Friday, February 24, 2017

Invisage Explains its 1.1um Pixel Global Shutter Operation

Invisage presented a paper "Device design for global shutter operation in a 1.1-μm pixel image sensor and its application to near infrared sensing" by Zach M. Beiley, Robin Cheung, Erin F. Hanelt, Emanuele Mandelli, Jet Meitzner, Jae Park, Andras Pattantyus-Abraham, and Edward H. Sargent at SPIE Physics and Simulation of Optoelectronic Devices XXV Conference held on Jan. 30-Feb. 2 in San Francisco.

The global shutter operation is based on QuantumFilm QE dependence on the bias voltage:


  1. Replies
    1. this means that QD can emit light like a LED, so you can illuminate the scene like a LED and then fast switching back to get an image... That's cool!

  2. Albert Theuwissen - Harvest ImagingFebruary 25, 2017 at 9:46 PM

    Can you integrate the next frame while reading the previous one ? I don't think so with this concept. Then you do have a situation like the ones with full-frame CCDs, while real global shutter devices can combine the intregration of the next frame with the readout of the previous one takes place. This is the case for interline-transfer CCDs, frame-transfer CCDs and 4T-CMOS imagers. Some call this architecture also a pipelined snapshot, or pipelined global shutter.

    1. I think it is possible in this architecture since the circuits are under the "quantum film." This gives more space for the frame buffer and readout while integrating the next frame. Maybe they don't do this but they could do it. The main issue will be read noise, although I understand there are other issues that are currently worse - e.g. dark current on a non PPD node. I am still optimistic that the NIR response of the "quantum film" will prove itself as a market niche for Invisage.
      The figure shown above has some issues and needs work!

    2. I think they tried to combine showing peak sensor QE at 940nm and charge integration into the same plot as a function of bias voltage. While that means you'll see negative QE on a plot, it's not that odd if you think about it in the time domain of integration time vs non-integration time.

    3. Albert Theuwissen - Harvest ImagingFebruary 26, 2017 at 10:53 PM

      To Eric : I do agree that you can put a memory node in every pixel, but why would you switch off the sensitivity in that case ? I can not recall any CCD or CMOS imager with a global shutter that shuts off the sensitivity of the pixel while you store/shift the pixel info into the memory node.

    4. In the paper, they target it specifically to structured light 3D sensing, where a light pattern is projected for a very short time. That way, the average projection power is kept low, while the high peak power can over-power a sunlight (behind a narrow bandpass filter at 940nm).

      If this use case, they do not need the full pipelining. They have a plenty of time to read the array out between the short structured illumination pulses.

      Other use cases might be more problematic, as you noted.

    5. Vladimir, in this case, there is less need to switch off the absorption. Every snapshot pixel can be used for this purpose.

    6. True, any global shutter pixel with short enough exposure time works for that. The advantages of Invisage are (a) 1.1um pixel with GS and (b) decent QE at 940nm.


All comments are moderated to avoid spam.