A frustrating evening

[almcl]

Oh the frustration!

Last night was the first clear night for 10 days, so I set out to do some simple imaging and to try out my new-to-me-home-made-cardboard+sealingwax EL flats panel.

Weather was actually better than forecast so after taking nearly 90 minutes to get focussed and on target, I settled down to a run of 30 x 90 sec lights.  For a change, the guiding on the  AZEQ6 was reporting sub 0.5" oscillation (hooray!)  We got to the end of the first run with no major drama so I set about a further 30.  Halfway through this I discovered, while looking at the images in DSS, that somehow the focus had slipped and we had stars with holes in the middle.  Stopped the run, slewed to a nearby bright star and reapplied the Bahtinov.  Slewed back to target (no easy job as CduC and APT had suddenly decided that they wouldn't both use J2000, grrh!) and started the run again.  

At the end of what should have been the 90th light, the dew shield was removed and the flat panel was applied.  30 flats and 30 dark flats followed and then all was packed away.

This morning armed with only the last 30 (in focus) lights, 30 flats and 30 dark flats, it was time to post process.  Calamity, the first stretch of the stacked image revealed the most horrific vignetting:   

Tried stacking in ASTAP, results much the same - the flats are over correcting, but why?  Delving into the forums here, I saw a recommendation to use bias frames if this happens.  I had some and so tried a stack with flats, dark flats and bias and lo and behold the vignetting vanished!

So I now have a 45 minute image (below) but it could have been so much better with 3 times the data.  Holmberg IX would show up better and HIP48635 A + B might be resolvable (they are on the indidual subs), but when will the next opportunity be, I wonder? 

Rant over.

[ollypenrice]

I can't pretend to understand the over-correcting flats problem because we were once plagued by it using an SX camera/Nebulosity/MAC computer (none of them mine. I was just the man on the ground.) We never got to the bottom of it.

The flats procedure is different for CCD and CMOS. With CCD you can shoot flats then use a master bias as a dark-for-flats. It will work fine. With CMOS you must make dedicated darks-for-flats at the same exposure settings as your flats. As I understand it bias are not useful in CMOS imaging.

I don't think your final image tells us anything about the effectiveness of flats/no flats because it is black clipped, so the faint brightnesses which are corrected by flats are absent from the image in any case.

Above all, don't throw anything away. I'm sure you have a good image in there somewhere. One work-around I tried with our over-correcting rig was to make a stack with flats and a stack without. I would then average the two of them to get a compromise before working on the gradients in software.

You have my sympathy: the over-correcting flat is something which has still defied every explanation I've read so far. One area to look into, though, is the flushing of images between captures. I don't remember the details but it seemed a promising line of investigation when we were fighting this problem.

Olly

[almcl]

Thank you, Olly.

Here's a stack (from ASTAP which doesn't use Bias/Offset) where bias files were used instead of the Dark Flats  (hopefully not quite so black clipped - min pixel value is ~4):

 

Whatever the cause, it seems that both DSS and ASTAP need the Offset/Bias files on this occasion, which goes against much of what I have read for the ASI 2600 MC.  Downloading files from the 2600 has proved 'interesting' and generated a fair amount of activity both on the ASI site and the APT forum.  

Edited February 19, 2021 by almcl