Sunday, September 18, 2011

Samsung Alters Buffer Pixels to Improve Dark Rows/Columns Performance

Samsung patent application US20110205405 proposes an improvement of dark columns and rows ability to withstand a bright local spots in active area close to them. The dark columns/rows themselves can be effectively shielded from the light by metals. However, if there is a very bright spot in the active array next to the dark rows or columns, the charge can bloom into the dark area and distort the dark signal. To somewhat alleviate this, Samsung proposes to connect TG and RST gates of the few buffer pixels on the border between active and dark pixel area to VDD:

The idea is quite simple and straightforward and can be effective, to a degree. However, I did not see it patented anywhere. As an improvement of it I would couple FD directly to VDD too, in addition to TG.


  1. totally nul. I can just add a Nwell gardring which is more efficient :)

  2. But in this configuration feedthroughs from TG and RST are absent in the dark pixels, thus producing dark levels completely inconsistent those in the regular pixels. How can this scheme work?

  3. @ "feedthroughs from TG and RST are absent in the dark pixels..."

    These are buffer pixels sitting in between the active pixels and dark pixels. They are not working in a sense that their output is completely wrong and, basically, thrown off.

    @ "totally nul. I can just add a Nwell gardring which is more efficient :)"

    Adding Nwell as a buffer zone between active and dark pixel can work, in principle, but it has few disadvantages in comparison with Samsung idea:

    1. The Nwell destroys the array rows/columns periodicity and few neighboring columns/rows would produce wrong results, both in active and dark pixel areas.

    2. Quite often the PD implant is deeper than Nwell, so it might work better for red light blooming in FSI sensors or for blue/green blooming in BSI.

  4. I repeat again that this patent is totally nul like that I can patent a pixel with 100 mosfets in parallel in order to reduce the disparity. You will never find this one, but what is the usefulness???

  5. Please consider :
    - this is not yet a patent, but still a patent application,
    - the decision whether this will ever become a patent is in the hands of the patent office and not in the hands of Samsung in tis case,
    - taking into account what was announced last week about Canon, it is better to have too many patents than too less.

  6. In addition to what AT said, remember that engineers in most companies are encouraged and rewarded for filing invention disclosures. So, some guy has an idea, good, great or not so good, and files a report. In most companies, the report develops a life of its own, going thru a patent committee that may or may not understand the technical idea in depth, and then on to patent attorneys who draft the application and submit it. So, when you look at applications, they are not much more than an idea that may or may not ever get used.
    It is silly to get excited about these applications. Wait for granted patents, even tho patent examiners also have little idea about prior art or if an idea is good or bad. That said, best not to get excited until you see it being used in a product.
    Meanwhile, good to see these new ideas surface early, good or bad, and we all get a chance to reflect on their usefulness.

  7. to Anon,
    Why do you keep referring to the "US guy"? There is a whole team of engineers employed by Samsung in the US and hundreds in Hwaseong and Giheung in South Korea working on image sensors and cameras.
    Next time, just look up the patent application. It is easy to find the inventors' names.


All comments are moderated to avoid spam and personal attacks.