Some Flat Masters Work, Others Not So Much

[sreilly]

I have pondered this for a long time and have tried different things without any real resolve. My imaging train has remained the same for years and I have been taking panel flats for many years as well. First I should list my system:

I use an Optical Guidance Systems 12.5" RC. From the rear cell I have:

1) the flattener attached
2) MMOAG Off Axis Guider w/SBIG RGH for my STL-11000
3) Optec 3" Pyxis Rotator
4) SBIG CW8
5) SBIG STL-11000M

I image mostly LRGB and those exposures are typically -20 degree at 900 seconds each. The narrow band images are -20 degrees and 1800 second exposures. I take a library of dark and bias frames at -20 and 900/1800 darks. The masters are generally made using anywhere from 20-50 exposures. Master Bias frames are typically from 75-100 bias frames and produced by these guidelines http://www.pixinsight.com/tutorials/master-frames/index.html

When I run into a set of images that seem to have issues flat fielding I will generally try to make a new set of flats and try again but here lately that seems futile. For some reason the flats just don't seem to work. Now to cover my bases I have tried using both east and west masters to see if it makes a difference in the calibrated image but overall not much and neither satisfactory. I had seen in my early days an issue with the software not knowing East side from West side depending on how/where the telescope was parked. When I parked the telescope with the OTA level facing North and the counterweight shaft end facing east (AstroPhysics mount park position 1) it seemed what was my west masters worked for the east images and vice versa.

These days I park the mount with the OTA facing south and the counterweight shaft end facing west (AstroPhysics mount park position 4)

If I have a problem image my process is that I create a plan and have the telescope slew to the object in question at the desired PA and take a short exposure, download, plate solve and then any corrections for PA or object orientation is corrected. Then I'll proceed with new flats now that I'm 100% sure the system is at the desired position. This has yet to solve any of the issues. The flats are minimum of 1 second exposures, 16 each side (E/W), calibrated and combined using the above mentioned process. 99% of the time this works great but that last 1% I'll have a series of images that have poor results. The exposure times are calculated to give a 25,000-28,000 ADU level.

Results for this current series of images is typical of attached. Images taken in 2018 came out without these issues. Not sure what I'm missing.....

[pfile]

when you calibrate the flats, what masters do you use?

rob

[ngc1535]

Hi Steve,

I would like to add something you may not be considering. Your flats may not always completely characterize the color of the sky. Light pollution (especially), moonlight and airglow are the biggest offenders. You might find (you didn't mention) that the images you might have problems with are objects at high air mass (or with a predominance of data taken at high airmass). This can make the effect I suggest even more significant. I do not find many discussions on this... but I suspect it is more important than people might consider. DBE of course can help when the flat errors are smooth and "regular" (symmetric).

So even if you do everything "right"... there can still be issues beyond your control.
-adam

P.S. Flatteners have a way of making complex illumination profiles.

[sreilly]

Thanks for the reply Adam. Interestingly enough the vignetting I expect but the doughnuts that don't get corrected are the biggest issues, hence the plate solve to verify the PA. That's what really stumps me the most. Same imaging train and flats taken the next day so no major changes there.

[sreilly]

when you calibrate the flats, what masters do you use?

rob

Rob I use the master Bias frame that was taken at the same temperature. Seems the STL-11000 did best with bias taken at the same temperature. The master dark is the 900 second and optimize is checked. See attached....

[sreilly]

This may help as well.

[ngc1535]

Thanks for the reply Adam. Interestingly enough the vignetting I expect but the doughnuts that don't get corrected are the biggest issues, hence the plate solve to verify the PA. That's what really stumps me the most. Same imaging train and flats taken the next day so no major changes there.

Yes- my suggestion would make dust motes not calibrate out as well. Professional observatories that deal with light pollution often take night sky flats to compensate for this effect and truly have properly (spectroscopically) illuminated detectors. You can prove me wrong (or correct) by taking night sky flats and seeing if the resulting flat correction is better. You will need to look this up...but basically taking many many images from the same night (time) and rejecting all of the stars (median should be OK) to produce the image. The hard part is getting enough sky counts for a good flat.

-adam

[sreilly]

Yes, I was thinking about the fact I failed to mention the method I get my flats which is by means of an Optec Flatman XL Panel. I've been fooling with using only darks or bias for calibration to see what difference this might have. I also have to wonder about how closely the filters are being repeatedly positioned. So many variables that it boogles my mind it works at all sometimes. I'll look into these night flats, never heard of them before. Just hope it isn't anything like the left handed smoke shifter we had the new kids looking for in scouts. One kid thought he really had one till I told him what he had was a left handed smoke shifter which doesn't work nearly as well. Oh I was a real s---- at times back then...........

[pfile]

if the flats are very short it is worth a try to just use the master bias only.

i have been meaning to try fabian neyer's method for correction of regular flats with night flats. my own systems suffer from what adam is describing - the flattener/reducer i use almost certainly leaves weird large scale round artifacts in the lights that are not corrected by my panel flats. but individual calibrated subs look OK to the eye; it's just when everything is stacked and the SNR is high that you can see these artifacts in the background.

rob

[sreilly]

After much digging I think I've found the culprit and it isn't the flat. Seems my master dark is the root of the problem. I take flats several times a year and create a master. This one is from just 3 months ago and consists of 58 frames. What I see is a light gradient that causes vignetting and punctuates the doughnuts in the flat. I tried an older dark master and it was much better so now I've taken and created a master flat of 10 fresh darks to test. So far the difference is the vignetting is mostly gone (normal) and the doughnuts are missing as expected. I'll post examples shortly.

[sreilly]

So as promised here's a screen shot of the new dark calibrated image vs the old dark calibrated image. Only change is the master dark.

[pfile]

yeah i'd say 9 times out of 10 overcorrection by flats is caused by bad calibration of the flat subs, so it's good to hear that new darks saved the day.

rob