Is this a problem with my flat frames

[billyharris72]

Hi all:

Here's an LRGB effort from a couple of days back, stacked in AstroArt and adjusted to bring out the issue I'm talking about. It seems the edges (especially bottom right) are brighter and redder.

Is this likely to be a problem with my flat frames? I was thinking maybe it has over corrected somehow, but is this a thing that can happen? Flats were taken with the same gain and offset as the subs (on an ASI1600mm), so wondering how over correction could even happen.

Another option I thought of was maybe the flats themselves have an issue. I've attached one - I don't think it looks too bad, but could there be an issue with the filtering? I guess the question is does light wavelength have a significant effect on vignetting. The flats were all taken with a luminance filter, using a flat panel made from a LED tracing pad - if the light was biased towards the blue end of the spectrum could this have the effect of producing a flat that would over correct for red? Just a thought - really have no idea.

Thoughts much appreciated - has anyone else encountered this, and if so how did you solve it?

Billy.

 

EDIT before I've even posted. Just looked at the flat again. The scope looked reasonably well collimated in my Cheshire (checked the day before, so could be out but should not have shifted much), but is that the secondary spot I see near the bottom of the frame?

[bobro]

The flat is totally black towards the corners and totally white at the centre. A flat needs to be an intermediate brightness level in order to correct an image. Perhaps a lower gain would help? I'm not familiar with the ASI1600mm or AstroArt, however DSS for example correctly applies flats even when taken with a different ISO to the main image.

[daz]

As bobro says, the flight is too bright in the centre and too dark at the edges. Also, you'll need to take flats for EACH filter - not just use the L filter for all flats.

The ring you can see at the bottom of that flat frame is a dust bunnie - exactly what you are supposed to see in the flat. It will help eliminate vignetting, dust bunnies and other illumination artefacts.

 

[Art Gecko]

I would recheck your collimation... I had some images go a bit like that, with uncontrollable patches of colour and shade... turned out to be my secondary mirror that had slipped and I hadn't realised while collimating the primary... I may be wrong in this instance, but worth checking... and old carpentry saying "measure twice, cut once"

hth, Art

[billyharris72]

Thanks all for the advice.

Bobro & Daz (good avatar) - I will go back and check the exposure on the flats, though I think the apparent high gradient (from total black to total white) is a result of my poor choice of visualization settings on the fits file. I've attached one with less extreme black / white points and the x axis ADU profile (saturation at 64k) below. Total light fall off from centre to edge is about 25% or so, with the centre slightly less than half saturation (is this correct or should I be exposing more / less)? Wasn't aware of the need to flats for each filter - bit of a pain but hopefully it will make a difference.

Art - agree 100%  that I need to recheck my collimation. Usually do it every session but was reluctant to remove the camera as I didn't want to have to refocus etc.

I've dug a bit further and the whole image is a complete mess to be honest. There is significant over correction, so I tried stacking without flats and got this.

As you can see the colours are totally uneven, with some parts of the image green and some red. No idea how this could come about, as I was shooting R,G,B in a repeating sequence using an electronic filter wheel.

Anyone seen this before?

Cheers,

Billy.

[michael8554]

The black corners of the flats may not necessarily be an exposure error if the chip is totally uniluminated there?

Are your focuser, nosepiece, filters, too small for the 1600?

Michael 

[Oddsocks]

Hi Billy.

Couple of things to look at / clarify...

The uneven vignetting pattern may be a poor collimation or the method of collimation used that results in either a centred or off-centred alignment of the secondary. Either method will result in a good apparent collimation but the centred method produces an uneven vignetted field while the off-centred method produces an even vignetted field. Other possibly causes of an uneven vignetted field include optical path tilt due to a loose coupling between camera and flattener, poor coupling between flattener and focuser or focuser draw tube alignment to the focuser body etc.

I found Vic Menards booklet 'New Perspectives on Newtonian Collimation', available from FLO, a great help when trying to understand the theory behind Newtonian collimation and tune up a fast (f3.8) Newt' though I never achieved the best the telescope could deliver until I began using an autocollimator eyepiece as described in later chapters of the booklet.

https://www.firstlightoptics.com/books/new-perspectives-on-newtonian-collimation-vic-menard.html

When using a LED panel for flats, placed directly on the front of the OTA, there is a risk the flats will be contaminated by indirect illumination of the focuser entry port and focuser draw tube as shown below. The light path from flats source to the camera when taking the flats should be identical to the light path taken for the lights, any differences will show up when the flats are used to calibrate the lights with the effects being more or less pronounced depending on the diameter of the OTA, distance the focuser is from the end of the OTA, focuser port diameter, focuser draw tube diameter, type and quality of the matt-black paint or flocking used around the focuser entry port etc.

The focuser entry port and draw tube must be absolutely black and non-reflective if you wish to avoid this. The focuser entry port and draw tube can be flocked with flocking paper or painted with a pigment based matt black paint such as Matt Black High temperature Barbecue paint from B&Q:

https://www.diy.com/departments/rust-oleum-stove-bbq-black-matt-matt-stove-bbq-spray-paint-400-ml/128291_BQ.prd

If using the spray paint in the link, spray some of the contents into a glass jar away from the telescope and apply to the focuser entry port with a brush, do not use the spray paint directly from the spray can on the telescope because of the obvious risk to the optics and bearings/bearing surfaces of the focuser mechanism.

Or simply take steps to move the LED panel further away from the front of the OTA. A wrap-around matt black dew shield of the kind used by SCT owners can be fitted to the front of the OTA and the LED panel placed on the end of that, or a cardboard packaging tube flocked or painted inside with black-board paint can be used, pointing the telescope into one end of the tube and the LED panel at the other. Obviously, these techniques are more difficult to apply to a Newt with it's long OTA and I still think that sky flats taken either before the session begins during twilight, around half an hour to an hour after the sun dips below the horizon, or pre-dawn flats taken an hour to half an hour before sunrise, will give the best flats. Using a wrap around dew shield on the front of the OTA will help guard against stray light entering the focuser port indirectly even when taking sky flats and is essential when battling nearby sources of light pollution such as street lights and house lights etc that can shine obliquely into the focuser entry port depending on the orientation of the OTA.

Flats must be taken at the same focus as the lights so mark the focuser draw tube position against the focuser body with a felt pen or blob of Tippex fluid next time the camera is bought to actual focus on a star and then make sure the focuser is set to the same mark if taking the flats at a different time.

I see that you are using Astroart for calibration and having used that myself for several years I know that program can produce excellent calibration results. It does need BIAS frames though and will not produce a correctly calibrated flat if you omitted to take and use them. Missing the BIAS frames does lead to an overcorrection when applying the flats since the BIAS frames contain the cameras pedestal offset value which can not be determined from lights, darks and flats alone.

If you missed the BIAS frames then you can take them anytime as long as the BIAS frames are taken in pitch darkness and ideally at the same temperature as the lights though with the new CMOS cameras that seems less important than it once did with their very low dark current. If you did not use BIAS frames originally try taking some now and go back to the original data and recalibrate using the new BIAS frames. 

You can substitute a master BIAS for a master DARK with these low noise CMOS cameras as long as you use mount dithering combined with a bad pixel map and a sigma reject combination method when combining the LIGHTS. If you can't use mount dithering then stick to using matching exposure time LIGHTS and DARKS and the shortest possible exposure time for the BIAS. For the majority of commercial and free image calibration programs the BIAS frames can not be omitted if calibration is to be correctly applied, for the new breed of CMOS cameras DARKS may be optional.

If you look at this recent thread it shows the results of omitting BIAS frames during image calibration:

HTH

William.

[billyharris72]

Thanks William, that's some really useful advice. I'd never considered that point about the light path from a flat panel, but it's really obvious now you mention it. Might have to experiement with either sky flats or moving the panel out a bit. My focuser tube is flocked, but even so it will still reflect some light.

I'm thinking I will try again with the flats, this time using the normal AstroArt preprocessing menu. Rather than follow the usual route I did a prior median combine on my flats, darks and flat darks (I shot flat darks rather than bias, though that should contain the bias signal also) to create masters and then worked from them - I'm now wondering if i have slipped up in that process somewhere (I can't remember exactly what I did but I used the preprocessing menu and I can see a couple of routes that could lead, for example, to the flat dark inadvertently being subtracted twice).

Also, thanks for the recommendation of the Menard book. I've faffed around buying that before and ultimately decided it was probably overkill; now seriously reconsidering.

Billy.

[billyharris72]

Hello again:

I may have made a bit of progress - not at fixing the image but at least understanding the problems. Having looked at the luminance frame, it is not too bad; there is a gradient (probably LP from my site) but no sign of significant over correction from the flats. The problem is with the colour subs, and especially the red - rather reinforces Daz's point about needing to take flats for each filter. It is a bit dark in the bottom right corner, but it's not totally black (think this might be a collimation issue). I reckon the gradient tools in Astro Art could fix a lot of this though (this is just pre-processed).

What's of more interest to me now is the strange difference between the colour subs, especially Red and Green. It almost seems that the centre of illumination is different for each filter, despite them being shot in a rolling sequence (R,G,B, repeat). Here's the red sub (stretched to highlight the apparent centre of illumination).

And now the green.

Any ideas on what could cause this behaviour? I don't think it is a mechanical issue with the filter wheel; I checked the filters and they are all nice and snug.

Thoughts appreciated.

Billy.

[Oddsocks]

I don’t think that is so unusual, I see the same with my camera using very high quality dichroic filters though not so severe. It may be that the filters are not sitting perfectly level in their holders, or the threads in the filter wheel disk for the filter holders are not perfectly perpendicular to the plane of the disk so that the filters are tilted slightly, it may even be that the disk detent position is slightly variable for each filter slot and the disk is not stopping in exactly the same spot for each filter. Could even just be manufacturing tolerances of the filters.

As Daz mentioned, using per-filter flats will fix that.

Below is an extreme stretch of RGB flats taken with Astrodon filters and a QSI 583 camera, even with these filters (and a poorly collimated and flocked RC scope) there is a noticeable difference in the shape and distribution pattern of the flats.

[JohnSadlerAstro]

Hi,

Sorry to resurrect an old thread, I wondered if your issues have been solved @billyharris72? I had similar looking flats, although using a 130P-DS the coma follows the flats most of the time so I got horrible coma on one side of the image. (Coma was bad in the same areas as the dark patches in the flats). The flats were working fine at removing gradient, but I was not happy at all with the amounts of coma introduced.

It looks to me as if your secondary may be at the wrong angle, so the light cone is falling away from the centre of the camera sensor. 

This was a stretched flat from my 130:

Although through the Cheshire the secondary looked in perfect alignment, I made the adjustments that were logically required and they worked on the flats, with the Cheshire still looking good after tweaking the primary. I'm still a bit unsure as to the extent of the relationship between coma and secondary collimation, though.

The main issue was the angle of the secondary in rotation and its pitch angle. Neither of these seem to be indicated by the Cheshire, so I'm hoping that this will have sorted the issue. Of course the real test will be a star field! ? 

John

[billyharris72]

Hi John. I've not had the chance to do much imaging since then and have mostly been with the DSLR (more chance of catching the gaps in the clouds / taking advantage of the brief window before the South Wales dew clobbers my mirrors).

I'm not sure I've really nailed flats yet, but have changed 2 things for the better.

1) Binned the laser collimator and moved 100% to the Cheshire, with a big improvement in collimation. I've gone for no secondary offset, but at f5 it's not strictly necessary (vignetting is slightly uneven but fixed by flats, coma is distributed around the centre).

2) Ditched the panel and moved to sky flats (miles better).

I'm surprised that the secondary was having such a big effect on coma for you - I think it should be quite insensitive to small misalignment. Were you using a laser at all (that would translate any small secondary misalignment to the primary and mess things right up)?

Billy

[JohnSadlerAstro]

Hi,

I was using a Cheshire--the main issue seems to be that I took the scope apart for flocking in the winter, so had to completely remove everything. For some reason it did some long-term messing up with the mirrors. I'm a bit worried about the coma though--Im hoping it will have gone away with the adjustments.

John

[ollypenrice]

I think there are at least two issues here. Firstly the off centre illumination does suggest that the optics need attention. I'll leave that to those who understand Newtonians...

However, accepting that the illumination is indeed off centre there is also the question of why the flats were over-correcting. Now this problem does come up on occasion and I've been plagued by it on a rig I used to operate here. I never got to the bottom of it but my reading up suggests the following lines of enquiry:

- Flats not dark-subtracted may over correct. Using a master bias as a dark for flats should sort this out.

- Be careful to use the same capture mode when taking flats and lights. Some capture software has a fast download option. Don't use it.

- Flushing between subs may not be adequate in some capture software. Greatly increasing the delay between subs seems to have helped some people.

Olly

Edit. One other thing: posting screen-stretched flats can be very misleading. We can tell nothing at all about the real level of vignetting by looking at a stretched flat. The best bet would be for the poster to state what the ADU values were in the corners and the centre. That would tell us the range of the vignetting. A screen stretch can turn a 10% fall-off into a 90% fall-off.

[Adam J]

2 hours ago, ollypenrice said:

 

- Flats not dark-subtracted may over correct. Using a master bias as a dark for flats should sort this out.

That's exactly the problem, but you are a little off on the solution. You shouldn't ever use bias on a CMOS as in the case of the ASI1600, due to unstable bias, with CMOS cameras always calibrate flats with dark flats. Flats should be more than a second long ideally more than two seconds. 

[ollypenrice]

7 hours ago, Adam J said:

That's exactly the problem, but you are a little off on the solution. You shouldn't ever use bias on a CMOS as in the case of the ASI1600, due to unstable bias, with CMOS cameras always calibrate flats with dark flats. Flats should be more than a second long ideally more than two seconds. 

Sorry, yes, I'm so used to CCD that I'd forgotten about CMOS bias. Clearly it makes sense, in that case, to use longer subs for flats.

Olly

[billyharris72]

Have to say that I use bias frames and have not had any major issues. I know the bias pattern can change, but as I understand it that's more likely when the camera is restarted. Any tests I've done (standard deviation on mean or median stacked bias) shows a clear fall in standard dev in proportion to square root of the sample size. Defect maps based on these tend to be almost identical, suggesting the noise is random and the underlying pattern quite stable.

Also worth bearing in mind that with flats the aim is less removal of noise (though this is good) as removal of the mean value of what you might call "ambient" bias value - you don't want to be dividing your pixel values by something that's had a constant added.

I do agree that dark flats are probably better than bias, but on the cooled 1600 the thermal noise is so low that they won't be very different from a stacked master bias. I do think it helps to try to shoot all the calibration frames each time (without restarting the camera) to catch fluctuations in bias, but again, "library" files seem to work better than nothing.

I'm interested in this though - any more info on the subject? The only other thing I can think of is amp glow in the edges (which my cam does have a little of), but that should result in undercorrection.

Billy.