Further to my previous thread here, I have bitten the bullet and swapped out the focuser on my FSQ106 with a new FTF3515 from Feather Touch. A very nice piece of hardware it is (if a tad expensive!). The improvement in the data quality is huge with what now appears to be a nice flat field as measured by the tools in Pixinsight and examined visually.
I am, however, still having difficulties with star alignment when registering the images taken either side of the meridian so there is definitely something awry with the data. I've tried RegiStar and Pixinsight and both give me the same (bad) results. That is that the bright stars do not quite line up in the registered images. Blinking back and forth shows the bright stars to be not aligned. If I take this data through the full image integration and channel combination process I know that I will finish up with coloured flares around the stars which is not good.
There is a link here to a Dropbox folder which contains pre and post meridian flip images. I have saved the raw and calibrated subs in *.fits format and *.xisf format.
The attached images show the output from the aberration inspector script and the FWHM eccentricity script in Pixinsight. These indicate that the data is clean and the field is relatively flat with no obvious tilt or field curvature. If I take the time to peer very intently at the brighter stars in the images I can maybe see some slight elongation that is different in the pre and post flip images but it is very small and I am needing to zoom in to 200% to pick anything. At the end of the day though, there is something in this data which is causing the brighter stars in a registered image to not line up.
I'm working on the assumption that I may still have some residual tilt that my eye and the PI tools are not picking up. This being the case, I would like to hand over this data to my learned colleagues on this forum to see if anyone wants to forensically analyze this data and see what can be found.
FWIW, I've downloaded an evaluation copy of CCD Inspector to give it a run. Attached are the curvature map plots extracted from same. As I am not familiar with the use of CCD Inspector, I am hesitant to try and interpret these results. If others who are more experienced could chime in and interpret this for me this would be good.
The plots are one for the pre meridian flip image and one for the post meridian flip image. I've set the image parameters in CCD Inspector to make my configuration which is an image scale of 2.09" and pixel size of 5.4 microns (QSI683).
Last edited by Ryderscope; 09-02-2021 at 02:34 PM.
Congrats on your focuser solving that issue. Unfortunately its all too common a story with FSQ's. Let's hope they finally got on top of it with the EDX4. My EDX3 was really good but obviously there is sample variation with them.
I just registered your subs in CCDstack 2 with the CCDIS plugin.
Looks perfectly fine to me. Perhaps your registration software is the issue here.
I find CCDstack 2 and the CCDIS plug in to be very good. It only fairly rarely fails (usually when the subs are too far apart in alignment. Then I do a manual align and then redo CCDIS and it usually works.
Bah!..ditch that CCD Inspector/PI analysis palava its bogus and not what people look at, it is the final image that counts and from those mozaics you posted...well?...in the immortal words of Gary Coleman.. "watch you talkin bout Willis..?"
If you have some registration issues it is your registration software, or how you use it, that is the likely culprit
Congrats on your focuser solving that issue. Unfortunately its all too common a story with FSQ's. Let's hope they finally got on top of it with the EDX4. My EDX3 was really good but obviously there is sample variation with them.
I just registered your subs in CCDstack 2 with the CCDIS plugin.
Looks perfectly fine to me. Perhaps your registration software is the issue here.
I find CCDstack 2 and the CCDIS plug in to be very good. It only fairly rarely fails (usually when the subs are too far apart in alignment. Then I do a manual align and then redo CCDIS and it usually works.
Greg.
That's very interesting Greg. I will download a trial version of CCDStack and give it a shot.
Bah!..ditch that CCD Inspector/PI analysis palava its bogus and not what people look at, it is the final image that counts and from those mozaics you posted...well?...in the immortal words of Gary Coleman.. "watch you talkin bout Willis..?"
If you have some registration issues it is your registration software, or how you use it, that is the likely culprit
Mike
Thank you Mike for bringing things down to reality. I've never had this sort of a problem before and accept that I may be getting down to minutia. I do suspect though that if I run the full data set through the workflow it will show up in colour flares around the stars. I do agree though the sometimes we can drill down into the sub pixel level and get lost in the haze
I just registered your subs in CCDstack 2 with the CCDIS plugin.
Looks perfectly fine to me. Perhaps your registration software is the issue here
Greg.
Hi Greg, I've downloaded CCDStack and the CCDIS plug, aligned the two frames and then ran the blink function. When I blink them I am seeing exactly the same problem. I've uploaded a MP4 video file of what I'm seeing in the CCDStack window. Click here to access video . Note that it seems to be only the brighter stars that are misaligned.
I only used the .fits files. XSIF does not work in CCDstack. I think its an image format for a processed image.
I only use .fts when registering. The only operation I do on the RAW files are calibration -darks/flats/biases depending on the camera).
Your lettering system - I assumed r means registered, so I didn't use those.
So in your dropbox you should really only have the RAW fits files, nothing registered once before. Calibrated fits are fine.
I have never seen CCDstack do that. Only if they in fact did not register.
Is the blinking of what? It looks like one is preregistration and the other is post registration. Did you do a combine? CCDIS tells you when registration fails. Did it say it failed?
Is this what you did?
I am downloading those files again and will run CCDStack CCDis on it
with the accurate box ticked.
Also why do all the fits files have the same file name? Are they copies of one image or are these different images?
I only used the .fits files. XSIF does not work in CCDstack. I think its an image format for a processed image.
I only use .fts when registering. The only operation I do on the RAW files are calibration -darks/flats/biases depending on the camera).
Your lettering system - I assumed r means registered, so I didn't use those.
So in your dropbox you should really only have the RAW fits files, nothing registered once before. Calibrated fits are fine.
I have never seen CCDstack do that. Only if they in fact did not register.
Is the blinking of what? It looks like one is preregistration and the other is post registration. Did you do a combine? CCDIS tells you when registration fails. Did it say it failed?
Is this what you did?
I am downloading those files again and will run CCDStack CCDis on it
with the accurate box ticked.
Also why do all the fits files have the same file name? Are they copies of one image or are these different images?
Greg.
I also only used the fits files, the calibrated versions. Did not do a combine, only used the align (register) function. The steps implemented were:
- Open images *_2021022707_c_cc.fits & *_2021023456.fits
- Stack -> Register -> CCDIS 'align all' -> Apply 'Apply to all'
- Image select tool 'Blink'
My understanding is that this should blink the open images which have now been aligned.
For clarity I have removed the registered files and XISF files from the Dropbox folder.
Also why do all the fits files have the same file name? Are they copies of one image or are these different images?
Greg.
Sorry, I just saw this comment. Attached is a screen shot of the directory of the Dropbox folder that I have linked to. There are two subs and I had included the raw and calibrated version of each. Were you registering the same image maybe but the raw and calibrated version?
Just to be safe anyway, I have removed the calibrated versions from the Dropbox folder. There are two files contained therein:
NGC3293_V2_300sec_Green_20210207_02 2707.fit (pre meridian flip)
NGC3293_V2_300sec_Green_20210207_02 3456.fit (post meridian flip)
HI Rondey.
Could it be that the bright stars have bloomed a little, and that is what you're not seeing line up when registering?
Regards
Josh
By blooming, do you mean over saturating the pixel well causing it to leak charge into adjacent pixels? If yes, maybe but I don’t see that with the QSI camera on my TSA120 so my presumption would be probably not. The QSI is an anti blooming CCD as well.
Having looked at the raw files I’m wondering whether it is a collimation issue. Given that the stars look pretty good (appears to be some tilt... or is it!) I’m thinking that the FSQ is collimated but not into the centre of the mechanical axis, this may answer why the image seems to rotate around the top left corner.
It may also answer why the smaller stars remain quite still across the FOV but the larger ones are moving. If you look at the bright stars without any stretch their central PSF remains largely unchanged but as you look at the brighter flare, that’s what’s “moving”. I’m thinking that this is how the light is collimated, light further from the central optical axis has to move further across which gives it a different incidence angle. If the optical axis isn’t central to the mechanical then when it’s flipped 180° it isn’t symmetrical. Although the stars are in the same place the lights angle of incidence is different; non-symmetrical.
Hmmm, quite plausible Colin but hard to tell. I do note that the image angle reported in the fits headers of the pre and post meridian flipped images shows that the change was 179.49 degrees rather than 180. I am assuming that this is due mount mechanical issues and probably a red herring. It could explain why the two images appear to rotate around a centroid in the corner.
Nevertheless, my working theory at the moment is that there is a small collimation issue combined with some tilt. I will do some more work on this to see if I can confirm this. I am preparing myself for the possibility that the OTA may have to be sent to Tak for service. If this is necessary I will be asking some more questions as I have now idea how one goes about that.
I finally got those 2 files - thanks for separating them.
I aligned them with various CCDstack and I see what you mean. The small stars are fairly static but the brighter ones are moving slightly.
I also did a star alignment in PixInsight and got the same result.
Odd. Is a lens loose in your lens cell or something? Odd its only on the brighter stars.
Greg.
Thanks Greg, there is nothing loose are far as I can tell and it is only a relatively small aberration but still worth chasing down if it is possible to address. It is looking more to me like a combination of some small tilt and collimation issues. Now that I have CCD Inspector I can do some more objective tests to see.
And on that note, I would like to ask a specific question. When I run CCDI it shows a tilt of 9% - 10% and a collimation error of around 4". Are these sorts of figures indicative of underlying problems that need to be chased down or are they small enough that you wouldn't normally bother? Is there a benchmark for this sort of testing? Sorry for asking basic questions but I am not that familiar with using CCDI.
Something worth trying the next night you can is when you cross meridian rotate the camera 180°. This will keep the subject in the correct orientation and may help narrow things down.
I've conducted a rotation test by taking a 60 second image, rotating the camera about 90d and then taking another 60 second image. Guiding was operating fine at the time.
Attached is a zoomed in annotated screenshot of the stars in the centre plus a copy of the CCDI curve plot. Whilst the elongation in the stars is only small it is definitely there and definitely moves with the rotation. This indicates that the issue lies with the OTA which is what I was suspecting.
So the main question here is whether I will really need to do something about it as the aberrations may be such that they don't show up in the final processing. I'm going to finish the data capture on my current object and see how it lands.
I had a chance to blink through your images in CCDStack, I see what you mean. So, all you did between these two image is a meridian flip? To me, image ....23456 is much softer than the other, almost out of focus I would say, but I see there is only minutes between the images, same temperature and the same focus position. But, something has to be moving. If the orientation of the camera to the OTA has not changed between images, then that says to me, there is some flexure somewhere.
All stars in the image are affected by it, but it seems the aligning algorithm has chosen the smaller stars to align on, making it look like the larger ones are moving.
Without shifting anything, can you take 2 or 3 images on a row of, say 60 sec exposure or less, pointing more or less east about 45 deg altitude, then flip the meridian and do the same pointing west, don't rotate the camera. Use the RED filter. do you notice any elongation of the stars or shifting focus, and if so, is the orientation of the elongation different in each side of the meridian with respect to everything that is in front of the camera?
In your last post, you rotated 90 deg, is the rotator after the focuser? if it is, when you say the cause could be with the OTA, that would also include focuser sag if there is any.