Tuesday, August 28, 2012

Omnivision Announces NTSC Sensor for Automotive Applications

PR Newswire: Omnivision announces the OV7955, a low cost NTSC-resolution image sensor for mainstream automotive applications such as rear-view, surround-view and blind spot detection systems. The output is both analog and digital - a DAC is converting its digital output back into the analog domain:

"In light of the imperatives of the Cameron Gulbransen Kids Transportation Safety Act, we anticipate heightened demand for rear-view camera systems. To address that need, we are bringing to market an image sensor that is 40 percent smaller than competing products, and that is offered at an extremely competitive cost," said Jeff Morin, automotive marketing manager at OmniVision. "We are extremely pleased with the OV7955's low cost, small footprint and exceptional performance, and have already begun shipping this sensor to major automotive manufacturers from around the world for incorporation into their mainstream automotive designs."

The new sensor is based on 6um-sized OmniPixel3-HS pixels with sensitivity of 16 V/Lux-sec, maxSNR of 38dB and dark current of 16 mV/s @ 50C. The OV7955 is housed in a compact 5.7 x 5.4 mm aCSP package.


  1. Only 6300 electron well capacity for a 6.0um pixel?
    It appears that OV has made a decision to boost pixel gain at the expense of full well.


  2. please note 16v/lux.s !

  3. OV low cost entry though losing to Aptina 5.6um pixel sensors in sensitivity, dynamic range (5x less), and max SNR (38 dB vs. 45 dB). Aimed at some niche low cost automotive applications?

    1. Yes, Aptina has 16.5V/lux*s since sometime last year (at least sampling), I think.
      OVT Sensors SNR is really a bit low - which is also true for several other of their Omnipixel-3 Sensors...

  4. 16 V/lx*s sounds outrageous, but for such a large pixel (6um), and CG on the order of ~200+ uV/e, it is possible (depending on their QE peak and spectrum - I did an estimation from other, unrelated, QE/responsivity data that I had available).

    That high CG results in a very large voltage swing at the FD (~1.3V), which is limiting the max signal.

    They clearly targeted sensitivity over all other parameters. Back-up cameras have to work in bright light too, but I'm not sure if anybody's scrutinizing or prioritizing their image quality under bright light.


  5. OV's data page shows 12 V/lux.sec sensitivity.

    1. Interesting discrepancy. The data page is dated by June 2012. May be they have increased the sensitivity since then?

  6. A. Suppose FF=100% at 555nm, the enegrgy accumulated in 1s will be 36e-12/683= 5.27e-14J.
    B. Suppose QE=100% too, then 5.27e-14J will give 0.45*4.27e-14= 1.9e-14C.
    C. Suppose 1V excursion on FD, 16V/lux*s needs CFD=1.18fF.
    D. FWC = 7375e

    so the sensivity and DR match their claim if the global QE=80%. But I think that 12V/lux*s is more realistic (global QE=62%)

    Your comments please ??

    -yang ni

  7. Nice article. This is so much more than i needed!!! but will all come in use thanks!!
    Steering Rack


All comments are moderated to avoid spam and personal attacks.