What is wrong with my flats?

[Phil Leigh]

I was looking at the large spots on the right hand side... they haven't moved which rules out focus/rotation.
I still think that the solution is some combination of the flat being too bright and the bias frames being too noisy.

A master light created without bias frames would be helpful...

[pfile]

Hi.
I have no idea why the bias has more signal than your camera. Maybe it's just how this camera performs? Same with the pattern you described.

It's strange what you say about the flat that it doesn't match the light, the really bad specs are actually on the ccd and not the filters so they shouldn't move during filter changes. The camera wasn't moved or rotated but I did a very small adjustment to the focus one time but can that really move the dust motes that much? How do people do that refocus after filter changes for example?


If I calibrate just one light with the posted bias and flat I get the same overcorrected vignetting as the master shown in the first post.

/Daniel

well the entire camera vibrates a little bit when the filter wheel is moved, so in theory it could cause dust on the CCD to move a little bit as well.

focus is not super critical for vignetting. i don't have any experience with newts but given the optical distortions inherent in newts i wonder if focus could affect dust shadows at the corners of the image more than at the center. this is certainly something you could try to characterize.

anyway we have had recent threads here with bad flats where some attempt was made to modify the flat to fix it up. strictly speaking this is the wrong thing to do, and to be honest i don't think it has anything to do with pixinsight's calibration - the methods for calibrating a frame are very much fixed and every program is going to do pretty much the same thing. the only real "freedom" is whether or not and how much to scale a dark frame. of course the flat scaling factor has to be computed but i think this should be the same no matter the program. but the bias signal (and dark signal) in a flat can throw that scaling computation off, which is why i'm thinking about the bias frame so much.

i still think there is something funny about the bias frame, but i guess you'd have to ask other SX users (or SX themselves) about it. although my camera is also based on the KAF-8300M all of the surrounding electronics are going to be different, so it's hard to say if your camera is performing normally.

anyway, outside of the bias, phil's suggestion to experiment with flat exposure is certainly something you can try.

by any chance are you using a fast readout mode? for me fast readout is not supported by my imaging capture program so i never use it. takes about 1.5 or 2 sec to download a bin 1x1 frame over USB2.0.

rob

[DanielF]

OK - not sure about the Bias frame - personally I would discuss with the manufacturer and/or other users of that camera.

I tend to agree with Rob - if you subtract your bias frame from your flat (to get a calibrated flat)  the flat ends up with the Bias pattern noise superimposed on it...

I'm not seeing the same thing as Rob regarding the flats moving vs the lights... they seem to line up pretty well to me.

I would try running the script again without supplying any bias frames and see what happens.

This is the master light created with the PP-script but without a bias frame (and then AutoSTF)


As you can see overcorrection is still occurring.

I was looking at the large spots on the right hand side... they haven't moved which rules out focus/rotation.
I still think that the solution is some combination of the flat being too bright and the bias frames being too noisy.

A master light created without bias frames would be helpful...

Here's a master light without bias. But still dark calibrated.
https://dl.dropboxusercontent.com/u/8802413/M45%20problem/Flat_master_no_bias.fit

/Daniel

[DanielF]

well the entire camera vibrates a little bit when the filter wheel is moved, so in theory it could cause dust on the CCD to move a little bit as well.

focus is not super critical for vignetting. i don't have any experience with newts but given the optical distortions inherent in newts i wonder if focus could affect dust shadows at the corners of the image more than at the center. this is certainly something you could try to characterize.

anyway we have had recent threads here with bad flats where some attempt was made to modify the flat to fix it up. strictly speaking this is the wrong thing to do, and to be honest i don't think it has anything to do with pixinsight's calibration - the methods for calibrating a frame are very much fixed and every program is going to do pretty much the same thing. the only real "freedom" is whether or not and how much to scale a dark frame. of course the flat scaling factor has to be computed but i think this should be the same no matter the program. but the bias signal (and dark signal) in a flat can throw that scaling computation off, which is why i'm thinking about the bias frame so much.

i still think there is something funny about the bias frame, but i guess you'd have to ask other SX users (or SX themselves) about it. although my camera is also based on the KAF-8300M all of the surrounding electronics are going to be different, so it's hard to say if your camera is performing normally.

anyway, outside of the bias, phil's suggestion to experiment with flat exposure is certainly something you can try.

by any chance are you using a fast readout mode? for me fast readout is not supported by my imaging capture program so i never use it. takes about 1.5 or 2 sec to download a bin 1x1 frame over USB2.0.

rob

I don't think there's something wrong with the calibration in PI either. I'm gonna try to take new flats with a bit shorter exposure and see if that helps with the vignetting (the dust motes will be hard to align up perfectly again but I'll give it a try.)

I'll post a message on the SX Yahoo group and see if I can get an example of a bias from another identical camera to compare to mine and see if there are any differences.

I'm not using the fast readout mode.

/Daniel

[Phil Leigh]

I wouldn't worry about the dust motes...the 1st objective has to be to get flats that correct the vignetting. Once you have that cracked, the dust will vanish from your subsequent images (you may have to write this image off to experience, frustrating though it may be).

I'd still like to get to the bottom of the root cause of the serious vignetting you have. Are you using 1.25 inch filters?

[DanielF]

I wouldn't worry about the dust motes...the 1st objective has to be to get flats that correct the vignetting. Once you have that cracked, the dust will vanish from your subsequent images (you may have to write this image off to experience, frustrating though it may be).

I'd still like to get to the bottom of the root cause of the serious vignetting you have. Are you using 1.25 inch filters?

Hi. I agree, to get the flats to do its job is top priority.
I'm using 36mm filters, which are larger than 1.25 inch, but a bit smaller than 2 inch. (36mm is 1.42 inches)

/Daniel

[DanielF]

Ok, I took new flats with shorter exposure and this time I got this result after calibration
(single calibrated sub)


This is a new flat sub. Does it look better than the previous ones?
https://dl.dropboxusercontent.com/u/8802413/M45%20problem/Flat_L_New_001.fit

Still not good in other words.
Any ideas? Could it be my light-box that isn't providing an even illuminated surface?

/Daniel

[pfile]

that is certainly possible - can you try to make some sky flats just as a comparison?

[DanielF]

Yes, but that has to wait until tomorrow. It's dark here now in Sweden.

I'll post the result as soon as I can.

/Daniel

[pfile]

ok. along the lines of even illumination - what is the light source in the box?

fluorescent bulbs do flicker at the mains frequency, so you need to make sure that your flat exposure is long enough to capture multiple cycles. if it is an EL panel it will flicker at 10khz or 20khz or something like that. less of a problem for CCDs than for DSLRs which can make very short exposures with good illumination.

another pitfall is that some types of shutter give artifacts for short exposures. so if the light source is really bright, you may be getting artifacts from the shutter. my STT-8300M is supposed to have an "even illumination" shutter but flats < 1s do start showing some funny artifacts.

rob

[Zocky]

Daniel, maybe your problem is not in flats. I thing you have a problem inside your system that is causing uneven illumination. Recently I have similar problem. I have spacer ring which was not black, and was causing internal reflections.
Can something in your system cause such a thing?
Maybe your secondary mirror is not big enough, and due to that causes uneven illumination?

[Phil Leigh]

I've not seen anything like this...
Your original flat master vs the new flat you just uploaded - the vignetting has changed!

I can only think of two possibilities:

1) the light box illumination has changed
2) something in the optical train has changed

Your 36mm filters should be fine - are you using the StarlightXpress filter wheel?

The vignetting is not symmetrical which implies something is not aligned in the optical train (but we need to see a sky flat to rule out your light box)

[DanielF]

Hello again.
Quick reply, my little daughter needs to eat dinner soon.

This is a sky flat sub that I took.
https://dl.dropboxusercontent.com/u/8802413/M45%20problem/sky_25_001.fit

And this is a light sub calibrated with 10 sky flats (plus bias and dark).
(AutoSTF applied)


So I still have the same problem…
I'll get back to your other questions as soon as I can.

/Daniel

[Carlos Milovic]

To me both flats (light box and sky) match pretty well, so I would discard a problem with them. Also the bias seems to work well in them to remove the pedestal.

The problem seems to be with the lights. Do you have a dark over there? Maybe the dark current is high enough to create a new pedestal signal that is messing the normalization done by the bias.
I substracted the mean of the bias to the image, additionally to the first bias substraction and that worked pretty well. There is still a gradient, but seems to be light pollution or normal sky gradient.

Also there may be a inconsistency with a gain somewhere...

[DanielF]

To me both flats (light box and sky) match pretty well, so I would discard a problem with them. Also the bias seems to work well in them to remove the pedestal.

The problem seems to be with the lights. Do you have a dark over there? Maybe the dark current is high enough to create a new pedestal signal that is messing the normalization done by the bias.
I substracted the mean of the bias to the image, additionally to the first bias substraction and that worked pretty well. There is still a gradient, but seems to be light pollution or normal sky gradient.

Also there may be a inconsistency with a gain somewhere...

Here's the master dark that I used. (it's 20x5 min total)
https://dl.dropboxusercontent.com/u/8802413/M45%20problem/dark-BINNING_1-EXPTIME_300.fit

I don't fully understand what you did (or how) to get a better result, but that sounds good. There probably will be a fair amount of light pollution in the image.

/Daniel