search

desihub / desispec

DESI spectral pipeline
BSD 3-Clause "New" or "Revised" License
35 stars 24 forks source link

nightqa support 2amp mode #2184

Closed sbailey closed 6 months ago

sbailey commented 6 months ago

This PR updates the amp bookkeeping in nightqa to support 2amp mode, in particular tested on z3 data on 20240221:

https://data.desi.lbl.gov/desi/spectro/redux/daily/nightqa/20240221/nightqa-20240221.html

Note that the amps are labeled correctly on the dark image: image

And the previously bad CTE offsets (from 20240220): image

Now don't have a step across the middle-of-the-CCD up/down boundary, though there is still an uncalibrated offset.image

Admittedly, those above/below labels still reference amps C and D, basically using those labels for quadrants even though the data weren't really read out through C and D. Does anyone care enough to suggest alternate labels?

I'd like to merge this today so that it is ready for nightqa tomorrow. @schlafly @araichoor @julienguy

UPDATE: Julien explained it to me: amp C (left) has an even worse CTE problem than amp B (right), thus the choice to use AB instead of CD. But the type of problem that C has is calibrated out in this plot in a way that the problem in B is not. However, when doing skysubtraction the extra offset in B is fit out so we should be ok... Regardless, that doesn't impact this PR for fixing nightqa in two amp mode (prior to this PR, it crashes).

araichoor commented 6 months ago

@sbailey : apologies, I ve missed the pinging yesterday.

I ll try to have a look at the label thing next week (I also need to make an update for the ctedet plot, for the code to look first for the "ctepreproc" file; https://desisurvey.slack.com/archives/C01HNN87Y7J/p1707930809499039).