[RCC] Seeking critique of star roundness...

13 Nov, 2017 20:14
Hello. First of all, I have not a clue what I am doing. I feel like a dust mote floating in a sea of overwhelming information!

I recently upgraded my imaging system to include an Astro-Physics AP1100GTO with absolute encoders. I replaced the entire focuser on my Takahashi FSQ-106EDXIII with a Moonlight WR-35. It is very robust. At the same time I added a Takahashi QE .73x focal reducer. I verified the math on the backspacing of the latter, and had several other imagers confirm I had the correct back focus distance. However, it turns out one piece of my imaging train between the QE and the CCD chip (an adapter ring) is 0.65mm too long.  I do not know if this is critical enough to be an issue. But I digress.

When I first installed the new equipment, I had a lot of very odd shaped stars, primarily on one corner of the field of view (the lower left). I subsequently installed a camera plane adjuster between the WR-35 focuser and the QE focal reducer in an attempt to make sure the chip was perfectly parallel to the scope. There is no detectable flexure in my imaging chain.

Attached is a test image I shot recently. It is five 600s guided images combined in PixInsight, with just histogram transformation. When I performed BLINK I was amazed that there did not appear to be any field drift. All five images were perfectly aligned. I did perform a StarAlignment. However, I am not impressed with the roundness of the stars themselves.  There still seems to be corner issues. Only one focus routine was made prior to the five images being taken. I am not so much concerned about the star bloat as I am the out-of-roundness. And please do not consider the lack of data or the prettiness of this series. It was just to check for optical errors. I'd be very interested in ways to statistically check the roundness of stars using either Maxim or PixInsight!

  M31 Test

I do not know if I am looking at guiding errors (the guiding seemed OK as far as the MaximDL graph showed; 5s, aggressiveness 6) or something more sinister like an out-of-collimation issue regarding the chip. Or an optical issue with the reducer. I'm in a dome so no wind. Great balance. Looking back at my first attempts at imaging earlier in the year without the reducer showed similar issues with stars being larger in one corner. So I wonder if perhaps there has been an issue all along and I just didn't see it or notice. Perhaps it isn't the reducer but that the chip is tilted relative to the plane of focus? I am still messing with the camera plane adapter, but unsure how to proceed. However, things seemed to get better upon adding it (perhaps quite by chance). Maybe it's poor polar alignment or a tracking issue I am seeing?

Maybe I'm being too critical in what the system can do? I welcome all honest critiques of either my methods or machinery! I did upload the raw .FIT file but am not clear on how to share this.

Kind regards, Chris
Edited 13 Nov, 2017 21:08
10 Mar, 2018 12:26
Hi Chris.

I'm pretty new to these forums, but noticed that your post has not yet been replied to. I expect that's because your image actually looks pretty good, especially near the centre of the field. There may be a little bit of drift or tilt/flexure in the imaging train … but only minor. Since the slight imperfections seems to be mostly constrained to the top-left and bottom-right, I suspect tilt or flexure is at play. Are all of the spacers/adaptors between scope and sensor threaded, and is your focuser well adjusted and tilt free?

Would be interested to hear whether you have it fully resolved.

Edited 11 Mar, 2018 23:32
10 Mar, 2018 17:08
First, your image is pretty good.  I see the issue you describe if I zoom all the way in.  I think it’s a guiding error that is compounded when stacking frames.  Check your polar alignment.  Also, dither frames , .5 in Y axis and see if that helps.  Otherwise, you’re  on the right track.
10 Mar, 2018 18:30

While I don't have all of the answers I might have a few.  Statistically determining the roundness of stars in PI is very easy.  Run the Sub-frame selector script on your subs after you have calibrated them.  There is a roundness number assigned to each.  It is not a percentage of "out of roundness" like CCDinspector provides but it does sort the subs by roundness.  I do not know the math on the PI statistic but I believe the  ~.6 range is around 20% out of round.  That's the threshold I'm using to toss subs.

You should evaluate this issue by looking at individual sub-exposures rather than a stacked result.  The stacking only hides potential issues.

As mentioned, it is critical that you dither (randomly change the framing by a few pixels) each sub.  It is a bad thing to have them line up perfectly before registration because you will not be able to remove CCD artifacts, gamma ray hits, airplane trails, etc. in the stacking process.

There is something going on with the stars being fuzzy on one side and hard on the other.  But we would need to see the individual subs to start guessing.


10 Mar, 2018 18:37

From my experience of the Tak reducer (on both my FSQ85 and FSQ106) the 0.65mm will make a difference.  Also the three set screws on the reducer are notoriously difficult to set just right.  Also, depending on the temperature change during your lights focus could drift compounding any other issues affecting your corners as the FSQ106 does need regular focusing with temp change - I trigger an autofocus after 0.7 deg C change in temp.

I presume your test image is luminance and you have focused using luminance rather than a filter offset?  I ask, just in case you have used a filter offset method as this could be another source of star elongation due to an incorrect offset magnitude.  Just trying to be thorough.

However, despite saying the above, your corners are not too far off at all.

As a first measure you may simply want to persevere with tiny adjustments of the three set screws on the reducer.  If you are able to adjust your metal-to-metal reducer spacing as well, I would also do that (don't forget to factor in one third filter thickness additional spacing, 72.2mm + one-third thickness).


Register or login to create to post a reply.