Tuesday, June 22, 2010

Forza Announces HDR Sensor Development

PR Web: Forza Silicon announced that it is in a joint development effort with a leading image sensor foundry to bring to market the next generation of HDR CMOS image sensors.

These techniques all require extensive modifications to the traditional pixel design that can lead to degradations in sensitivity and dark-current performance. Modulating the integration time instead of the conversion gain can reduce the pixel modifications required” said Forza CTO Daniel Van Blerkom. “The next generation of HDR image sensors will require customization and optimization of the pixel, analog signal chain, and the image processing algorithms. We are happy to announce our co-operation with our foundry partner to develop this sensor IP.

Forza demonstrated an image-aware tone-mapped image set in its labs in Pasadena, California, which represents a close approximation of how a natural scene looks to the human eye.

I'd guess that Forza's foundry partner is TowerJazz.


  1. Pixim sensor has used already this method. What is new about their solution ?

  2. All the small CIS companies work with Tower now :)
    What are the reasons for this ?

  3. if it works, why does it have to be new?

  4. "Pixm sensor has used already this method".. Really? What method is that? I expect the Forza approach is completely different from the Pixim ADC-in-every-pixel approach.

    Of course tone mapping as a concept is hardly new, but there are many many ways to do this algorithmically incl. hardware and software.

    Frankly, Forza's design team is probably more advanced than any other on the planet, including Sony and Aptina. However, as a small foundry partner, their influence on processes for advanced pixels is small, compared say to another fabless company like Omnivision.

    Smaller foundries like Tower-Jazz have business models that are more flexible in order to capture business from smaller production runs. It is a good thing. Unfortunately, in small production runs, it is hard to get a grip on yield improvement for image sensors since they use a standard baseline process with tweaks added for particular sensors. Somehow the tweaks always seem to lead to lower yield. Also, leading edge designs tend to push the envelope for a particular process, also impacting yield.

    Personally, I am always impressed by the Forza team and have worked with many of the lead guys (gals) either at JPL, Photobit or as a customer at Siimpel. I wish they were part of my team at Samsung!

  5. Wow, what a great recommendation for Forza!
    As a matter of fact, me too have heard many good words about Forza team from assorted sources.

    Not to diminish Forza achievements, there are some other nice teams on this planet also doing superb job. To me, it's hard to rank the sensor design groups given different challenges and obstacles they are struggling with - then the net result is hardly comparable.

  6. Indeed, there are other excellent teams, some just down the street here in Pasadena, others coming out of FillFactory, and a few others. Forza's team is pretty large for a custom design company - I am guessing 50 people, mostly in design. So, depth and breadth-wise, for me, today, it is Forza.

    Knowing Alex Krymski and LinPing Ang well, they are also very talented and another team I usually recommend when asked. I don't know CMOSIS and Caeleste that well but they both have some outstanding people and for European inquiries, I usually send people in their direction as well. Sorry if I am missing others out there. Maybe ISW can add more specific categories to the nice list of companies on this page?

  7. @ " Maybe ISW can add more specific categories to the nice list of companies on this page?

    Well, this list was intended to keep track of obscure and hard to find names. Then I got few requests to add more known companies and added them too, one by one, by request. Still, many well known companies are not there, like Panasonic or Sharp. And some companies are not in business anymore, like 3DV or CmoX, but their web sites are still on-line, and I just keep them as hard to find links. Yet some other companies are not exactly sensor ones, like Cap-XX or Apical.

    There is a different, more complete list of active "core image sensor" companies here:


    What kind of categories should be in the list?

  8. Multiple exposure based HDR solution needs to read out several differently exposed images to form one output image. Pixim uses in-pixel ADC to transmit these images in high speed.

    Their design is still that only multi-exposure solution sold in volume...

  9. there should be a very strong electric field someone's head because his opinion on other design teams is always so polarized !

    Please screen out such kind of comments on this blog, it's a self promotion of small circle and misleading to general forumers ...

  10. @ "Please screen out such kind of comments on this blog, it's a self promotion of small circle and misleading to general forumers ..."

    Actually, I agree with Eric that Forza has a great team. There are other great teams on this planet, which, I think, quite competitive with Forza, but this is another matter.

    However, I disagree with Eric on his comments about the yield problems at Tower-Jazz. Tower makes full-frame DSLR and X-Ray sensors, some occupying a big part of the wafer. These kind of products are impossible to produce if defect density is not low enough. Now, if we compare yields of mobile VGA sensor with FF DSLR one, they are not directly comparable.

    Also, my sources tell that TowerJazz makes quite deep CMOS process modifications to improve image sensor quality. It's not just "tweaks" to the standard logic process flow.

    Also, in many cases Tower provides ready, silicon-proven pixel designs for its customers, and it accumulated a big pixel portfolio over many years. This greatly simplifies image sensor development for the smaller companies.

  11. Anon - that was funny about the electric field but it is not self-promotion. This blog entry IS about Forza and I was responding to the first 3 comments. I don't have any current business relationship with any of the parties I mentioned.

    ISW - I was trying to make a general comment about why people use smaller foundries like Tower-Jazz, in response to the 2nd comment. "Baseline process" means baseline CIS process, which then gets tweaked. It is a very good thing that Tower-Jazz does these tweaks as does other smaller-run foundries. And, it is a fundamental axiom to manufacturing engineering that yield and production-run volume go hand-in-hand. It has nothing to do with any specific foundry. I do agree with your last two paragraphs on Tower.

  12. For ISW -- suggestion on categories (re: 23/6/10 2314)

    -High volume (mostly)
    -Custom and niche (mostly)

    You could list a company more than once.

    This site and its keeper should get some kind of recognition for its increasing value to the image sensor community!

  13. These would be huge lists, if made more-less complete. Just image sensor companies, including defunct, would have, may be, 150 names or so. Optical components list can be even longer. Testing list would be at least 100 names, probably more.

    If I put everything on the left panel, it would be too long. Unfortunately, Blogger software does not support collapsible lists. I'll think what can be done instead.

  14. Well, I was really thinking you could just categorize the companies already in your list and add new ones as time goes on as you do already. You don't need to be the authoritative supplier-list keeper, unless you want to be.

    "The problem with feeding St. Bernard puppies is that eventually you wind up feeding St. Bernard dogs." - Marty Sokolowski, NASA HQ circa 1991

  15. Eric, I compiled a list of image sensor companies here:


    If it looks good enough, I'll add it to the blog.

    As for the whole ecosystem list, including camera module and lens makers, package vendors, software and test equipment, AF motors, simulation tools, etc - it's too big a task. And there is a little sense to have grossly incomplete lists.

  16. Dear Image sensor world,

    Could you please add NIT in the "custom design" list too ?


  17. Is Kodak still making image sensors? The company name is not on your list.

  18. Kodak is under niche market category, same as DALSA.

  19. The list looks great! Sorry for the delay, was driving from California to New Hampshire.

    You might add Luxima (not sure why this is different from Alexima or if it is still alive).


All comments are moderated to avoid spam and personal attacks.