SemiIsrael published an article about Advasense. While it mostly reiterates known stuff, there are two interesting quotes:
"The company has changed its business model during the economic downturn of 2009 and finances itself by commercializing its FCP (selling non-exclusive licenses to strategic partners). The funds are used for the development of new CIS technologies and products."
"The company is developing new generations of the FCP technology that pave the way to sub micron pixel size with excellent image quality. A 12Mpixel imager with 0.9u pixel size can be housed in 1/4" optical camera module – the same physical size used for a 5Mpixel 1.4u pixel imager."
aptina and advasense are working together. i am very impressed with this combination and look forward to seeing their products.
ReplyDeletethanks ISW for posting this. :o)
what's the use of so many pixels in such a small place?
ReplyDeleteWhere in the post one can understand that ADVASENSE is working with APTINA ?.
ReplyDelete> what's the use of so many pixels in such a small place?
ReplyDeleteNot all pixels are equal. Advasense 0.9um pixels have a high full well, essentially same as 2.2um pixels from few years ago. So, in good light the image quality approaches to one of the much bigger pixels, on pixel to pixel basis.
In low light there is no magic, each pixel gets less photons meaning lower SNR. As with other sensors on the market, one can combine few pixels output and trade the resolution for a better SNR. This way one can get back to the old days when resolution was lower and low-light performance better, but with one important advantage: no color interpolation artifacts (remember Foveon demos?). Since 0.9um-based sensor effectively oversamples the lens-limited image, the Bayer interpolation artifacts can be eliminated.
So, in some senses our 0.9um sensor generation can be superior to 1.4um and, probably, to 1.75um sensors. One should just combine them with the right image processing routines.
Vladimir Koifman,
Advasense
that aptina is working with advasense is obviously not addressed in the post--it is an aside that is relevant to an understanding of advasense being an important player in the cmos imade sensor space.
ReplyDeletemost on this board are probably aware of this since they have industry experience. the joint venture was discussed in an earlier ISW post which, as most of his posts, i found most enlightening. here's a link:
http://image-sensors-world.blogspot.com/2009/08/aptina-and-advasense-jointly-develop.html
I heard many rumours on ADVASENSE-APTINA joint development. No clear statement or press release saying that it's true
ReplyDelete>"So, in some senses our 0.9um sensor generation can be superior to 1.4um and, probably, to 1.75um sensors. One should just combine them with the right image processing routines."
ReplyDeleteI'm not so sure you could overcome dramatic diffraction effects at this scale with post-processing routines, and then surpass 1.75µm and 1.4µm sensors performances. Even at 1.1µm, optical sensitivity is reduced, and both optical and electrical crosstalk can strongly degrade the pixel efficiency.
F.H.
True, there are many barriers on the way to create a competitive 0.9um pixel. Advasense pixel use a heavy doped photodiode extending well below 3um depth. Most of this depth is neutral, rather than space charge commonly found in other 4T pixel designs. This ensures that photocarriers generated in one pixel do not move to its neighbors, reducing the crosstalk.
ReplyDeleteThe capacitive crosstalk between floating diffusions is another type of crosstalk that is reduced by a factor of 5 to 10 by the virtue of feedback controlled pixel operation.
You are right that diffraction requires an attention here. Our estimations show that a careful design can minimize pixel sensitivity degradation.
Good Luck - Advasense!!!
ReplyDelete