Flats not right-but why

[Rodd]

I am getting so tired of flats not working.  Yes, dust specks can move between the light frames and the flat frames invalidating the flats....but every time?  This is getting tiresome.  I have tried flats with the histogram about1/4, 1'3 and 1/2 and none work--there is ALWAYS dust motes or partial dust motes that remain.  Why?  I use 50-100 flats ranging from .5 sec to 2 sec.   So far the .5 sec ones work the best.  But others say its best to use 2-10 sec.  I use flat darks of the same temp and duration as my flats.  What is the secret?

I am not referring to false IFN cloud like blemishes.  I think they are from hazy sky (high clouds, poor transparency.  Its the dust specs that are infuriating.  How can they move EVERY time?  I shoot flats immediately after collecting lights.  I don't know what else to do

This is NGC 5905 and 5908--only the red and green channels.  It is an RGG image.  The FWHM of the green stack is 1.6.  I used all 161 subs.  A very consistent night.  Guiding errors remained .39-.42"  But the green flats did not work.  I had to do a lot of work to clean this image up.  I made it to see if I could salvage the data--or if I had to throw it out.  Imagine throwing out 5.3 hours of subs with FWHM of 1.8.  Its rare that seeing for me allows this.  The red stack has a FWHM of 2.6--and I was lucky to achieve this...it was a blustery night and I threw out 50% of the data.  Though far from a good background, based on this image, I think I will keep the green and proceed with blue then Lum.  But I am  really getting fed up with flats.  I was lucky that none of the dust motes were in the galaxies.

I use a Flatman panel.  Histogram levels are 1/4-1/2.  I use flat darks of same temp and duration.  What could it be?

C11 Edge with .7x reducer and ASI 1600 red: 158 120 sec; Green: 161 120 sec.  About 11 hours

 

 

 

Here's with a bit more red.  Cant decode which is better

Edited May 19, 2023 by Rodd

[andrew s]

Rod , I feel your frustration!  For flats to work the flat light path needs to match the light path from the sky as closely as possible.  I suspect you are getting additional light from the Flatman panel either from reflections or flooding in the C11.

Not sure where you have the panel in relation to the corrector but the further away the better to match the sky.

Regards Andrew 

Edited May 19, 2023 by andrew s

[bottletopburly]

Are you getting light leaks into your calibration frames ,i was and thankfully anther sgl member spotted the error and seems to have sorted once i shot new darks flats and  flats,Darks  ,i was using a newtonian though .

[Rodd]

17 minutes ago, andrew s said:

Rod , I feel your frustration!  For flats to work the flat light path needs to match the light path from the sky as closely as possible.  I suspect you are getting additional light from the Flatman panel either from reflections or flooding in the C11.

Not sure where you have the panel in relation to the corrector but the further away the better to match the sky.

Regards Andrew 

Not really possible to put it far away.  It is designed to sit on top of the tube.  I don't have a an observatory wall to hand it on.  Besides, then I would need a panel 3x the size.

[Rodd]

13 minutes ago, bottletopburly said:

Are you getting light leaks into your calibration frames ,i was and thankfully anther sgl member spotted the error and seems to have sorted once i shot new darks flats and  flats,Darks  ,i was using a newtonian though .

I don't think so.  The panel forms a tight seal with the tube.  Could be internal reflection-but it doesn t seem to be that--dust particles are not being canceled out--but there is no streaking or flares that would suggest a light leak

[andrew s]

1 hour ago, Rodd said:

Not really possible to put it far away.  It is designed to sit on top of the tube.  I don't have a an observatory wall to hand it on.  Besides, then I would need a panel 3x the size.

Then I don't think you will solve your problem.  Why need it be 3 times the size? The angular size of the c11 field is quite small.

Try some twilight flats or just experiment.  If you won't change anything how do you expect to solve the problem?

The high focal ratio of the c11 makes it sensitive to changes in light path.

Regards Andrew 

[ONIKKINEN]

You could post a flat, a darkflat, a dark, and a lightframe in raw .fits format to be analyzed by others. I have seen at least half a dozen cases where light leaks were to blame but its not necessarily always the issue. With newtonians its the clear usual culprit but an SCT is more closed so maybe not this time.

You could have mirror flop or focuser sag, or really anything at all in the light path moving between lights and flats and the result is bad calibration. This is testable though.

Slew the scope to extremes on each side and take a flat. I mean one at the zenith, one at maximum east and north, one at south, one at west etc. Point is to get the scope in as many orientations as possible.

Then divide one of these flats with another simply with pixelmath in PI (Flat1/Flat2) and see what you get. Thr resulting image from dividing any flat with any other flat from this test should be a featureless gray mess with no brightness difference to be seen even with stretching. If you get a gradient, you have mechanical issues and flats will simply not work until they are solved.

 

 

[Rodd]

55 minutes ago, andrew s said:

Then I don't think you will solve your problem.  Why need it be 3 times the size? The angular size of the c11 field is quite small.

Try some twilight flats or just experiment.  If you won't change anything how do you expect to solve the problem?

The high focal ratio of the c11 makes it sensitive to changes in light path.

Regards Andrew 

I tried sky flats. Terrible. They work well with fsq but not c11.  Maybe I need a diffuser 

 

[Rodd]

58 minutes ago, andrew s said:

Then I don't think you will solve your problem.  Why need it be 3 times the size? The angular size of the c11 field is quite small.

Try some twilight flats or just experiment.  If you won't change anything how do you expect to solve the problem?

The high focal ratio of the c11 makes it sensitive to changes in light path.

Regards Andrew 

The panel must be perpendicular to the axis. It has to be flat. The only way that I can do this is to put it on top with the scope pointing at zenith.  I just don’t see how I can do it otherwise unless I build a wall and hang it. 

[Rodd]

24 minutes ago, ONIKKINEN said:

You could post a flat, a darkflat, a dark, and a lightframe in raw .fits format to be analyzed by others. I have seen at least half a dozen cases where light leaks were to blame but its not necessarily always the issue. With newtonians its the clear usual culprit but an SCT is more closed so maybe not this time.

You could have mirror flop or focuser sag, or really anything at all in the light path moving between lights and flats and the result is bad calibration. This is testable though.

Slew the scope to extremes on each side and take a flat. I mean one at the zenith, one at maximum east and north, one at south, one at west etc. Point is to get the scope in as many orientations as possible.

Then divide one of these flats with another simply with pixelmath in PI (Flat1/Flat2) and see what you get. Thr resulting image from dividing any flat with any other flat from this test should be a featureless gray mess with no brightness difference to be seen even with stretching. If you get a gradient, you have mechanical issues and flats will simply not work until they are solved.

 

 

It’s probably mirror flop. Even with the mirror locked down with the mirror lock knobs and focusing with crayfish at back of scope I see movement of the focus star when I focus. I thought the mirror locks were suppose to lock down the mirror. How to fix this?

[Elp]

I point mine up too and put the panel on top of the dew shield. Its on lowest setting and I use two sheets of opal acrylic, one light grey and one dark grey all stacked up. RGB still takes flats less than a second though (F2), narrowband takes a few seconds. On my refractor the time is a few seconds sometimes up to 10 in narrowband. The exposure is calculated automatically via my asiair and the histogram is usually just under halfway to halfway at 3/4 peak intensity.

[ONIKKINEN]

35 minutes ago, Rodd said:

It’s probably mirror flop. Even with the mirror locked down with the mirror lock knobs and focusing with crayfish at back of scope I see movement of the focus star when I focus. I thought the mirror locks were suppose to lock down the mirror. How to fix this?

Not an SCT owner myself, so wouldnt know. Although i have seen others with the issue install an extra focuser on theirs and permanently lock the mirror so that might be an option.* I see you already have one, didnt read the comment properly..

But before spending who knows how much on who knows what you should test the stability of yours with flats in different scope orientations first.

Edited May 19, 2023 by ONIKKINEN
*

[Rodd]

Just now, ONIKKINEN said:

Not an SCT owner myself, so wouldnt know. Although i have seen others with the issue install an extra focuser on theirs and permanently lock the mirror so that might be an option.

But before spending who knows how much on who knows what you should test the stability of yours with flats in different scope orientations first.

Well, I already know there is mirror flop. I see it when I focus. The focus star shifts.  The mirror is locked down using 3 mirror lock knobs. I added a focuser to the back. I focus using a crayford shorty and the mirror remains locked.  Not sure why there is flop. But it’s definite. Not sure what the fix is, seeing that I already use the fix. But the fix doesn’t work

[andrew s]

13 hours ago, Rodd said:

Well, I already know there is mirror flop. I see it when I focus. The focus star shifts.  The mirror is locked down using 3 mirror lock knobs. I added a focuser to the back. I focus using a crayford shorty and the mirror remains locked.  Not sure why there is flop. But it’s definite. Not sure what the fix is, seeing that I already use the fix. But the fix doesn’t work

Why should the focus star shift if you are using a add-on focuser to focus? It is more likely to be play in the focuser or camera side of it than the mirror moving especially as it's locked down. 

Regards Andrew 

[The Lazy Astronomer]

15 hours ago, Rodd said:

focusing with crayfish

Well there's your issue. The segmented body of the crustacean inherently  introduces the potential for tilt into the system. 😁

On a less stupid note, how good are the mirror locks really? I would have thought it would only take a tiny movement to cause an issue, so I think conducting the test suggested by onikkinen would be a very worthwhile exercise.

[tomato]

Forgive the crude rendition, but if the flat panel is not too heavy, could you hold it place with a wooden baton and some tie wraps around the dovetail plate and finderscope shoe?

[Elp]

I'm guessing mirror flop is more of an issue the larger the scope you go, I don't really get it with my C6 even after meridian flip. I also don't use an external focuser, don't feel the need to.

Edited May 20, 2023 by Elp

[Rodd]

1 hour ago, The Lazy Astronomer said:

Well there's your issue. The segmented body of the crustacean inherently  introduces the potential for tilt into the system. 😁

On a less stupid note, how good are the mirror locks really? I would have thought it would only take a tiny movement to cause an issue, so I think conducting the test suggested by onikkinen would be a very worthwhile exercise.

I asked John Hayes about that-he is reknown for his C14 system and I thought he might have an idea.  He said "the Celestron mirror locks are not perfect, but they are giod enough to provide good calibration".  Based on conversations with him, and here, I suspect its the focuser, not the mirror flop.  Maybe its loose.  There are a number of set screws and thumb screws.  I wil tighten it up and see if that helps....over a langastino lunch

[Rodd]

3 hours ago, andrew s said:

Why should the focus star shift if you are using a add-on focuser to focus? It is more likely to be play in the focuser or camera side of it than the mirror moving especially as it's locked down. 

Regards Andrew 

That is what I am coming to believe as well.  My question is if there is shift in the focuser so that dust motes move--why not stars.  One might say that the registration process will align all the stars--which is true, but there would be registration artifactsv at the edges of the stack.  I get those becuase I manually align the FOV after focusing.  But If I shoot 100 subs and do not refocus (it happened the other night.  FWHM stayed between 1.6 and 2.1 for hours....in my sky, that is as good as it gets), there is no stacking artifacts.   So I am a bit unclear how a shifting focuser can impact calibration but not registration.

[Rodd]

1 hour ago, tomato said:

Forgive the crude rendition, but if the flat panel is not too heavy, could you hold it place with a wooden baton and some tie wraps around the dovetail plate and finderscope shoe?

Crude? You nailed teh colors!  It looks.....almost blike a photograph!  It would be possible to do as you suggest, though it is on the heavy side.  At this point, it would be verifying something we already know--dust motes move.  I think the first place to look, after everything everyone has said, is the focuser.  The fact that it doesnt happen all the time could be good news, or bad.  ohn Hayes seems to think that a draw tube focuser on a EDGE is not viable, for a couple of reasons.  1) this problem, and 2) the sensor does not remain precisely at 156.5.  Granted, focus travel is very small-fractions of a milimeter, but the principoal is not as sound as something like an Optec.  Unfortunately Optrecs cost over $2,000.  Not in the cards at this time.  I have been thinking about selling everything and starting over and building a "perfect" system.  If I lived in Bortle 1/2 and frequently had 5/5 seeing, I would consider it more seriously.  Robo scope might be the answer--it only costs about $1,600/year to get all data from 3 different scopoes.  $900/year for one scope.  I would be flooded with data if I had 3 scopes sending me data.  But I am not sure I would get the same enjoyment out of just processing.  I am not that fond of processing Hubble images--though its such a PITA to get the data that that probabkly has something to do with it.

[Rodd]

49 minutes ago, Elp said:

I'm guessing mirror flop is more of an issue the larger the scope you go, I don't really get it with my C6 even after meridian flip. I also don't use an external focuser, don't feel the need to.

The C11 is a bit different.  I had a stepper motor attached to the stock focuser and I experiemced serious flop.  It was most apparent when trying to focus during lunar/planetary imaging.  I would have to wait until the image returned to a stable position before analysing focus.  I made the change to a back end focuser mostly for that reason--I never considered how it coukld impact calibration.  I wish I could afford a 10" refractor.  Then I would be in Elysium

[Rodd]

17 hours ago, Elp said:

I point mine up too and put the panel on top of the dew shield. Its on lowest setting and I use two sheets of opal acrylic, one light grey and one dark grey all stacked up. RGB still takes flats less than a second though (F2), narrowband takes a few seconds. On my refractor the time is a few seconds sometimes up to 10 in narrowband. The exposure is calculated automatically via my asiair and the histogram is usually just under halfway to halfway at 3/4 peak intensity.

I use ASair too--though an older version I think.  It does not support dither, and I have never seen a flat calculator.  It doesnt take me more than a few secons to manually get the histogram where it should be.  A couple flats.  I find that SII and Ha can take 30 sec to achieve a good flat--and that is being generous becuase until recently, I was under the impression that 1/4 was good.  Now that folks have steered me to the understanding that 2/3 is better--it would take more than 60 sec to shoot a good Ha or SII flat.  With the FSQ sky flats work amazingly well.  There are no dust motes to speak of so its just illumin ation.  A cloudless sky just before dark works like a dream.  Not with the C11 though--I need a diffuser (fabric over the objective).  That is a pain to do--especially when time is pressing, which is the case for sky flats.  There is a very short window between too bright and too dark

[Rodd]

18 hours ago, andrew s said:

Why need it be 3 times the size? The angular size of the c11 field is quite small.

I dont think its the panel, or even mirror flop at this point.  All knowing minds point to the focuser.  Besides, if the panel was 10 feet away, or 5 or wahtever, the duration of my Ha and SII flats would go from 20 sec to minutes!  

[Elp]

A 30 second flat seems unusually long. Even when imaging F6.3 reduced the max the asiair works out in narrowband 12nm for me is around 10s, and that's for a dimmed flat panel on the end of the dew shield. I'd need to check the gen1 asiair (think it runs the same as all the others), in autorun mode when taking flats you leave the exposure time to auto, then when you set it off it will take various captures, then decide on the best duration and then completes the set number of flats at the duration it decided upon. Its usually always better than when I've tried to set the exposure manually as I take auto and manual then compare them when stacking.

[Rodd]

Just now, Elp said:

A 30 second flat seems unusually long. Even when imaging F6.3 reduced the max the asiair works out in narrowband 12nm for me is around 10s, and that's for a dimmed flat panel on the end of the dew shield. I'd need to check the gen1 asiair (think it runs the same as all the others), in autorun mode when taking flats you leave the exposure time to auto, then when you set it off it will take various captures, then decide on the best duration and then completes the set number of flats at the duration it decided upon. Its usually always better than when I've tried to set the exposure manually as I take auto and manual then compare them when stacking.

I'll have to try that.  Thanks.  Then again, I jusy checked, I am using ASiimg--not air.