Question about correct exposure for flats (overcorrecting flats)

23 Mar, 2018 06:40
Hi everyone,

I recently switched from a DSLR to a QHY163m + Baader LRGB filters. My first steps went Ok, I think - but I focused on very bright and diffuse objects (M42, Rosetta nebula).

Switching to objects with a lot of sky background, I noticed that my flats overcorrect across all the channels. I understand that this is most likely caused by underexposing them. So I am now trying to understand how to fix this (don't get a lot of clear nights in these parts, so would like to be prepared for the next one, when ever that may be).

My technique so far was to use the built-in function in APT to automatically determine the correct expose for each channel. APT suggests a target ADU of 20.000. This comes out at ~18.500 when checking in PI.

Two questions:

1) So if 20.000 is too little, is there any robust way, other than trial and error, to figure out what the correct ADU should be?

2) Using Pixelmath in PI with the expression "$T+0.2"  I was able to "boost" the L flat and "rescue" my luminance image. However, a similar approach (using a range of scaling factors from 0.05 to 0.5) for the RGB channels did absolutely nothing. How can that be?

Thanks for any and all input!


Example of the L channel prior to "flat boosting":
and the flat:
Edited 23 Mar, 2018 07:16
23 Mar, 2018 07:08
20,000 or anywhere near it should be fine.  That's around where CCDAutopilot suggests.  I've always let Autopilot acquire my sky  flats and have only had a problem when my scope has dewed during flat capture.  So all I can add is that I don't think exposure is the issue.

23 Mar, 2018 08:29
I would be interested as well. My asi has the same chip as the QHY163M and I too have problems with uneven stacking results that clearly appear on the corners. I tried 14k adu
23 Mar, 2018 08:33
I did some more digging; someone mentioned over on cloudynights that the readout mode for the QHY163m (and presumably ASI1600) differs between long and short exposure. My flats were taken at 0.5sec exposures, possibly that qualifies as "short". Anyone heard anything like that before? I suppose I could dim my flatfield box all the way down and take 2-3sec exposures instead. Uhm… It is weird though.
23 Mar, 2018 09:39
I have the exact same problem with the same QHY163m camera and same LRGB filters! It is driving me nuts. I was wondering if the fact I am calibrating very shot exposure light subs (60 sec) is also a contributing factor so I plan to image for longer on LRGB (targetting 2000 ADU in SGPro) and see if that helps any.
23 Mar, 2018 09:54
Oddly enough it works fine at 60secs for me, and produces the error at 120secs =) Seems to have something to do with signal intensity in the light exposures, maybe. I just don't know anymore.
Edited 23 Mar, 2018 09:55
23 Mar, 2018 09:56
I have read that the absolute minimum exposure for the QHY163 is 0.3 secs ( so I am going to use a buck converter to reduce my lightbox right down and shoot for longer flats as you suggested above. I have no issues with narrowband flats which are normally 2-4 secs on length so that seems to be a variable to target next
23 Mar, 2018 15:22
CCDAupilot guidelines state the flats should never be shorter than 2-3 seconds.
24 Mar, 2018 07:59
determining the right exposure for a flat should be done by a PTC (photon transfert curve), determining the linearity breakout point and exposing just few counts before. Generating a PTC is not that complicate… You can find many articles on the internet. Anyway, while is important to know a target adu value, is also fundamental to use the longest possibile exposure time to reach that target. This because of the shutter. If the exposure is too short, than you can have nonlinearity because of the shutter that, while closing, leave some areas of the sensor exposed longer than others. Your sensor have a typical full well of about 18-20K so a value about 12/15K should fit. To be precise the only way to determine it is a PTC.
Edited 24 Mar, 2018 08:07
24 Mar, 2018 13:00
These cooled CMOS cameras don't have mechanical shutters! I stopped taking flats with my QHY163m as my field illumination is great and I try to keep my filters clean. Didn't find flats helped, and any gradients are totally fixed by APP. If you do take flats, they should be matched by dark flats instead of bias frames, and as mentioned, stay above 0.3s (ideally 1s or more).
25 Mar, 2018 17:09
CMOS sensors have direct A/D conversion on pixel basis. In the conversion process they've the map of errors embedded inside the electronics. Anyway, if you want to do measures for scientific purposes, a flat field must be acquired for charaterizing the whole system. I don't know about this specific camera model but there are some CMOS cameras that have mechanical shutters. Obiuvsly, mine were just considerations on how to determine the correct values for generating a good flat field.
Register or login to create to post a reply.