Pixpolar announces an image sensor simulator. The intention of the simulator is to show the company's non-destructive CDS readout, but it can be used for other purposes as well. Number of photons per pixel can be set in histogram widget. Readout noise, dark current and exposure time can be set in the main settings window. The simulator works as an on-line application.
Thanks to JS for the link!
Is this pixel like the floating gate CCD readout circuit ??
ReplyDelete-yang ni
Hi Yang Ni,
ReplyDeleteThe simulator is completely general meaning that pixels featuring Non-Destructive CDS (NDCDS) read-out can be of the floating gate (Skipper Amplifier), Internal Gate (IG, e.g. DEPFET) or Modified Internal Gate (MIG) type. Pixels featuring Destructive CDS (DCDS) read-out can be of the CCD or 4T CMOS type. The properties of the pixels are defined by general parameters which are the read noise, dark current rate, quantum efficiency and DCDS / NDCDS read-out. Besides the exposure time, frame rate (optional) as well as the minimum and maximum scene intensities at the image sensor pixel level can be chosen freely. I hope that this helped to clarify things - if not please don't hesitate to ask.
By the way, Artto, you know that the multi-sampled floating gate amplifier goes back 15 or 20 years before its reinvention as the skipper amplifier? I hope you don't think those things are synonymous. It was conceptually invented about the same time as the CCD, but not for the CCD by Engeler, and then further applied to the CCD by Kosonocky but perhaps best by Wen subsequently.
ReplyDeleteHi Eric,
DeleteI wasn't aware of that - I knew only that the minimum overall read noise record (0.6 e) for the floating gate configuration was achieved with the skipper amplifier by James Janesick.
what is the fundamental advantage in NDCDS please ? Suppose that the exposure time is 1s, 2 different scenarios:
ReplyDelete1. read 10 times during the exposure and sum the 10 read results.
2. read 1 times by using 10X multiple samples
If the noise is white enough, do you think that the 2 results are equivalent ?
Thanks !
-yang ni
It depends on lighting.
DeleteIf there is enough light and all the individual read-outs are clearly in the shot noise limited region then it doesn't matter whether you read non-destructively or destructively.
On the other hand, if the read noise in a single read-out is proportional to the shot noise then, in the case of destructive readout, the read noise piles up and the signal-to-noise ratio decreases.
These cases can easily be tested with the simulator.
Ah ! Artto, how can you do NDCDS with a 4-T pixel please ?
ReplyDelete-yang ni
It's not possible with a traditional 4-T pixel.
DeleteHi Yang Ni,
DeleteNDCDS is not possible in the traditional 4T pixel since the signal charge is destroyed in the read-out process. In order to achieve NDCDS the following requirements should be fulfilled:
(a) two storage nodes are required inside silicon,
(b) complete depletion of the two storage nodes must be enabled,
(c) complete signal charge transfer between the two storage nodes must be enabled in both directions,
(d) coupling of the signal charge to the read-out transistor must be stronger in one storage node than in the other,
(e) the read-out bias configuration should be independent of the fact in which storage nodes the signal charge is situated, and
(f) beneficially signal charge can be transferred in one direction between the two storage nodes without altering the bias configuration of the read-out transistor.
In order to fulfill the NDCDS requirements in a MIG pixel at least one FET and an additional gate are required. By using different bias configurations on the two gates the signal charge can be transferred between two storage nodes situated underneath the two gates. Besides, different bias configurations on the two gates can be used for pixel selection, read-out and reset.