QHY8L FLATS - Problems in BPP

[pscammp]

Hello all,
   Just given my 'second hand' QHY8L first light. Due to a rather annoying hardware issue I only managed to get x10 Lights each at 180 sec of the M81 / 82 pair.

Got x10 DARKS and x10 Bias calibration frames the next morning along with x10 FLATS but after running it all through BPP I ended up with an image with very light edges, seem this appears to be a common problem as I read across the Net but still cant seem to find a definitive answer to the problem.

I will attach a Gimp produced .jpg image of how my final integrated Light looks in Pixinsight with STF applied (unlinked) as the file seems to be a massive 210Mb in size (    What the hell's that all about  ) so I cant really do anything with it, Dropbox complained !

I done my FLATS in the usual way using Sequence Generator Pro's FLAT's wizard using an EL panel which gave me exposure lengths of way less than 1 Sec and they over brightened the edges of the frame. After reading and reading on the net some were saying with a QHY8L, FLATS need to have an exposure time of over 4 sec's due to the dual readout issue. So with multiple pieces of white paper I kept adding these until I got above the 4 sec mark - Still bright edges !

More reading told me that my ADU of 25000 may be way too much so then I took a library of FLATS which had ADU's ranging from 10000 to 30000 with x10 exposures in each folder. The FLAT exposure times ranged from 4.1 sec all the way up to 19.6 sec so I could try different sets to find the 'sweet spot' for my camera.

The Over-illumination seen in the attached pic looks identical using the 10000 ADU flats AND the 30000 ADU flats and everything in between - Im confused to hell, my Canon 1000D used to produce perfect FLAT correction every time in Pixinsight  ??

I have a master bias, dark and flat which was used to calibrate the x10 lights in the attached image in case anyone wants me to load them up for inspection, these are all 23Mb in size so if anyone wants me to load them up, just ask.

In spite of the over illumination problem and the fact that I was a little out of focus on the Lights, for only x10 Lights, im quite chuffed with the resulting image for a very inexperienced Astro photographer, a reasonable amount of detail in there I thought ?

One thing to note about the flats library, they were all done with the camera at zero degree's but the Lights were actually at approx. 99.1 degree's ref SGP's Framing & Mossaic wizard, not sure if this makes much difference but dust bunnies seemed to be in the same place on Lights and Flats.

So the questions:

1. Will the fact that I used Flats at zero degree's when light were at 99.1 cause what is seen in my image as the illumination is not only the corners  ?

2. There appears to be differing opinions on the net - Should I make sure Flats are exposing above 4 sec when using a QHY8L ?

3. With a library of Flats ranging from 1000 ADU all the way to 30000 ADU - How can the BPP result look identical in all of them ?

4. HELP !

Ow please, someone help me here, this is driving me to drink !

Many thanks for anyone who can help me get sorted.

Regards
Paul

 

[pfile]

the flats need to be calibrated with a matching master dark or if they are short enough, they can be calibrated with a master bias. i don't see mention of this in your description; if you don't calibrate the flats one way or another the scaling will be wrong.

flat minimum durations are very camera dependent and are usually tied to shutter mechanisms - if you have an iris type shutter, the exposure needs to be long enough or else the motion of the shutter causes some parts of the flat to be underexposed. i don't know about the QHY8L.

flat ADUs around 1/2 well depth should be fine.

i don't think your issue is the rotation but it could be... it really depends on the optical train.

rob

[pscammp]

pfile,
   Always there so you are   

QHY8L has no shutter so there is no issue there.

Not sure what BPP does internally as in "does it bias subtract the Flats, master dark etc"

What do you mean by 'matching master dark', is this a Master Dark at exactly the same exposure time as the Flats ?

What I did do is follow 2 different tutorials to do it all manually, this was the first followed to the letter:

http://www.pixinsight.com/tutorials/master-frames/index.html

Same illumination result !

Then I found this one:

http://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html#Section2

Followed this to the letter except cosmetic correction and the drizzle stuff and again, same illumination result !

Ref 1/2 well - Might learn something here - Would I be right in assuming Full Well would be 65,535 so 1/2 well would be around 32000'ish  ?

Optical train is a 10" newt, in the Moonlite focuser is the QHY8L with a 2" extension tube to a Skywatcher coma corrector which has an IDAS
P2 Light Pollution Filter screwed to the front. My Flats library was done with the focuser cranked out to the same position as when the Lights
were exposed but obviously the camera angle would have been different.

Regards
Paul

[pfile]

yes by matching i mean a master dark that matches the duration of the flats.

i think BPP should do the right thing, but i am not at all sure about that. i know that it tries to scale darks when it can't find a matching dark. in my case my flats are short so i'd rather just calibrate them with a master bias; my reluctance to go spend all the time to figure out exactly what BPP does in this case has prevented me from using it.

yes 1/2 well is around 32k ADU, if the camera has 16-bit output and a completely overexposed sub has 65535 in all pixels (you can test this yourself of course)

if you are sure that the flats were properly calibrated then i would start wondering about reflections in the optical train.

can you post a few bias subs, a few flat subs and some lights?

is there any chance the camera gain settings differ between the lights/bias/dark and flats?

rob

[pscammp]

pfile,

Thanks, will have to have a go at master Dark which matches the Flats & see if it makes a
difference instead of using the Bias / Superbias to calibrate the Flat.

No chance of my QHY8L gain or offset differing between lights and the calibration frames, SGP
always remembers the last used settings when connecting to the camera and I do check now and
again. I used the following tutorial to get the Gain to 10 and the Offset to 117. These figures give
me approx. 600-700 ADU at the bottom end of the histogram and just touching 65535 ADU at the
top.

The above ADU figures were derived by following a tutorial by Craig Stark as the instructions laid
down in the camera's manual were useless to say the least:

http://www.stark-labs.com/help/blog/files/GainAndOffset.php

Currently loading up x2 Lights, x2 Bias, x2 Darks and x2 30000 ADU Flats from the library I put
together the next day, I'll post a link when completed.

You will Notice in the file names that the Lights were done at -24c and the others were at -4c, all
frames were captured at -25c below ambient but when the Lights were captured outside ambient
temp was +1c and the next day when doing the calibration frames inside the house was much much
warmer at around +20c. I wonder if this temp offset makes a difference  ?? 

Reflection is a possibility I suppose, the QHY8L as standard comes with an IR blocking front window,
some, when using a Light Pollution Filter have suffered seriously bad reflection issues when a really
bright star is in the frame. My 8L now has a plain glass front which is said to solve this problem ref
the QHY forums. The IDAS P2 LP filter is a reflection type filter so it reflects non passed wavelengths
back up the tube so this could be a real possibility but wouldn't be sure now to confirm this !

Regards
Paul   




 

[pscammp]

Ok,

So here goes, these are random pairs of frames:-

Lights 1 & 2 (Debayering order will be - GBRG - for QHY8L):
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Light_180sec_-24.7C_0018.fit
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Light_180sec_-24.7C_0024.fit

Bias 1 & 2:
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Bias_0.01sec_-5.2C_0003.fit
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Bias_0.01sec_-5.2C_0009.fit

Darks 1 & 2:
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Dark_180sec_-4.1C_0002.fit
https://dl.dropboxusercontent.com/u/4938813/Subs/M81_Dark_180sec_-4.1C_0004.fit

30000 ADU Flats 1 & 2:
https://dl.dropboxusercontent.com/u/4938813/Subs/Flats_Test_Flat_19.57sec_-5.1C_0002.fit
https://dl.dropboxusercontent.com/u/4938813/Subs/Flats_Test_Flat_19.57sec_-5.2C_0007.fit

Hope you notice something significant Rob !

Regards
Paul

[oldwexi]

Hi Paul!
If you Stretch one of your lights and one of your flats there is an off Center Position
of your camera visible.
Checking the brightness distribution on the flat,  the center of brightness is in the lower middle of the
flat.
If you do the same with your light,  the brightness center is left of the middle.

It looks like you have turned the camera 90 degree  in comparison to the lights.
In addition with the off center postion of your camera this results exactly in a brighter part
in the flat where your dark area in the calibrated light is visible.
Make new flats in the same position as the lights have been shot.

In addition for the future centering of your camera in the optical path will deliver better results.

Here the screenshots
The top shows the offcenter flat
below the brightnessdistribution with rangemask (Center middle below)
below the brightnessdistribution of the light (Center left middle)
below the difference between these two brightness Facts, which results
          during calibration exactly to the darker areas where the flat is to bright.
below the calibrated Image with the dark areas because of different camera Rotation.



Gerald

[pscammp]

Gerald,

Many Thanks for taking a look

You are right, the Lights were taken at 99.1 degree's I think ref the plate solve in Sequence Generator Pro and
I will admit that when I put together my Flats test library these were all done at zero degree's. In my living room
there was no easy way to get the camera back to the 99.1 degree's I needed but then again, a 90 degree twist in
the right direction would have been better than nothing. Didn't really think it would have made THAT much differ-
ence but I trust your judgement way more than mine, it's a learning curve you know and I have to remember the
really important things like this !

Ref the off center -
Does this look to you like it could be sorted out with 'Focuser tilt' or does it look more serious than that, the reason
why I ask is because If I zoom in close on one of the Lights, on the right side near the edge there are comet looking
stars which would be an indication of coma but on the left they are not showing comets. The Moonlite focuser has a
grub screw in each corner for tilting. Or, failing that, maybe the secondary mirror isn't lined up correctly in the optical
path, or maybe collimation is out, or a bit of everything 

You guy's are the experts, what is the off center more likely to be caused by, where should I start to solve this

Many Thanks guy's
Paul

[oldwexi]

Hi Paul!
I have not so much hardware experience to answer your question
WHY the optical path to your camera is off center. My main experience is  Image processing.
And my C14/Hyperstar is perfect centered out of the box, so no need so far to
collect extensive experience in that area.
Anyhow reading your facts i "assume" it is  a bit of everything...

I am sure there are many here who can give better advice in hardware questions.

However, i think a simple DBE with carefully set points can get rid of most of the uneven background.

Gerald

[pscammp]

That's fine Gerald, many thanks for your help.

Newtonians are not so kind with stuff lining up though, hopefully someone can help based on the pics
you posted above although when I get a chance I'll start doing some experiments using flats to monitor
the position of the center of illumination. Short of further advice I'll post any positive results I get.

Regards
Paul

[pfile]

i guess if you have some focuser sag or other "non-stiffness" in the optical train then the rotator angle can really mess up your flats. another problem could be the angle of the OTA itself (with respect to gravity) differing between the lights and the flats. the problem with that is that of course the angle of the OTA changes all night, which could make all of your lights have slightly different illumination profiles...

rob

[gvanhau]

Hello Paul

I know QHY8L is a bit tricky because of the two exposures on images of less than 4 sec.

For me the best method has been using the manual calibration tools istead of the script.
What I do is:

Asuming  flat duration less than 4 secs. Light duration more than 4 Secs.

Masterbias=integration of bias
Calibrated flat= Flat-MasterBias
MasterFlat= integration of calibrated Flat
MasterDark=integration of Dark ; Dark should match temp and duration of light.
Calibrated light= (Ligt - MasterDark)/MasterFlat.
No auto dark scaling can be used.


If you find that your Calibrated Light is over corrected (light corners, darker center) then you may explore this
http://pixinsight.com/forum/index.php?topic=9312.msg59752#msg59752 work arroud I've found to compensate.

Hope this hepls.
Geert

[pscammp]

Sorry guy's, been away for a few days and no chance to get on the internet......

Well, when I get a chance I have lots to try out.

Ill start with redoing flats library at more or less the right angle of about 99.1 degree's & see how that turns out.

Then, I need to do some testing using flats to find out why my bright spot is not in the center of the images and
as a result my Lights also wont be centralised either. Don't think it's flexure but I can test for this by setting up
my scope inside, slewing to different positions and taking a Flat and see if the central glow moves positions.

Once here, and all else fails, I can try out this workflow Geert has suggested including what is shown in the link.

As usual, I will post images of any results I get as it's always useful in case anyone else stumbles onto this thread
with similar problems.

Many Thanks all for your help

Stand By.......

Paul

[pscammp]

gvanhau,

  I'll give your manual workflow a go just as soon as I get a clear night and the time to get fresh data.

One thing I need to point out though which I have noticed with current data:

On the night when I took the Lights the Camera's temp was about zero degree's when I was ready to image and I
took the camera's temp down 25 degree's to -25c.

The following day when I did the Darks etc, this was in the house where the camera's temp showed as approx. +20c.
I also took the Camera's temp down 25c to approx. -5c.

This means my Lights were taken with camera's sensor at -25c and the darks were taken at -5c so the temp between
them is approx. 20c.......Not Good !

The fact that the lights and darks (at same exposure time) is 20c apart, what effect will this have in the real world when
processing to produce my finished Master Light 

Regards
Paul

[pscammp]

oldwexi,

I believe I have solved the off center illumination issue by way of the tilt of the focuser, the Moonlite focuser has a
small grub screw on each corner to adjust tilt, a test Flat's bright spot is now central.

I also tried the scope at many different angles to see if the bright spot moves but it stayed put in the center and
collimation is also smack on at these different angles too.

Just need some clear sky's when I'm not at work now !

Cheers
Paul