Tuesday, August 18, 2009

Rumors Galore

The stuff below is unconfirmed rumors and should be treated as such. Although I did some cross-checks to verify them, I can not guarantee they are correct.
  • Toshiba and ST 1.4um FSI pixels show very high QE, approaching 70% in green. That good FSI pixels set too high bar for BSI competition.
  • Sony: In addition to the recently announced 1.4um pixel based camera modules, Sony has another module having a lot of advanced high-speed functions integrated in the camera module form factor. The module has a high speed 5-frame memory and implements a multi-frame low-light image stabilization. The module also has face detection feature integrated.
  • Microsoft "Project Natal" demo and PDKs distributed to the game developers are based on Primesense 3D solution. Aptina design win for the next generation Xbox360 is in fact related to the Primesense device.
  • DxO Labs is up for sale. The company failed to raise money recently and forced to look for a buyer. There are 6 potential buyers doing due diligence at DxO. The rumor is that the acquisition deal is near.


  1. That good FSI pixels set too high bar for BSI competition.
    So you think BSI won't be able to compete with high efficiency FSI? I'm not sure I have enough information about the cost of the Toshiba/ST process to draw that conclusion ... much less the cost of BSI. Also, what about noise?

  2. > So you think BSI won't be able to compete with high efficiency FSI?

    It appears that 1.4um pixel in 65nm FSI process can perform better than the first generation 1.4um BSI pixel. Both Toshiba and ST are working on 1.4um BSI pixel which is supposed to be better than their current FSI. So, next year my claim will not necessarily hold. But as it stands today, at 1.4um FSI delivers better performance than BSI. Talking about future, certainly BSI is the way to go for 1.1um pixels and beyond.

    > Also, what about noise?

    ST and Toshiba noise is low enough to be non-factor in SNR10 - they both reach 100Lux.

    Talking about cost, I agree that ST and Toshiba have a challenge here. Even more so since manufacturing in Japan and Europe is more expensive than in Taiwan and Korea. I'm not sure what they plan to do with it.

  3. Can you share where can see Toshiba/ST's high QE and SNR10 reach 100Lux.
    Thanks a lot!!

  4. Oh, this is easy to guess - Nokia new phones. I can not think of a better place to show up.

  5. Dear Image Sensor,

    thank you very much for your site : daily news for many people, always relevant comments, always good information.

    Regarding this specific topic and cost issues, and due to BSI process complexity, it is not sure FSI sensors manufactured in Japan or in Europe exhibit higher costs than BSI manufactured in Taiwan, as an example. Added to good performances of Toshiba/ST FSI sensors, as you say, the question of BSI need for 1.4µm is very relevant then. This is a difference between technical performances and pure marketing, where the first to propose a new process, even not needed by the customer, will make very high margins.

    No real doubt about BSI need for 1.1µm however...

    Having said that, image sensor manufacturing process is quite complex and it is again not sure that a well designed BSI process, even manufactured in "high-cost" area, won't compete with BSI process manufactured in "low-cost" areas.

  6. Despite laying off 20% of its workforce in 2009, DxO's B2B division is still not profitable by now. DxO indeed failed to raise more money from investors and was recently forced to raise a bank loan of a few millions. They're certainly in a weak position.


All comments are moderated to avoid spam and personal attacks.