Monday, November 08, 2010

Sony Announced Super 35mm-class HD Video Sensor

Sony has developed "Super 35mm equivalent" CMOS sensor for the professional Digital Cinema camcorder line. The "Super 35mm" array size is stated as 1.7-inch type, with diagonal of 27.1mm. The sensor is capable to deliver 1080p60 HD video and relies on column-parallel ADC architecture. Its pixel size is said to be 4 times bigger than that of DSLRs providing SNR of 63dB.

This is the first time Sony's digital cinema line uses CMOS image sensor. The previous generations relied on CCDs.

11 comments:

  1. Any idea about pixel size ? or pixel number ?

    ReplyDelete
  2. if it is 1920 x 1280 pixels and the diagonal is 27.1mm and the format is 16:9 then the pixels are 12.3 microns

    if that is 4x bigger than the DSLR pixels then that says the DSLR has pixels of about 3 microns. That seems believable to me.

    ReplyDelete
  3. 63 db S/N is 1400:1, implying a full well of 2 million electrons. That is really unlikely in a 12.3 micron pixel. Maybe it is on the order of 100k electrons. Then 63 db DR would put the noise around 70 electrons, not too unlikely for CMOS without reset noise subtraction.

    ReplyDelete
  4. sorry, not following the computation:

    63 db S/N is 1400:1, implying a full well of 2 million electrons

    ---
    can you explain the 2 million electrons conclusion? what read noise did you assume, 1428 electrons and if so, why?

    ReplyDelete
  5. 12.3 um pixel size in combination with PPD ? Could be challenging w.r.t. image lag. But is it really 12.3 um ??? A.T.

    ReplyDelete
  6. @ "...implying a full well of 2 million electrons."

    Actually, 2M electrons not that impossible like it might sound. Indeed, assuming 1V effective voltage swing on the PD, one needs about average 0.32pF to achieve it.

    For 1um PD one can get 1.5ff capacitance. For simplicity, let's assume the whole caps come from the area and neglect the peripheral component. Then 12um PD it would be 144 times more, or 0.2pF. Now, if Sony uses very abrupt junctions and more than 1V effective swing, I would not rule out that 2 million electrons are real.

    Once we have 2M electrons stored on the PD, one can read them in a single shot in 3T pixel or by parts in 4T one.

    @ "can you explain the 2 million electrons conclusion? what read noise did you assume, 1428 electrons and if so, why?"

    This is shot noise, not read noise. In most cases shot noise dominate when calculating SNR.

    ReplyDelete
  7. They mention that the area of the pixel is 4x the area of a 16.2 MPixel SLR sensor (area, not pitch). A 16.2 MPixel APS sensor for SLR has about 4.8 micron pixel pitch, that brings pixel pitch to 9.6 micron for a 4x larger pixel.

    On the other hand, super 35 film format is 24.89 mm wide, wider than Sony's sensor.

    ReplyDelete
  8. the text I read above says "Its pixel size is said to be 4 times bigger than that of DSLRs"

    ---
    size could mean linear dimension or it could mean area.

    ReplyDelete
  9. This comment has been removed by the author.

    ReplyDelete
  10. Yes, but the original Japanese text said "area".

    ReplyDelete
  11. The picture in the post too suggests that the pixel areas are compared.

    ReplyDelete

All comments are moderated to avoid spam.