Gpixel presents GLUX9701BSI, a 1” format BSI sCMOS sensor with a resolution of 1.3 MP (1280 x 1024) and large 9.76 μm x 9.76 μm pixels. The sensor is the first in a new family targeting extreme low-light imaging for both surveillance and scientific use.
GLUX9701BSI supports a dual-gain HDR mode, achieving dynamic range of 90 dB by combining 1.5 e– RMS read noise and 50 Ke– full well charge. A dedicated low noise mode further optimizes imaging performance with read noise of 0.8 e– and power consumption of 180 mW. Dedicated circuit and process engineering improves noise uniformity of GLUX significantly over previous sCMOS models with a close to Gaussian noise distribution behavior as can be seen in below figure.
The sensor offers two types of data output: 4 channel sub-LVDS and MIPI (CSI-2, D-PHY). The default frame rate at HDR of 30 fps can be achieved using both channel options with dedicated operation modes achieving frame rates of up to 120 fps.
A picture below demos the new sensor performance at 25 fps with <0.01 lux light intensity (measurement equipment limited).
A demo video is taken at 25fps under strarlight:
Gpixel's new Youtube channel also has a company presentation video:
For those who followed the comments below the recent postings about layoffs at onsemi: can you spot the difference in product portfolio, roadmap, technology and company environment? Can you name a single onsemi industrial cis that you would design in today vs gpixel? The aptina legacy in the low end probably. But no way in the higher end...
ReplyDeleteThere is definitely a huge overlap between their respective portfolios and both the output velocity of new sensors as well as the price performance point seem to be in favor of Gpixel...
DeleteThe layoffs at ON Semi won't be helping in that regard...
Quantization noise at 50ke and 12 bit is already 6e-. How can they achieve 1.5e read-noise??
ReplyDeleteby HDR, the high gain channel has far less FwC. Then HDR combination is needed
DeleteThen wont there be a transition point noise? I think this needs to be included somehow in the noise estimates. For some applications it should be fine, but for some others, it may have an impact.
DeleteIt depends exactly when and how they transition. They may or may not show an "SNR dip"... which is what you are worried about. Everyone doing HDR has this issue to deal with.
ReplyDelete