Tuesday, November 15, 2011

CMOSIS CMV Roadmap

CMOSIS showed a roadmap of its CMV series of fast sensors with global shutter at Vision 2011 fair in Stuttgart, Germany on Nov. 8-10, 2011:

17 comments:

  1. Big mono ? :) I wonder if they will cost equaly "big".

    ReplyDelete
  2. well - when talking about FPS, the resolution is somehow interesting:
    CMV 300: 648x488
    CMV 2000: 2048x1088
    CMV 4000: 2048x2048
    CMV 12000: 4096x3072

    I found 2 things worth mentioning at my Vision-Exhibition visit:
    - USB3.0 is everywhere. It seems to be not yet ready (software standardisation ongoning), but more or less all camera manufacturers are pushing it. It seems to be a lot cheaper than GigE. One company is offering a 1.3MP global shutter USB3.0 camera for 189€... IDS was showing a PC with 8 cameras (1.3MP, >25fps) at one onboard-USB3 port of a core i5 PC (using a Hub). Constant grab/display with less than 25% CPU load
    - a lot of camera manufacturers use the e2v EV76C560 cmos sensor.

    ReplyDelete
  3. Sensor Spec - Slide 10:
    SNR spec given as 64 dB is incorrect. This is the dynamic range (32Ke- fullwell/20e- dark noise).
    Max SNR is 45dB for this sensor.

    Parasitic Light Sensitivity (PLS) is 1/50000, which is much lower than sensors from other companies which quote something like 1/500. what would be the reason for very good PLS in CMOSIS sensors.

    ReplyDelete
  4. In CMOSIS snapshot sensor, the image signal and the reset signal are stored on two capacitors. A substraction is applied during the readout. The leakage in these 2 capacitors are very similar, so the PLS is low. But the problem is when there is a strong illumination, you can have a "black sun" effect. And the black sun effect is not better than other CMOS sensors.

    ReplyDelete
  5. @ Piotr Maliński

    Prices are considered reasonable for their performance by our customers.

    @ Anon - FPS

    The CMV naming is steered by resolution and marketing.
    CMV = CMOSIS Machine Vision
    CMV300 = ~300k pixels
    CMV2000 = ~ 2M pixels
    CMV4000 = ~ 4M pixels
    ...

    As soon as we introduced the 70Mp, we decided to abandon the lengthy row of zero's and called it CHR70M - CMOSIS High Resolution 70 Mp.

    @ Anon 2.

    Indeed, on slide 10, DR was meant.

    The origin from the good PLS lays in the CMOSIS patented 8T pixel architecture, where reset and signal are stored in dedicated light insensitive capacitors. See

    ReplyDelete
  6. See ... http://www.cmosis.com/assets/images/Global%20shutter%20CIS%20for%20machine%20vision.pdf

    ReplyDelete
  7. I've tested CMV2000 and the black sun comes quite early.

    ReplyDelete
  8. "dedicated light insensitive capacitors"
    a capacitor is always light insensitive. what is the sensitive is the PN junction attached to this capacitor.

    ReplyDelete
  9. @Tim Baeyens in my case in astrophotography anything bigger than 2/3" and mono is Kodak... and big Kodak powered cams aren't cheap, and they are from the pro/lab branch ;)

    When Basler will have the GigE version of Ace with CMV I'll probably get one for some tests from the local Basler partner. 2048x2048 would be handy for lunar and solar imaging. I'm curious of its noise and artifacts performance in such conditions.

    ReplyDelete
  10. @Piotr- they don't call it the black sun artifact for nothing ;)

    ReplyDelete
  11. "Anonymous said...

    "dedicated light insensitive capacitors"
    a capacitor is always light insensitive. what is the sensitive is the PN junction attached to this capacitor."

    A MOScap is light sensitive. And btw even MIMs can show dielectric absorption ;) So your statement is false.

    ReplyDelete
  12. It's just a matter of filters and keeping good histogram fill. Then black sun artifact maybe won't be a problem... except maybe of sunspots - http://www.rkastrofoto.appspot.com/site_media/astro/orig/sun-25-09-2011/spot-0846-b.jpg ;)

    ReplyDelete
  13. "A MOScap is light sensitive. And btw even MIMs can show dielectric absorption ;) So your statement is false. "

    if the D/S are connected to GND and the signal stored on the gate will be insensitive to light is the MOS is in strong inversion mode. Am I wrong again?

    ReplyDelete
  14. Strange to see 12Mp sensor samples only available in March 2012 as it was introduced last year in 2010: any issue with it?

    ReplyDelete
  15. "if the D/S are connected to GND and the signal stored on the gate will be insensitive to light is the MOS is in strong inversion mode. Am I wrong again? "

    Sadly, yes wrong again. The fact you connect it to ground does not make the capacitor insensitive to light. Your ground should provide the current as function of light. If the capacitor would be insensitive to light, the ground current would not change....that's the theoretical part you're probably lacking but for the rest reply I agree from the engineering point of view.

    ReplyDelete
  16. Well, what matters is the deplacement current inside the capacitor. If there is a current flow at D/S areas, there is no deplacement current inside the capacitor. In this case, what is important is the charge stored on the gate which will not sensitive to light. Am I wrong again?

    ReplyDelete
  17. Claiming that the light sensitivity doesn't matter as your ground node supplies the light sensitive current is okay by me (from a practical standpoint). Claiming that they're insensitive is not. Think about it. ;)

    ReplyDelete

All comments are moderated to avoid spam.