Quote:

Originally posted by noah katz Roger, "The failure could just as likely be one of the trace lines that goes bad or the failure could be localized to a multi-pixel group." For the vast majority of people who have complained about bad pixels, it's been isolated pixels.

This is true but these are post production failures or panels/chips that make it out of the QC phase so they probably don't count in the traditional yield measurements. These are not the panels/chips that failed before even being cut from the wafer or larger glass substrate It is the failure modes of those parts that don't even make it to assembly/mounting that determine whether an increase in feature density would help increase yield.

 

[rogo]

Chip size is the most important thing. But number of features and complexity of processing circuits also matter.


I'm not sure it'd be cheaper to go down this route. I think it's cheaper to master making things like LCOS imagers. They could eventually have huge yields like regular silicon does -- in theory.


Mark

 

[Art Lloyd]

Ignorant question: Is it the cost of the chips that is the major factor in projector pricing? How much do the manufacturers pay for TI's chips or for LCOS panels? Art

 

[rmostad]

Quote:

Originally posted by Art Lloyd Ignorant question: Is it the cost of the chips that is the major factor in projector pricing? How much do the manufacturers pay for TI's chips or for LCOS panels? Art

Well that would be the 64 million dollar question. I don't think you will get any good information on that for a while. It might be possible to get concrete pricing information of LCD panels for projectors as there is probably a wider group of manufactures and LCD projection technology has moved down the ladder. Of course quantities purchased will also effect the price.


Mark:

Shrinking chip size is definitely the best way to reduce rejects. Of course there are some limits to how small the chips can get. I don't think you would have companies lining up for a 0.5 cm DLP chip, even if it was 1920X1080 ;-)

 

[tbhausen]

Wouldn't more pixels decrease the fill ratio due to more inter-pixel spaces per overall image size? Also, all these pixels would require some major processing power. I like Joe Kane's approach--720p beats 1080i, so let's just do away with the interlaced formats, perfect the 720p devices (e.g. HD2 dimple fix) and live happily ever after.


I'm really to the point of wondering how or why pq could or should get any better. Viewing HD on my SP7200 from any reasonable distance leaves pq to the mercy of the provider (HDNet live events-spectacular, DBS 480i can be horrible). Most of the talk of technical improvement should focus on OTA and recorded sources--the display devices are most often hald back by the source.


Todd/Indy

 

[Palladin]

Quote:

Originally posted by Art Lloyd Ignorant question: Is it the cost of the chips that is the major factor in projector pricing? How much do the manufacturers pay for TI's chips or for LCOS panels? Art

Hi Art,


I know Mr. Wigggles had posted the amount for just the licensing fee on the TI chip, which I believe was in the 7 figure range, but don't recall the thread. As to the chips themselves, based on the pricing of the DLP RPTVs (e.g.. Samsung), its a safe bet that the cost is signicantly less than what we may have believed when it was showing up only in FPs.


___________________________

Palladin


No matter where you go, there you are

 

[noah katz]

"Wouldn't more pixels decrease the fill ratio due to more inter-pixel spaces per overall image size?"


I think that as the pixels get smaller, so do all the other features.


"Also, all these pixels would require some major processing power."


In the example I gave, 4 adjacent pixels display the same information. The idea isn't to display more information, but to dispaly it with redundant pixels so that bad pixels go unnoticed.

 

[DaViD Boulet]

Quote:

I like Joe Kane's approach--720p beats 1080i, so let's just do away with the interlaced formats, perfect the 720p devices (e.g. HD2 dimple fix) and live happily ever after.

that's a very short-sighted approach on the part of Joe Kane. The real answer is not having to make a choice between 720P and 1080I...but to simply choose 1080 P.


Any digital projector based on a 1920 x 1080 chip would be inherently progressive. Processing is not an issue. The Sharp 10000 already properly deinterlaces film-based 1080I material by applying 3-2 pulldown (in other words...consumer-priced 3-2 reversal for 1080I is a reality) and scaling is easy math to either 720 or 1080 resolution. In fact, scaling looks *better* the higher you go because you put more pixels between you and the source resolution...so all things being equal an image scaled to 1080P will look smoother and more "analog" than one scaled to 720P even if the added resolution adds no percpetable detail.


Oh...did I mention that 1080P beats out 720P by more than *twice* the resolution?

 

[tbhausen]

Quote:

In the example I gave, 4 adjacent pixels display the same information. The idea isn't to display more information, but to dispaly it with redundant pixels so that bad pixels go unnoticed. [/b]

Fair enough, but you'll lose brightness due to the gaps between the four pixels.


Todd/Indy

 

[tbhausen]

Quote:

Originally posted by DaViD Boulet that's a very short-sighted approach on the part of Joe Kane. The real answer is not having to make a choice between 720P and 1080I...but to simply choose 1080 P.

Who's broadcasting 1080P? Considering bandwidth issues, will we see it anytime soon? I'm a realist, and to me 720P seems the best solution for high motion (e.g. sports) for the forseeable future.


Todd/Indy

 

[HarmsWay]

Quote:

Originally posted by Art Lloyd Ignorant question: Is it the cost of the chips that is the major factor in projector pricing?

I would guess that in terms of production cost per unit, the optics would be more expensive than the chips. Development cost for the chips are obviously much more expensive than for optics though, so it would come down to number of units to spread the development cost over. Projectors are still very low volume, so they've got to share chips with other manufacturers or with their RPTVs. All we need to do is convince all our friends that they need to buy high res. projectors.

 

[DaViD Boulet]

Todd,


Sure for live events like sports, 720P is a great compromise...not full 1080 HD res but no interlacing uglies... No argument here.


However, for *film based material* which I would imagine would encompass much (or at least some) of what an HT enthusiast would watch on his/her projector...any "interlaced" 1080 film-sourced signal could be properly deinterlaced to a "real" 1080P signal with 3-2 pulldown. Why not have a 1920 x 1080 projector to watch it on? I'm sure 720P upconverted to 1080P would look spectacular as well.


In other words...all 1080I "film source" material *is* 1080P material. Thinking of it that way...there's a lot of 1080P broadcasting going on right now


Also, 1080P at 30 frames per second uses about the same bandwidth of 720P at 60 Hz...so there are other options for "live" broadcasting giving our current bandwidth limitations once HD cameras can be designed affordably to do the job.


-dave

 

[tbhausen]

No argument with your points, either. Heck, for me whose vision is not the best anyway, I think I'll be much happier with the state of technology of display devices for quite awhile. Broadcasters, OTOH, have a LONG way to go. Sure is a lot of garbage out there...


Todd/Indy

 

[KBK]

"Put all your equipment in the back of the room next to the projector. You will have short cable runs and you can upgrade cables easily. Run only speaker wire to the front. You didn't put your equipment in the back of the room when you had it hooked up to your TV did you? I have just saved you thousands of dollars. I wish I would have done it."



I hardwired the PJ and the HTPC togehter, wt ithe HTPC right under the PJ. No connectors, everything was soldered.


The best way to get away from bad broadcasting, is to stop watching all brodcast media of any kind, period That's what I did. If they only produce crap, then shut it ALL off......

 

[JHouse]

Quote:

Originally posted by KBK I hardwired the PJ and the HTPC togehter, wt ithe HTPC right under the PJ. No connectors, everything was soldered.

So other than making it hard to swap out equipment, did the hardwire help in any way?