Horizontal Lines in Flats

[gnomus]

I have come to the conclusion that there is something amiss with my flats. 

I see a distinct horizontal line at the top and bottom of the frame.  Like this on a Ha flat (I chose this one because it is a bit more distinct than on the LRGBs I do):

When I first started seeing these I convinced myself that this must be something that I was doing wrong.   However, I have now taken numerous flats through different scopes (ED80 and WO Star 71).  I have tried using one of those Aurora panels and more recently an A3 LED panel from Amazon.  I keep seeing these lines.  I use an Atik EFW2, an Atik 383L and 36mm unmounted Baader filters in a 7 position wheel. 

I wonder if this could be having an affect on my images.  Here is a 23 x 5 minute stack of Blue of M45 with minimal processing.  I have cropped the frame very slightly, then done a DBE, some slight NR and a masked stretch:

I think I can see the same horizontal line at the top of this image.  Incidentally, due to an error on my part I was unable to get flats for this stack and so all that is seen here is a quick process of bias only calibrated frames (no darks, no flats).  Note also the halos around the stars which I have asked about elsewhere: http://stargazerslounge.com/topic/260308-m45-blue-filter-haloes/

I was careful to orientate the Baader filters correctly, that is with the arrow pointing towards the telescope (i.e. mirror side away from the CCD chip).  Tonight, however, I found this post on a PixInsight forum (it is the second post on the thread): https://pixinsight.com/forum/index.php?topic=4776.0   

This interested me because I have always used the Reducer with the ED80 and the WO Star 71 has got built-in rear elements.  I now wonder if I should try orientating the filters 'incorrectly' in my wheel.  Before tearing down my 'permanent' set up for the umpteenth time, however, I thought I should ask whether anyone had come across anything like this before and, if so, what can be done to fix it.  A while back, I did see another thread here which appeared to show flats that look very similar to mine.  That chap was also using Baader filters, an EFW2 and a 383L.  Unfortunately, I cannot find that thread now.

Thanks in anticipation of your help.

[swag72]

What ADU do you use for your flats Steve?

[gnomus]

What ADU do you use for your flats Steve?

These were set so that the White point was at ~31000. 

[swag72]

On the QSI site it recommends 35.7k as an average ADU with the 683 which I think is the camera you have?

[gnomus]

I have the Atik 383L. It is the same chip. Would that few extra ADU make a difference do you think?

[swag72]

Sorry Steve for some reason I thought you had a QSI ..... I'm not sure that those figures would be the same for the Atik even with a similar sensor. Hopefully someone can come along soon with the 383L and tell you what they use

[ollypenrice]

I have the Atik 383L. It is the same chip. Would that few extra ADU make a difference do you think?

I doubt it.

A very slight dark horizontal band in the flats may be, predictably enough, turning into a very slight horizontal bright band in the lights. That's what flats set out to do, after all.

Have you tried a stack without flats, just to confirm that the flats are indeed creating the light horizontal band (which I can only really see in the top.) I'm not certain that what I see in the flats and the stretched lights are related, though they could be. A stretched light without flats would decide this.

Regarding the dark band in the flats: it may be electronic in origin and I suspect that it is. Something I would try, myself, would be to run a master flat through Noel's Actions 'Horizontal Banding Noise Reduction' and the re-apply them. If you don't have Noel on board then send me a Drop Box master flat and I'll run it and return it. It would be instructive to see the result.

In looking at your stretched M45 you should not forget that this is a phenomenally deep stretch, the like of which only entered the world of imaging maybe ten years ago. Compare your stretch with this recent APOD. http://apod.nasa.gov/apod/ap150617.html But it would be good to lose that light band for sure. I try to show below the soft bright horizontal band I feel I can see. I wonder if you/others see it?

Olly

[gnomus]

I doubt it.

A very slight dark horizontal band in the flats may be, predictably enough, turning into a very slight horizontal bright band in the lights. That's what flats set out to do, after all.

Have you tried a stack without flats, just to confirm that the flats are indeed creating the light horizontal band (which I can only really see in the top.) I'm not certain that what I see in the flats and the stretched lights are related, though they could be. A stretched light without flats would decide this.

Regarding the dark band in the flats: it may be electronic in origin and I suspect that it is. Something I would try, myself, would be to run a master flat through Noel's Actions 'Horizontal Banding Noise Reduction' and the re-apply them. If you don't have Noel on board then send me a Drop Box master flat and I'll run it and return it. It would be instructive to see the result.

In looking at your stretched M45 you should not forget that this is a phenomenally deep stretch, the like of which only entered the world of imaging maybe ten years ago. Compare your stretch with this recent APOD. http://apod.nasa.gov/apod/ap150617.html But it would be good to lose that light band for sure. I try to show below the soft bright horizontal band I feel I can see. I wonder if you/others see it?

Olly

Thanks for responding Olly.  I do not have Noel's actions (yet) and so here is a link to that Ha flat (FITS): https://www.dropbox.com/s/7fwre980tlyoi0e/ha%20Flat.fit?dl=0

I can definitely see the horizontal line in the image that you annotated.  That image is nearly 2 hours of blue without flats.  I thought I could see the line in roughly the same place I see it on my flats.  I wondered, therefore, if the same issue could be producing this artefact on the flats and on the lights.  I went back through some earlier data and found a similar line in a version of M31 (excuse the horrible processing - I just wanted to make the line clear):

The fascinating thing is that whilst this was shot with an Atik 383L (through an ED80), it was not shot with the same 383L as the Ha flat linked to above.  There seems to be no consistency: in my recent M45 attempt (ongoing) I see no line in this 13 x 5 minute exposure shot through the red filter and calibrated with red flats:

I know that the orientation of my reds are different to my blues (yet another tear down and reconstruction of my 'permanent' set-up that went awry - doh!!!).

Processing is one of my many areas of weakness and I wonder if this is all down to me 'over-stretching' my images - it is usually the case that my stars are blown out to pure white, which does suggest this to me.

Regards

Steve

[ollypenrice]

Firstly I processed the flat you made available for contrast and, yes, something is certainly going on. Both the top and bottom long sides show dark bands from the chip edge to the area arrowed in red. There is at least one sub-band in each as well. (Red dots.)

When I look at this I'm unable to decide between an electronic and an optical explanation. Maybe the electronics buffs could comment? To my eye it could be either, unfortunately.

Now the problem is not present in the last image taken through a red filter and with red flats. It is common to suffer from internal reflections more severely in blue than in red, so maybe the problem is optical.

However, I'm lost on the camera saga. Is this a different camera from the one which took the bad Ha flat? If so, does the new camera behave itself in blue? Basically the present camera is OK in red? So how is it in blue?

Olly

[gnomus]

Firstly I processed the flat you made available for contrast and, yes, something is certainly going on. Both the top and bottom long sides show dark bands from the chip edge to the area arrowed in red. There is at least one sub-band in each as well. (Red dots.)

ha Flat bands.jpg

When I look at this I'm unable to decide between an electronic and an optical explanation. Maybe the electronics buffs could comment? To my eye it could be either, unfortunately.

Now the problem is not present in the last image taken through a red filter and with red flats. It is common to suffer from internal reflections more severely in blue than in red, so maybe the problem is optical.

However, I'm lost on the camera saga. Is this a different camera from the one which took the bad Ha flat? If so, does the new camera behave itself in blue? Basically the present camera is OK in red? So how is it in blue?

Olly

Thanks Olly. The M31 was taken back in September with my original 383. I then noticed that I had an issue with streaks/stretching of hot pixels in my darks. This was discussed in a thread here around Nov/Dec from memory. The new 383 is what I used to take the new Ha flat and the offending M45 blue. Weather has been so bad that there have been few opportunities for imaging round these parts, so I have only just noticed the Flats issue. It seems to me that if this was an isolated issue relating to my current camera then it would be extraordinary to get the same issue with a different camera.

Did you see the link to the PixInsight forum in my original post (relating to orientation of the filters)?

I think I will try doing a flat for each of my filters to see which ones show the issue and which do not. I may do that tonight because, believe it or ot, the forecast ain't so good.

[gnomus]

OK.  I've run a set of flats through all of my Baader filters.  I do not see any horizontal lines in my LRGB filters.  I see them in each of the narrowband filters.  The position of the horizontal line is slightly different in each NB filter.  In the OIII filter, I can just about convince myself that I can see 2 lines top and bottom (as indicated).  Looking down the scope, there is nothing there that has a straight line except, of course, the window behind which is the CCD chip.  I am wondering whether I am getting some sort of reflection of the window in the NB filters.  I am quite certain that I put these in the correct way round. This has to be optical doesn't it?

I am still at a loss to explain the horizontal bright band in the Blue channel posted above.  Here are the flats.  These are all exposed to around 32000 ADU and tweaked a bit in PI to show the defects.  Incidentally whilst determining the correct exposures to get ~32000ADU I tried various settings and the horizontal lines appear irrespective of ADU.

[ollypenrice]

Yup, I agree. Progress! Baader are a good firm and would, I think, be genuinely interested in this. I'd send them the files for comment.

It has to be optical in my view but it may be the fault of reflectivity somewhere in the system rather than strictly the filters. I think this rules out camera electronics.

I think that, for now, I'd put the blue problem down as - well, a red herring. (About the least convincing suggestion I've ever made, in the circumstances! However, I do think it might be best set aside. Besides, if it is a reflection problem it would most likely strike in blue.)

Olly

[gnomus]

Thanks. Before contacting Baader, I will strip the wheel down just to check that the filters are orientated correctly. But before that, I will gather more R,G & B to complete my M45. I'll redo the blues completely. Would Lum be advisable on this target?

Thanks again for your advice.

[Oddsocks]

Have been following this with some interest.

You have almost excluded camera electronics but not quite.

When you take the flats, to exclude an electronic source they need to be all taken with the the same exposure time, I am guessing at the moment your RGBL flats are taken at only a fraction of the time that was used for the narrow band flats?

If necessary, add neutral density absorbers to your flat field light source for the RGBL flats, a few sheets of plain printer paper works well, and try to match more closely the exposure time for the RGBL and narrow band flats.

It is interesting that in the KAF 8300 the well the chip sits in within the semiconductor package is asymmetric and contains a stepped ledge at the top and bottom long edges but only a single ledge at the short sides, also, there are only reflective gold connection wires along the long sides of the sensor unlike many other chip designs that have gold wires on all edges of the sensor, image attached below, if the cause is reflection artefacts then that might explain the two-stepped artefact that you see in your flat images.

If it is a reflection artefact between the filters and the chip then adding a temporary spacer tube between the filter wheel and camera to increase the distance between the two will change the position and width of the artefact.

I notice that at least one manufacturer using the KAF 8300 adds a blackened mask inside their camera to shield the edges of the sensor well and only the active area of the chip is visible when looking into the camera window.

 

[Oddsocks]

Another simple check for a possible reflection artefact between the sensor and filter is to impart a slight tilt into the filter mount, if you are using unmounted filters then add a tiny sliver of single thickness paper between the filter wheel recess and the edge of the filter to tilt it a fraction of a mm. at the sensor plane any reflection artefact will be shifted quite a considerable amount.

Note: for greatest effect you would need to add the paper sliver beneath the filter to correspond with the long edge of the sensor chip.

[gnomus]

... I am guessing at the moment your RGBL flats are taken at only a fraction of the time that was used for the narrow band flats?

...

In fact, these flats were taken with broadly similar exposure times.  Even with a sheet of dark grey perspex over my LED panel, I have to use different thicknesses of diffusion material over the objective lens to give me a reasonable length exposure (so that the shutter does not become an issue).  In any event, as I said, whilst determining the 'correct' exposure (to give ~32000ADU) a range of exposure lengths had to be tried - the line was identical with each NB filter regardless of exposure.  

[ultranova]

Hi Steve, did you ever get to the bottom of the problem,

I have just posted a picture in the imaging section with the

same problem,

http://stargazerslounge.com/topic/260917-ic-405-flaming-star/

My Ha flats don't really show the dark band at all,

my Ha flats are, if I remember are around 15 seconds long

Paul

[gnomus]

Your issue looks very like mine. I plan to check that I have the filters set in the wheel the right way round. I haven't done it yet because I was finishing off an LRGB run.

These must surely be reflections of the camera chip window don't you think?

[Oddsocks]

These must surely be reflections of the camera chip window don't you think?

That is my take on it too and it will be interesting to read what your findings are.

I had similar problems using a Trutek filter wheel, Starlight Xpress SXH9 and Astronomic filters around ten years ago.

In my narrow band images you could see clearly a reflection of the CCD chip socket and in Ha you could even see the edge of one of the circuit board components adjacent to the CCD socket.

In LRGB there was no reflection visible at all, most clearly present in Ha and SII and less so with OIII.

At the time, as Sara found more recently, if you removed the (screw-in) Ha filter from the filter wheel and screwed it directly into the camera nose piece the artefacts totally disappeared.

I reasoned that as the reflections were absent with the filter directly mounted in the nosepiece then the reflections with the filter mounted in the filter wheel were due to non-filtered off-axis light leaking around the filter wheel housing and illuminating the camera window, the illuminated components inside the camera port then being reflected back off the NB filter to form an image on the CCD chip, the comparative difference in exposure time between LRGB and narrow band allowing a large proportion of the reflected image to contribute to the final NB component.

Since the highly reflective narrow band filters send the non-target light back up the OTA I assume this unwanted light then bounces back and forth between the various optical components in the OTA and gently illuminates the inside of the filter wheel.

Opening the TruTeK wheel the internal surfaces of the wheel compartment were only lightly anodised and not really black, more a grey shade, so I used flocking paper on the upper and lower compartment faces and also completely flocked the wheel on both sides.

This made a big improvement and reduced the reflection artefacts by around ninety percent but did not completely eliminate them, in a twenty minute Ha exposure you could still just see the CCD socket.

In the TruTek wheel the physical gap between adjacent filters was quite small so I figured the remaining off-axis light could be leaking via neighbouring filters in the wheel and removed all but the Ha and covered the other filter ports temporarily with flocking paper, now even with 20 minute Ha exposures you could not see a reflection.

I did think about making a collar to fit inside the filter wheel around the entry and exit ports to reduce the gap between the wheel and the wheel cavity upper and lower faces, to reduce the risk of stray light passing through the adjacent filter ports, but at that time I hadn't a lathe so tried a different approach and made a flocking-paper mask to sit on the CCD window with just a square cut-out in the centre to expose the CCD active pixel area, the rest of the sensor, sensor socket (which was bright pink) and the PCB were all covered by the mask, and that solved the issue for me with the SXVH9 -TruTek wheel combo.

Since moving to the QSI with internal filter wheel and Astrodon filters I don't have any reflections or artefacts at all, at least that I can see, but in the QSI the filters sit immediately above the shutter and CCD window so there is not enough of a gap between filter and CCD to allow a reduction-reflection to appear in the image.

If you tilt your filters a fraction of a millimetre in the wheel and the position of the line artefact move in your image then that would positively prove/disprove the reflection theory.

HTH.

[ultranova]

Thanks HTH, does sound a very solid theory and definitely worth  

looking into, look like a flocking I shall go,

Just strange the amount of people that have the Atik 383l and

Atik Ewf2 filter wheel and do Narrow band this is not reported more

often, maybe there other variables to consider,

First thing I am going to do is move my light pollution filter

back between the camera and filter wheel, as after checking Ha images

before I moved it  there is no black band then

if its still there then ill move off onto the extreme

solutions

Paul

[gnomus]

OK. I've double checked. I took my filter wheel apart and took the NB filters out. I had them seated correctly. I replaced them and redid some flats. They are still showing the horizontal lines. I have had this now with two telescopes - an ED80 with focal reducer and a WO Star 71.

I think I will e-mail Baader.

[Oddsocks]

First thing I am going to do is move my light pollution filter

back between the camera and filter wheel, as after checking Ha images

before I moved it  there is no black band....

solutions

Paul

I know many imagers do use LP filters permanently in the light path in conjuction with NB filters in the wheel but I don't see any filter manufacturers actually advocating this approach. Possibly ok with LRGB but the problem is that LPand NB filters are interference filters, very similar to optical gratings, and using both effectively stacked together will create constructive and destructive wave patterns at the sensor.

Adding a LP filter double-stacked with NB filters will have unpredictable outcomes.

A possible work around if you must have the LP available for the LRGB in the wheel is to add a second filter draw in the optical path so that the LP can be withdrawn when NB is in use but personally, if LP is serious problem, I would stick to NB only and reserve LRGB for those occasions when imaging under clear dark skies.

Spent much of the 1970-1980's trying to fix optical interference patterns in lens systems attached to image intensifiers which output light in a single wavelength.

A shift of the position of a lens or beam splitter relative to the image intensifier output screen of less than 0.001mm would introduce standing wave patterns into the camera image and with the image intensifier housing over a meter in length this would easlily occur with changes in ambient temperature, had to be dealt with using temperature compensated optical beam paths.

The orientation of the patterns, thickness of the bars, stationary or moving, had so many variables such as separation between lens elements, progressive or interlaced scanning of the camera, thickness of lens coatings etc that you never quite knew what to expect when setting up a new Image intensifier the first time....made for an interesting job though.

If you haven't seen this document by Alan Homes at SBIG cameras it contains some interesting and useful information regarding flat field problems that can afflict astro imaging.

http://www.sbig.com/about-us/blog/flat-fields-the-ugly-truth/

William.

[gnomus]

I had a response from Martin at Baader Planetarium.  He says:

... after seeing your images, showing strictly horizontal lines at the very border of the images it gets clear that there is no filter defect. This would have a different pattern.

Also a electronical defect is very unlikely, it would stray over the whole image.

I know such patterns from very narrow, rectangular constrictions in the optical path.

Or from highly reflective parts alongside the optical path.

F.e. a rectangular filter holder or cover. It is indeed possible that there is a brightening effect, resulting from grazing reflections off such a element.

Less likely, but possible: it could result from the sensor body itself, having a unlucky distance between surface and filter.

For checking if it is a optical sensor interaction it would help to use another camera with different sensor, albeit difficult to realize.

Solutions to this are:

-better matted parts inside all adapters and holders. 

-changing the distance between filters and sensor (extra tube or else) to move eventual reflection knots

-as trial: flip the filters, even with more reflective side pointing to the sensor it could be better (surely more reflection, but also different behaviour and maybe improving).

I am convinced that this must be a reflection of the camera sensor window bouncing off of the back of the filter back down on to the sensor itself.  I would like to try Martin's suggestion of changing (increasing - I can't make it any smaller!!) the distance between the filter and the sensor.  Does anyone know of an adapter that would allow me to do this?  I may also try reversing the filters.  A different camera would be a bit of a stretch right now.  If I get anywhere I will report back. 

[ultranova]

Thanks William  / Steve, made for some very interesting reading on both accounts

If I come up with a solution I shall report back

Paul

[gnomus]

I have now tried the suggestions from Martin at Baader.  Reversing the orientation of the filters made little difference.  I thought that the Ha was a little better (but perhaps the line was now more prominent at the bottom of the frame).  So I put the filters back in correctly.  I then found a 5mm extension tube.  I put this between the camera and the filter wheel.  This seems to have fixed the problem.  I no longer see any lines:

   

I imagine that the vignetting may be a little worse with this set up.  And I am sure that the camera and filter wheel were not designed to be used with a 5mm spacer between them.  I need to try this out with some actual images, and this will, of course, have to wait until the present storms pass.  Preliminary test frames seem OK, however.