Calibration - some frames do; some frames don't

[kor]

Hi all,

I could use some help.  I'm having some problems getting my frames to calibrate.  17 of 35 frames finished calibration with some bias signal still in them.  The 35 frames were taken over a few months.  The 17 that did not calibrate correctly were taken all on the same night.  Other frames were also taken on that night and calibrated OK.

I don't see any issues in Process Console; no warnings or failures, reasonable dark scaling factors, etc.  But when I look at the calibrated frames the left edge clearly still has some glow.

Here is a frame that did not calibrate:
Raw: <http://www.kor-astro.net/calibration_issue/Light_G_0030.fit>
Calibrated: <http://www.kor-astro.net/calibration_issue/C_Panel-1_Light_G_0030.fit>

Here is one that calibrated OK:
Raw: <http://www.kor-astro.net/calibration_issue/Light_G_0031.fit>
Calibrated: <http://www.kor-astro.net/calibration_issue/C_Panel-1_Light_G_0031.fit>

Here are the master bias, dark, and flat.
<http://www.kor-astro.net/calibration_issue/Master-BIAS_-20C.fit>
<http://www.kor-astro.net/calibration_issue/Master-DARK_-20C.fit>
<http://www.kor-astro.net/calibration_issue/Master-Flat-G.fit>

The light frames are 120 secs. 
The Master Dark is 50x600 second subs. 
The BIAS is 200 subs
The flat is 25 subs of about 1.5 seconds.

The camera is QSI683wsg with filter wheel.
The frames that did not calibrate are green.

I know there are some calibration issues with some of the QSI683 cameras--I don't think mine matches the banding that they were seeing, but I will also send QSI a question.  So, can anyone give me some pointers?  Does it look like a camera issue or do I just not know as much about PixInsight as I thought I did?

Thanks - Shane

[mschuster]

Shane,

The residual in the light could be either amp glow or light pollution gradient or a mix of both. Although I agree it does seem to look like residual amp glow.

As a test, you might want to expose 10 new biases and create a new master bias. No need for a lot of frames. Do a calibration with only the new master bias (disable master dark and master flat). Does the problem remain? If not, then I would say the bias in the light drifted and your old masters were not able to remove it. If it is still there, then maybe it was light pollution in the first place or bias drifted yet again.

Thanks,
Mike

[kor]

Thanks Mike,  I'll give that a shot.  The master bias and dark frames I used are new; I made them last week after I noticed the calibration problem (at that time I was using an older master bias and master dark done at -25 instead of the -20 of the light frames--I thought that might be the problem ...)  I'll do another new one and try a bias correction only.  It's going to be cloudy tonight anyway

- Shane

[kor]

OK, I tried bias-only calibration with the new 10 frame BIAS and got the same results. Anyone have any more ideas?

Here are my process icons building the master bias, master dark and master flat, and calibrating the green frames:
<http://www.kor-astro.net/calibration_issue/Calibration_Subset.xpsm>

- Shane

[mschuster]

Hi Shane,

Please post a link to the new master bias. I want to run PixelMath "new_master_bias - old_master_bias + 0.5" to see how much the bias level has changed, if any, and also to see if any difference is the same left/right across the frame or not.

Over a year+ period, my QSI683 dark bias level drifted just a little bit, about 15 DN. The drift was the same across the frame. Such drifts are to be expected for the electronics we are using.

You also might want to run PixelMath "light - new_master_bias". This is what ImageCalibration does when only a master bias is selected. If the problem is still there (most likely), then the issue is with your light and not the calibration process.

Thanks,
Mike

[kor]

Hi Mike, and thanks again for your input.

Here is the master bias (10 frames) that I made this morning:
<http://www.kor-astro.net/calibration_issue/Master-Bias-New.fit>

For a timeline:
Light frames that did not calibrate where shot on Sep 16
Light frames that did calibrate where shot July 11 through Sep 19, including many on Sep 16
The "old" bias and dark frames where taken Sep 27 and used on all frames taken between Sep 15 and Sep 19.  Older frames used a different set of bias and darks shot at -25C.
The flats where taken Sep 15.

I don't know if any of that is useful.

I ran that same pixel math operation this morning after I tried the standard calibration.  The glow IS present using the PM.  So maybe the camera just lost its mind between 11:15 and 12:55 on the night of Sep 16 ...

- Shane
And the "new" bias frames where taken Oct 4

[mschuster]

Thanks Shane,

The masters are nearly the same, a bias level change of about 2 DN (within spec, not an issue for this camera) and no left to right departure (this is good).

My conclusion: your light did in fact calibrate. What looks like residual glow from an assumed incorrect calibration is very small (~10-15 DN). IMO this is a light pollution gradient in your light and not a calibration issue. If you have not done so, check out the DynamicBackgroundExtraction process to remove these types of gradients post calibration. I think these types of gradients and worse are to be expected, and as long as you do not have a lot of nebula across the whole frame that would confuse DBE, this process is the standard way of removing them.

Thanks,
Mike

[kor]

Hmmm.

Well, I knew I could probably fix it with DBE.  In this case, I could probably just crop it off, since it is part of a mosaic and I have that portion in the frame next to it.  I was just trying to figure out what the real problem was so I could make sure it didn't happen again. 

Thanks for the help, Mike.  I'm not sure I'm going to finish this image this year--I lost most of June and July to the excellent monsoon season we got this year and am running out of sky!  If I do have to finish it next year, I'll probably just omit these frames and get some new ones.  If I can get enough time in to finish it in the next month or so, I'll see if I can salvage these.

Thanks again - Shane