Thursday, September 29, 2011

Samsung Expands its BSI Line on High and Low Ends

Business Wire: Samsung announced 1.2MP 1/8.2-inch CMOS image sensor SoC, the S5K8AA. The new sensor is based on 1.4um BSI pixels and aimed to camera modules shorter than 3mm for the slim mobile devices. The S5K8AA supports 30fps HD (720p) video and 60fps at VGA resolution. It fitts into a 5.5 x 5.5mm camera module with an extreme 2.8mm low profile.

Samsung’s S5K8AA image sensor is currently sampling to qualified customers.

Business Wire: Samsung also announced 1/2.3-inch 16MP image sensor, the S5K2P1, based on 1.34um BSI pixel and designed for high-performance, advanced smartphones as well as digital still cameras and camcorders. The S5K2P1 is fast - it supports a native 16:9 format video at up to 8.3MP resolution at 60 fps and full 16MP resolution for point-and-shoot images at 30 fps without shutter lag effect.

According to TSR, the DSC market is expected to reach 154M units in 2012 where a substantial portion of up to 48.2 percent is forecast to be adoption of CMOS image sensors as the main sensor device.

Samsung’s S5K2P1 image sensor is currently sampling to select customers. Its mass production is scheduled for November 2011.


  1. Why no low light sensitivity number? Low light sensitivity is biggest selling number for bsi sensors. It seems not all bsi sensors have the same low light sensitivity. Am I right?

  2. Low-light sensitivity is one selling point. Another one is reduced stack height, which is clearly what the company is promoting here. To me this suggests that the target market (handset OEMs) views end-product thickness as more important than low-light sensitivity as a competitive differentiating feature.

    In a literal sense you are correct stating that "not all bsi sensors have the same low light sensitivity". In the figurative sense - that the underlying technology is just not as good as a competitor's BSI in terms of sensitivity, so the company is omitting discussion of sensitivity - I don't think you are right.

  3. Is the primary measure for low light-sensitivity mV/lux-sec? If so, why don't companies include this in their spec sheets?

  4. mV/lix-sec is not the whole story. One needs to know QE, conversion factor, read noise and color crosstalk. Then one needs to calculate an optimized color conversion matrix and use it to normalize the above numbers, so it could be compared between different sensors. For an experienced engineer it takes a couple of days per sensor, mainly the color conversion matrix optimization.

    So far the best figure of merit I'm aware of is Nokia SNR10 - best, but not ideal.


All comments are moderated to avoid spam and personal attacks.