in the prehistoric days when we used to use analog outputs, we thought of this quantity as "Conversion Gain"... as in how many uV of output signal we could get for every arriving electron. How does this relate to efficiency, I am clueless.
To us dinosaurs, charge conversion efficiency evokes a calculation of how many electrons are actually collected and converted to a voltage from a total number of electrons that are generated by incoming photons...
Actually, most people I know, including myself, still use "Conversion Gain" or "Conversion Factor" terms. The first time I saw "Charge Conversion Efficiency" was in DALSA blog. I wonder if anybody else uses this term.
I agree with the above posts. I use the term conversion gain, when referring to the conversion of electrons to uV at the FD node. I use the term conversion factor for converting electrons at the FD to LSB at the output. In Dalsa's terms, you could become more "efficient" simply by cranking up the analog (or digital) gain - which you could say that, but I don't agree with the terminology, because I think what you are really trying to get at is the capacitance of the FD node.
Sorry for the question: can someone explain me how can CIS processes assure good maching (hence low PRNU) of very small FD capacitance? As an analog designer, I am used to matching problems even with larger capacitances (e.g. >50fF)...
CIS pixel array is an ideally repeating cell pattern. This greatly improves the matching, as compared with a generic analog circuits. One of the claims I've heard is that even transistor mismatches in the array are better than Pelgrom - hard to believe, but the claim came from a well-known very reputable person.
Another contributing reason might be the bad pixel correction implemented on many commercial sensors. If some pixel deviates too much from its neighbors, it's interpolated out. This improves the statistics a bit.
Vlad,
ReplyDeletein the prehistoric days when we used to use analog outputs, we thought of this quantity as "Conversion Gain"... as in how many uV of output signal we could get for every arriving electron. How does this relate to efficiency, I am clueless.
To us dinosaurs, charge conversion efficiency evokes a calculation of how many electrons are actually collected and converted to a voltage from a total number of electrons that are generated by incoming photons...
perhaps I will go hibernate now! :)
Actually, most people I know, including myself, still use "Conversion Gain" or "Conversion Factor" terms. The first time I saw "Charge Conversion Efficiency" was in DALSA blog. I wonder if anybody else uses this term.
ReplyDeleteI agree with the above posts. I use the term conversion gain, when referring to the conversion of electrons to uV at the FD node. I use the term conversion factor for converting electrons at the FD to LSB at the output. In Dalsa's terms, you could become more "efficient" simply by cranking up the analog (or digital) gain - which you could say that, but I don't agree with the terminology, because I think what you are really trying to get at is the capacitance of the FD node.
ReplyDelete-DP
Sorry for the question: can someone explain me how can CIS processes assure good maching (hence low PRNU) of very small FD capacitance? As an analog designer, I am used to matching problems even with larger capacitances (e.g. >50fF)...
ReplyDeleteCIS pixel array is an ideally repeating cell pattern. This greatly improves the matching, as compared with a generic analog circuits. One of the claims I've heard is that even transistor mismatches in the array are better than Pelgrom - hard to believe, but the claim came from a well-known very reputable person.
DeleteAnother contributing reason might be the bad pixel correction implemented on many commercial sensors. If some pixel deviates too much from its neighbors, it's interpolated out. This improves the statistics a bit.