It is not only an interesting project, but also under excellent guidance of prof. Meynants, an authority in the field. Amazing, I started in 1983 with my work at Philips Research, and one of the first topics i was focusing on, was noise reduction in our CCD output stages. Now we are over 40 years further, and still noise seems to be a hot topic. In my trainings the chapter about noise is entitled : "NOISE : you love it or you hate it!".
There is lots of room at the bottom when it comes to read noise. Getting someone to pay for this research in academia is the difficult part. The Dartmouth work starting about 10 years ago was covered by Rambus (mostly) DARPA (some), and NASA (a little bit). Devices without impact ionization gain have achieved input-referred noise medians in the 0.19-0.22 e- rms range using single, destructive-read CDS with high conversion gain and CMS for subsequent electronics noise reduction. (see review paper: https://doi.org/10.1109/TED.2022.3166716 )
The mechanism of noise (1/f etc) in small source-followers may (still) not be well-understood and MOSFET (and JFET) transistor noise study over decades has been like peeling an onion with new mysteries arising at every technology node. (Yes, the trap-ensemble model is probably not correct in my opinion.)
For single photoelectron detection, (e.g., dense quantum qubit readout) we need to drive read noise to the 0.15e- rms level for low error rate. As I mentioned recently at TU Delft, and also mentioned above by Aaron, it is really the noise tail part that needs to be tamed.
Plenty of room at the noisy bottom, and I look forward to seeing continued progress in this area. We are still investigating at Dartmouth, too.
Sounds like an interesting project. Coming up with a way to reduce the tail in the read noise distribution of photon counting CMOS would be great.
ReplyDeleteIt is not only an interesting project, but also under excellent guidance of prof. Meynants, an authority in the field.
ReplyDeleteAmazing, I started in 1983 with my work at Philips Research, and one of the first topics i was focusing on, was noise reduction in our CCD output stages. Now we are over 40 years further, and still noise seems to be a hot topic. In my trainings the chapter about noise is entitled : "NOISE : you love it or you hate it!".
you should say : noise you love it or you hate it, it's there forever !:)
DeleteIt is not only there forever, imagine that your parents hadn't made any noise 9 months before you were born ...
DeleteThere is lots of room at the bottom when it comes to read noise. Getting someone to pay for this research in academia is the difficult part. The Dartmouth work starting about 10 years ago was covered by Rambus (mostly) DARPA (some), and NASA (a little bit). Devices without impact ionization gain have achieved input-referred noise medians in the 0.19-0.22 e- rms range using single, destructive-read CDS with high conversion gain and CMS for subsequent electronics noise reduction. (see review paper: https://doi.org/10.1109/TED.2022.3166716 )
ReplyDeleteThe mechanism of noise (1/f etc) in small source-followers may (still) not be well-understood and MOSFET (and JFET) transistor noise study over decades has been like peeling an onion with new mysteries arising at every technology node. (Yes, the trap-ensemble model is probably not correct in my opinion.)
For single photoelectron detection, (e.g., dense quantum qubit readout) we need to drive read noise to the 0.15e- rms level for low error rate. As I mentioned recently at TU Delft, and also mentioned above by Aaron, it is really the noise tail part that needs to be tamed.
Plenty of room at the noisy bottom, and I look forward to seeing continued progress in this area. We are still investigating at Dartmouth, too.