Meaning of gain parameters in desi.yaml

[dkirkby]

The gain parameters in desi.yaml are all given as:

gain: 1.0 # electrons per ADU

Since the value is 1.0, this question is mostly academic but I would like to be sure I am using this constant correctly in specsim in case someone ever runs with a different value.

The units indicate that I should divide the simulated electrons per bin by the gain. Is that correct?

[moustakas]

Right.

[dkirkby]

Ok, thanks. I normally think of an amplifier gain as (output signal) / (input signal), but I'm glad the units are correct at least.

[djschlegel]

I've always advocated re-naming the quantity that astronomers use as the "loss".
At least then the meaning is no longer ambiguous.

[sbailey]

The amplifier gain is (output signal) / (input signal) since in real life it is used to convert raw ADU (amplifier digital units) coming out of the CCDs into electron units, upon which we do the rest of the calculations. i.e.

electrons = (raw data CCD ADUs) * gain

It only seems backwards because our simulations are in electrons (photons) and then we divide by the gain to get back to really really raw ADU data. specter jumps through some hoops to ensure that we get the integerization right at the photon (electron) shot noise, and then it is converted into ADU and re-integerized like the CCD electronics do, and then re-multipled by the gain to get the final "electrons" that one would work with for extractions. I welcome review of that logic, however.

[dkirkby]

Ok, but the input signal to the front-end hardware amplifier is electrons (from e-h pairs) collected on a CCD gate, not ADUs.