Monday, June 06, 2022

Review of Quanta Image Sensors for Ultralow-Light Imaging (IEEE TED June 2022)

[As mentioned in some recent comments on this blog, the latest (vol. 69 no. 6, June 2022) issue of IEEE Transactions on Electron Devices has many interesting papers related to image sensors. I will post summaries here in the coming days.]

In an invited paper in the June 2022 issue of IEEE TED, Jiaju Ma et al. write:

The quanta image sensor (QIS) is a photon counting image sensor that has been implemented using different electron devices, including impact ionization gain devices, such as the single-photon avalanche detectors (SPADs), and low-capacitance, high conversion-gain devices, such as modified CMOS image sensors (CIS) with deep sub-electron read noise and/or low noise readout signal chains. This article primarily focuses on CIS QIS, but recent progress of both types is addressed. Signal processing progress, such as denoising, critical to improving apparent signal-to-noise ratio, is also reviewed as an enabling co-innovation.

https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9768129









9 comments:

  1. Another review paper! This is just one in a long string of ... papers that survey the literature in a general way and ...

    YL

    ReplyDelete
  2. thanks for sharing. And thanks to mr Fossum, his invention literally changed my life. I joined Micron in 2004 on DRAM, then switched on image sensor in 2005, first smartphones, then automotive (with a stint in DSLR); so much learning and fun through all the challenges. Thanks to mr Fossum.

    ReplyDelete
  3. For me, 0.19e-@340uV seems small considering signal line to SF feedback etc. I'd like to see the separation of noise components. Can I see a demonstration somewhere?

    ReplyDelete
    Replies
    1. https://link.springer.com/book/10.1007/978-3-642-18443-7

      Delete
    2. The referenced book by Seitz won't help you with a demonstration, and anyway, it also did not anticipate deep-sub-electron read noise in CMOS image sensors. You might be able to buy a camera from Gigajot and then it is easy enough to measure the noise yourself. Further, one cannot collect the experimental photon-counting histogram of Fig 3 without this level of noise. (Fig 3 is measured for a single pixel and 0.12e- rms read noise was extracted.)

      Delete
    3. Was the noise measured for a chosen pixel, or averaged from multiple pixels?

      Delete
    4. What is the difference between the reported cases as Fig. 3 shows 0.12 e rms while the table shows 0.19 e rms at room temperature? (and both "CMS" 8)

      Delete
    5. I am not an author on reference 32 but it is open access and I looked it up. The noise at 25C and 8 CMS cycles was 0.19e- rms peak (at the peak of the distribution, a now-common metric) and 0.22e- rms median, with 0.12e- rms listed as best single-pixel noise. At 20C and 8 CMS cycles the peak drops to 0.17e- rms and the median to 0.20e- rms. Please refer to the IEEE EDL paper for more details, including the full distribution.

      Delete
    6. Thanks, I see it now. Interesting choice to report rms for a "best single-pixel noise".

      Delete

All comments are moderated to avoid spam and personal attacks.