Details
-
Type:
Story
-
Status: To Do
-
Resolution: Unresolved
-
Fix Version/s: None
-
Component/s: Calibration Products Production, cp_pipe, ip_isr
-
Labels:None
-
Story Points:5
-
Epic Link:
-
Team:Data Release Production
-
Urgent?:No
Description
Aside from the banding (see DM-33752) there is also lots of mysterious (to me, at least) unstable vertical bias structure in LATISS, or at least there seems to be. This ticket is to take some time and dig into the data a bit and see if the things we're seeing are real, look at failings in the ISR, and improve ISR on LATISS in general.
See, for example, some images taken back-to-back: dayObs = 20220405, seqNums = 539-586. Of these, in 551 and, to a lesser extent, 572, there are bright vertical lines starting at the bottom of the chip going up the columns. I don't think they're on amp boundaries, i.e. aren't conventional divisadero tearing, though they look very similar. It is also notable that the one in the amp 2nd from the bottom left is, I think, perfectly in line with the big columnar defect in the top left part of the chip. I wonder if this instability is why we correct that bad column so poorly, and if this is all connected. I also wonder/hope whether there is some hardware mitigation we can take for this stuff.
Nigh movie here: https://lsst.ncsa.illinois.edu/~mfl/observing_runs/2022-04/20220405.mp4
Attachments
Issue Links
Activity
I've seen similar features in dark frames before, and have assumed that the defect column in C11 (top row, second from left) is creating crosstalk features in all of the other amplifiers (explaining the period in columns and the reflection from top to bottom). I haven't investigated much more than that.
I think that's quite possible, yes. Maybe the charge injection in that column fluctates quite a bit, and that causes the streaks in others to come and go. Though that said, we don't have crosstalk correction on, so if that were the dominant cause, I'd expect them to be much more present in general, and more there than not, so I don't think that can be causing these particular fluctuations.
Anyway, it's both a good point, and definitely something to be aware of when looking into this, so thank you!
Note to future-us: having dug into some other transient things I've been (bright defects appearing out of nowhere) I realised that it was happening after long periods without taking an image, which actually makes a lot of sense now that I think about it. I wonder if this is responsible for a lot of this stuff. For example, see the bright defect which appears in the top middle of the chip 2023-02-14 seqnum 849 which wasn't there before, where that image was taken after a 49 min gap in observing, or an earlier instance in that same night in image 321 which was the first on-sky image, so hours after the last readout.
See this comment on DM-33752 about a presumed intermittent bright column in the middle of the chip too.