Newt Astigmatism: tilt or secondary mirror?

First of all, it's coma, not astigmatism.
Secondly, it is most likely due to tilt in the corrector.
Thirdly, I'm not sure the APM is that good in the corners and frankly I wouldn't use it.
Fourthly, if you want to investigate whether you have an astigmatic flat you need to assess it visually without correctors of sorts, or with a sensor at the center of the image, inside and outside focus.
Furthermore, collimation's final stages are best done visually with a high power eyepiece and never with the image train in-situ. On top of that the best collimator are the autocollimators. If you are serious about your newt's collimation get one and ditch the tri-bahtinov mask (not sensitve enough).
Finally, get a Gerd Neumann TCU to correct for slop/tilt in the image train: https://www.gerdneumann.net/english/astrofotografie-parts-astrophotography/ctu-camera-tilting-unit.html

Astigmatic 5th order coma behave like that and it is a by-product of the correctors with paraboloidal mirror (it wouldn't appear so with hyperboloidal mirrors). At any rate the shot without any corrector reveals tilt and consequential mis-collimation.

To test against astigmatism you need to look at the inside-outside Fresnel ring when dead-center otherwise coma very promptly takes over and masks the issue.

I can swear about autocollimators and never let me down if the center spot is properly marked (with a bright reflective marker, if possible). The barlowed laser is an useful adjunct and is both insensitive to tilt and decentering and will get you pretty close to dead-on collimation but I found it diffcult (ney, impossible) to obtain here were it also priced affordably which it isn't. I have the Moonlite focusers (old ones) and never once failed to produce in-focus perfect Airy pattern if seeing cooperates. I like the immediacy of the autocollimator and the fact that it can be used in daylight which spares me from wasting rare good nights.

I also have good center dot templates that can be used to check whether the center of the mirror has been marked properly. The point to final in-focus collimation is that it is insensitive to the minutiae of centering and delivers there and then, try that with a Roddier wavefront analyser. Besides, if using that software you need pay more than the actual autocollimator, considering the need for a AI-driven model for you actual reflector.

I can swear about autocollimators and never let me down if the center spot is properly marked (with a bright reflective marker, if possible). The barlowed laser is an useful adjunct and is both insensitive to tilt and decentering and will get you pretty close to dead-on collimation but I found it diffcult (ney, impossible) to obtain here were it also priced affordably which it isn't. I have the Moonlite focusers (old ones) and never once failed to produce in-focus perfect Airy pattern if seeing cooperates. I like the immediacy of the autocollimator and the fact that it can be used in daylight which spares me from wasting rare good nights.

Thanks so much Andrea. So just to make sure I understood you correctly: I have both coma due to miscollimation and tilt causing the issues seen in the images?
Are you saying that fine tuning with an autocollimator would be recommended here or would a start test be preferable for this level of tuning? A slight defocus start test is a little difficult to judge correctly for me given the secondary offset, when defocusing by a few waves as recommended in most of literature, I can’t tell if the central shadow offset is due to actual mirror offset or mis-collimation. 

So that’s why I built a single sector tri-bahtinov mask with the slits moved all the way to the perimeter of the aperture to maximize sensitivity. After checking collimation this ways I usually defocus slightly and check for even illumination around the defocused star shapes. 

To your point about immediacy, I also like a camera based solution better because I can be at the back of the telescope adjusting the primary and not have to go back and forth between eyepiece and back of the telescope. Not to mention having to take the imaging train in and out each time (also my newt is setup with the focuser pointing down when in zero position). Am I thinking about this the wrong way?

Do you think I should just get the catseye autocollimator or do you think that it won’t make much of a difference given perhaps lower grade coma corrector and tilt issues? Do you have any suggestions on how to fine tune the collimation using a star test while i wait for the autocollimator?

Thanks so much Andrea. So just to make sure I understood you correctly: I have both coma due to miscollimation and tilt causing the issues seen in the images?
Are you saying that fine tuning with an autocollimator would be recommended here or would a start test be preferable for this level of tuning? A slight defocus start test is a little difficult to judge correctly for me given the secondary offset, when defocusing by a few waves as recommended in most of literature, I can’t tell if the central shadow offset is due to actual mirror offset or mis-collimation.

So that’s why I built a single sector tri-bahtinov mask with the slits moved all the way to the perimeter of the aperture to maximize sensitivity. After checking collimation this ways I usually defocus slightly and check for even illumination around the defocused star shapes.

To your point about immediacy, I also like a camera based solution better because I can be at the back of the telescope adjusting the primary and not have to go back and forth between eyepiece and back of the telescope. Not to mention having to take the imaging train in and out each time (also my newt is setup with the focuser pointing down when in zero position). Am I thinking about this the wrong way?

Do you think I should just get the catseye autocollimator or do you think that it won’t make much of a difference given perhaps lower grade coma corrector and tilt issues? Do you have any suggestions on how to fine tune the collimation using a star test while i wait for the autocollimator?

With my 6" and 8" newt I just do the autocollimator  since the ceenter mask is very bright I can see up to 4 reflections. The 12" is quite dusky in the center because I had to take out the older center mark (as it was off-center) so reflections aren't as bright and I may manage 3 in the best of times. So I'd normally switch to EP between 250x and 300x and continue collimation with a start test. Now here the star test is carried out with just a tweak outside focus, just enough to reveal 1 maybe 2 rings (depending on the seeing) untill I feel I'm pretty simmetrical at the center of the FOV. If the seeing is good enough I then move on to in-focus checks and further tweaks to get a nice Airy pattern. My 12" focuser points inward directed toward the RA axis so where it points depends on the position but I get that it is akward going to and fro. Yet I rarely need to do more than 1 collimation per season/year so fair play. 

Cat's eye autocollimation and sight tubes are n.1 so I'd get one asap. This said, I feel the main issue here is tilt so I'd try to address that with a CTU. That's money well spent. And I'd star working with the GPU and when that's done move on to longer FLs.

I suspect that collimation is close but not perfect.  The only way to get there (imo) is with a star test .  When I do that - even crudely  -I generally get eccentricity down to 0.45 or better

Or could always just correct it all out using RC Blur Exterminator

Thanks Tim, but average eccentricity in PI for these images is definitely is less than .45, I think the reason you are seeing such big numbers is because the image is highly compressed and resampled by 50%. Also hocus focus reports similar eccentric and FWHM numbers live while taking the images.

If it wasn’t for BXR, I would have thrown this scope and quit the hobby a long time ago because of this scope 😂 I’m just trying to get the image as perfect as possible because I feel bad that my tiny 80mm refractor can achieve perfect star shapes  while the large 10” can’t. 

how do you do your collimation and what coma corrector do you use?

Ah I see -- I did wonder.

I am guessing that he image scale with the 80 mm is much coarser - so it is easier to get round stars - and its a refractor and all..F4 reflectors are difficult --much more so than F5s.

Now -- just in the last 2 weeks - I finally have a Howie Glatter laser and a tublug so I am fortunate.   Even so I still look at out of focus star images and make adjustments.  I use the SW aplanatic F 4 coma corrector.

But to be honest - the main use of my F4 12 inch scope is to look more closely at small (angular size) objects in more detail so when capturing in Sharpcap I use a reduced frame size but highly sampled (0.41 arcsec/ pixel)  - so not usually too bothered about the edges of the largest possible frames (where any residual coma would be greatest).  I  also tend to stack lots of short high gain subs - in accord with Robin Glover's calculations - in order to provide the maximum information at 0.41 for for BlurXt to work with.

What seemed surprising to me is that following BlurXt correction and NN deconvolution you can get great final images the quality of which depends on the SNR and FWHM  of the starting image (as you would expect)  but to not depend very much on the Eccentricity of the original image.   In other words the correction algorithm in BlurXt is so smart that it makes me wonder how much mileage there really is in being too fussy about the alignment.  If everyone starts to use this sort of software then it will signal a paradigm shift perhaps ?

It is obviously difficult to make well controlled comparisons (starting SNR different etc) but for example - over a few nights ....But the impression I get is of equally good end points from starting points that vary quite a bit (from ca. 0.69 down to 0.45) in average Eccentricity values.

so this one starting at E0.69 from 507  x 10s frames





ends up in about the same place (or actually rather better because of its better SNR ?) in terms of final image quality  than the following from 360 x 10s frames  averaging at about E0.44 .  Note star sharpening was set to zero in both cases.




Maybe I did waste money on that collimation equipment  after all ?!?

Tim

Tim Hawkes:

I suspect that collimation is close but not perfect.  The only way to get there (imo) is with a star test .  When I do that - even crudely  -I generally get eccentricity down to 0.45 or better

Or could always just correct it all out using RC Blur Exterminator

Thanks Tim, but average eccentricity in PI for these images is definitely is less than .45, I think the reason you are seeing such big numbers is because the image is highly compressed and resampled by 50%. Also hocus focus reports similar eccentric and FWHM numbers live while taking the images.

If it wasn’t for BXR, I would have thrown this scope and quit the hobby a long time ago because of this scope 😂 I’m just trying to get the image as perfect as possible because I feel bad that my tiny 80mm refractor can achieve perfect star shapes  while the large 10” can’t. 

how do you do your collimation and what coma corrector do you use?

Ah I see -- I did wonder.

I am guessing that he image scale with the 80 mm is much coarser - so it is easier to get round stars - and its a refractor and all..F4 reflectors are difficult --much more so than F5s.

Now -- just in the last 2 weeks - I finally have a Howie Glatter laser and a tublug so I am fortunate.   Even so I still look at out of focus star images and make adjustments.  I use the SW aplanatic F 4 coma corrector.

But to be honest - the main use of my F4 12 inch scope is to look more closely at small (angular size) objects in more detail so when capturing in Sharpcap I use a reduced frame size but highly sampled (0.41 arcsec/ pixel)  - so not usually too bothered about the edges of the largest possible frames (where any residual coma would be greatest).  I  also tend to stack lots of short high gain subs - in accord with Robin Glover's calculations - in order to provide the maximum information at 0.41 for for BlurXt to work with.

What seemed surprising to me is that following BlurXt correction and NN deconvolution you can get great final images the quality of which depends on the SNR and FWHM  of the starting image (as you would expect)  but to not depend very much on the Eccentricity of the original image.   In other words the correction algorithm in BlurXt is so smart that it makes me wonder how much mileage there really is in being too fussy about the alignment.  If everyone starts to use this sort of software then it will signal a paradigm shift perhaps ?

It is obviously difficult to make well controlled comparisons (starting SNR different etc) but for example - over a few nights ....But the impression I get is of equally good end points from starting points that vary quite a bit (from ca. 0.69 down to 0.45) in average Eccentricity values.

so this one starting at E0.69 from 507  x 10s frames





ends up in about the same place (or actually rather better because of its better SNR ?) in terms of final image quality  than the following from 360 x 10s frames  averaging at about E0.44 .  Note star sharpening was set to zero in both cases.




Maybe I did waste money on that collimation equipment  after all ?!?

Tim

Tim,
When you see elongated stars on axis, you’ve most likely got one of three issues:
1) A tracking error.  Check to see if the elongation is aligned with RA and if it is, that’s a likely problem.
2) Vibration.   Was there wind and how stable is you system?
3) A problem with your secondary.

Remember that it only takes a very tiny amount of power on a flat secondary to introduce significant astigmatic errors into the wavefront.  That’s why it is very important to have a very flat secondary used at 45 degrees on a fast Newtonian.  Power can come from the way the optic was figured but it can also come from the way that it’s mounted.  So if you can’t solve the problem with on-axis elongated stars, it might be worth checking how the secondary is mounted.  It shouldn’t be able to move but it cannot be stressed by the way that it is mounted.  While it’s possible to have an astigmatic or warped primary mirror, the secondary is probably the first thing to check.

Yes, BXT can fix this kind of stuff, but you’ll get even better results if you start with clean data.

John

John Hayes:
The corners are definitely not dominated by astigmatism.  You’ve got a LOT of classic 3rd order field coma in this image.  It appears to be fairly well balanced in the field, which indicates a problem with how your coma corrector is set up.  I agree with Andrea that you should ditch the TB-mask.  SKW is a much better way to balance the field but it won’t tell you much about how to adjust the optics.  However, it will be very useful to tell whether an adjustment makes things better or worse, which is really helpful for dialing out the coma.

John

If you are referring to the image with the bright central star, that was taken without any correctors so I guess the coma present is “normal”. Before stretching the image, the info is central star looks round, at last given the seeing conditions.

I really appreciate your comment about SKW, made me feel a little better about the investment

andrea tasselli:
Most emphatically BXT does NOT eliminate the massive reduction in MTF due to severe coma (or even astigmatic coma) not does serve as a substitute to the actual collimation of reflectors in general. While it does a darn good job in rounding the off-center distortions isn't really the same as if there were none.

By the way, did you but a PPU model or a permanent one?

I appreciate this comment very much, Russ said the exact same thing on many of his presentations on BXT…no substitute for good inputs that can be further improved using deconvolution or otherwise. That’s just a fact of signal processing

I ended up getting the Permanent model. After looking at the catseye bundle with shipping to Canada and import taxes, the permanent model ended up being a little cheaper. I also have to collimate the telescope before each use because I disassemble and move the OTA and imaging setup indoors after each night so the PPU would not have been cost effective in my case.

not sure why would you do a coarse collimation with the OCAL when it's by far the most precise collimation tool you can use. I have never managed to get as precise collimation with any other tool.

John Hayes:

Tim Hawkes:

I suspect that collimation is close but not perfect.  The only way to get there (imo) is with a star test .  When I do that - even crudely  -I generally get eccentricity down to 0.45 or better

Or could always just correct it all out using RC Blur Exterminator

Thanks Tim, but average eccentricity in PI for these images is definitely is less than .45, I think the reason you are seeing such big numbers is because the image is highly compressed and resampled by 50%. Also hocus focus reports similar eccentric and FWHM numbers live while taking the images.

If it wasn’t for BXR, I would have thrown this scope and quit the hobby a long time ago because of this scope 😂 I’m just trying to get the image as perfect as possible because I feel bad that my tiny 80mm refractor can achieve perfect star shapes  while the large 10” can’t. 

how do you do your collimation and what coma corrector do you use?

Ah I see -- I did wonder.

I am guessing that he image scale with the 80 mm is much coarser - so it is easier to get round stars - and its a refractor and all..F4 reflectors are difficult --much more so than F5s.

Now -- just in the last 2 weeks - I finally have a Howie Glatter laser and a tublug so I am fortunate.   Even so I still look at out of focus star images and make adjustments.  I use the SW aplanatic F 4 coma corrector.

But to be honest - the main use of my F4 12 inch scope is to look more closely at small (angular size) objects in more detail so when capturing in Sharpcap I use a reduced frame size but highly sampled (0.41 arcsec/ pixel)  - so not usually too bothered about the edges of the largest possible frames (where any residual coma would be greatest).  I  also tend to stack lots of short high gain subs - in accord with Robin Glover's calculations - in order to provide the maximum information at 0.41 for for BlurXt to work with.

What seemed surprising to me is that following BlurXt correction and NN deconvolution you can get great final images the quality of which depends on the SNR and FWHM  of the starting image (as you would expect)  but to not depend very much on the Eccentricity of the original image.   In other words the correction algorithm in BlurXt is so smart that it makes me wonder how much mileage there really is in being too fussy about the alignment.  If everyone starts to use this sort of software then it will signal a paradigm shift perhaps ?

It is obviously difficult to make well controlled comparisons (starting SNR different etc) but for example - over a few nights ....But the impression I get is of equally good end points from starting points that vary quite a bit (from ca. 0.69 down to 0.45) in average Eccentricity values.

so this one starting at E0.69 from 507  x 10s frames





ends up in about the same place (or actually rather better because of its better SNR ?) in terms of final image quality  than the following from 360 x 10s frames  averaging at about E0.44 .  Note star sharpening was set to zero in both cases.




Maybe I did waste money on that collimation equipment  after all ?!?

Tim

Tim,
When you see elongated stars on axis, you’ve most likely got one of three issues:
1) A tracking error.  Check to see if the elongation is aligned with RA and if it is, that’s a likely problem.
2) Vibration.   Was there wind and how stable is you system?
3) A problem with your secondary.

Remember that it only takes a very tiny amount of power on a flat secondary to introduce significant astigmatic errors into the wavefront.  That’s why it is very important to have a very flat secondary used at 45 degrees on a fast Newtonian.  Power can come from the way the optic was figured but it can also come from the way that it’s mounted.  So if you can’t solve the problem with on-axis elongated stars, it might be worth checking how the secondary is mounted.  It shouldn’t be able to move but it cannot be stressed by the way that it is mounted.  While it’s possible to have an astigmatic or warped primary mirror, the secondary is probably the first thing to check.

Yes, BXT can fix this kind of stuff, but you’ll get even better results if you start with clean data.

John

Thanks John.  You are exactly correct - as usual I have to say :-)  - and I know already that a large part of the problem is 1).  2) comes and goes on different nights. Elongation is indeed always close to the RA axis and is in the same direction whichever of my two Newts I mount.  Which points to the mount.   I determined a couple of years ago that while good for most tasks at not such exacting image scales as 0.406 it is a fact that my CEM 70 'jitters' in the RA axis.  So while everything looks fine and dandy on the 3s PHD2 guiding scale there is a subtle oscillation of about 2-3.5 px  going on in RA  on about the 1s scale which scatters light along that axis direction.- and so stacking 200ms subs produced round stars whereas stacking 2s subs didn't.  I did a simulation to show that the jitter could excatly account for the Eccentricity I was seeing.  On that basis I convinced FLO, my supplier,  to replace the original mount - which they kindly did.  The new version was better but still imperfect. -- so now I think my 'best' is about E = 0.45.   I did a detailed analysis and produced a short slide set on it - if anyone is interested or might have the same problem I could post it up?   It might be useful to someone because it is perhaps one of the less discussed issues around 'star shape'.

The problem really is how far do I want to take this ?  The OO UK telescope is relatively cheap (compared to other designs) and the only 12 inch F4 also  light enough to really work on a CEM 70.  It has some issues but it's really not at all bad for the money and it is the devil I know that I can tune up well.    Ditto do I really want to invest in a higher quality mount at 2-3X the cost to solve something that - in my book anyway - is a relatively minor problem that is at least adequately if not perfectly fixed by software?  At this point the whole debate becomes   philosophical and financial rather than technical and I guess that I am happy  - and actually quite amazed to get get images from my B7 backyard  of galaxies like M51 and M63 wherein it is possible to tick off and at least recognise most all of the features in NASA/ESA  HST images.  That's really not bad ..

But thanks for the tip also on 3).  Having said all the above there is certainly something satisfying about improving whatever you can and so I will also have a good look at the secondary.  I could well imagine that putting two much force into the 3 alignment or the up/down screw could do something nasty in terms of warping the mirror - so I will have a go at easing off the pressure a little.

On BXT -- I suspect that you are right that better images in = better images out.  Intuitively it seems like common sense.  But as yet I have no evidence  that Eccentricity is actually making a difference to final image quality that I can perceive across my image set -- but you would think that it should  and maybe I just haven't hit the quality level where it matters -- o I will keep trying to get a better quality comparison on that.
Tim

Hello,
I'm struggling to get good star shapes across images from my ONTC 10" F4, specifically around the corners. I tried everything and can't seem to locate/confirm the source of my problems.  Some information about the setup: 

General Setup and imaging conditions: ONTC 10" F4 with 80mm secondary, Evoguide 50DX guide scope with ASI178MM, CEM70 Mount on a tri-pier 360...2-2.5 arcsec average seeing, .5-.4 arcsec RMS guiding error
Imaging setup: Risingcam IMX571 mono, askar back focus adjuster, APM 1.5x coma corrector(transforming the F4 to an F6 @ 1500mm FL), Antlia LRBG-HSO 3nm filters, Moonlite CRL2.5 focuser.  

Imaging and telescope optical alignment procedure:

• Pre imaging session: Coarse collimation using OCAL 3.0 to center/align secondary mirror with focuser, align primary mirror and secondary mirror reflections
• At the start of imaging session: Center on a bright start, start a loop of exposures in NINA and use a tribahtinov mask to fine tune alignment of primary mirror
• Backfocus: Start with ideal back focus for the APM corrector (95mm), observe star shapes and adjust the back focus adjuster in or out to minimize radial or tangential star elongation around the edges of the image.

Below are the results of following the above procedures from a recent session of M101 (Luminance filter, uncalibrated, stretched, 50% resampled, and exported to jpeg)
  
Corners and a crop of the centre:

   
The top right and bottom left corner show signs of astigmatism, while the other edges show some stretching in various directions. The centre of the image has minimal artifacts but in no way perfect. Generally, the aberrations are most noticeable with the L/G/B filters, less so with R and H-Alpha filters. 

Based on the above, I don't think this is a collimation issue, at least based on the shapes of the central stars and the tri-bahtinov mask results. NINA's Hocus Focus Aberration Inspector reports good curvature results (especially after the BF adjustment procedure) but reports tilt beyond critical focus tolerance (about 0.1-0.06 degrees consistently regardless of telescope orientation). I bench tested my camera using a laser pointer and a rotary setup and can confirm there's a very slight tilt...it's there but barely noticeable.  Reflecting the laser off the sensor to a screen at a distance of about 2ft/60cm while rotating the camera, I see the sensor reflection tracing a circle about 1/16in or  1-2 mm in diameter. The risingcam imx571 doesn't have tilt adjusters so I'll have to buy to fabricate a solution to eliminate the tilt...I'm just not sure if such little tilt can cause these issues. 

Is the tilt of the sensor relative to the corrector causing the corner astigmatism and deformed start shapes in general? I read that astigmatism in Newtonians can be caused using badly shaped primary or secondary mirrors, more likely due to a bad secondary. I also read that some of the secondary mirrors used on the ONTC telescopes suffered from astigmatism. 

I'm not sure if it's title, a bad secondary, or something else with alignment or otherwise. 

 I would really appreciate any advice...this issue is driving me insane! 

Those cross shaped stars are astigmatism, and it is not caused by the tiny amount of tilt you have, when I checked my camera with a laser, the circle drawn by the reflected laser was approx 7mm, and that was not that noticeable in the final images, my astigmatism was caused by 2 things, the first being misaligned optics, basically the flattener was out of collimation with the main lens, and mine was an old refractor scope and it did not like the modern small pixel cameras at all, as they showed every little optical error.
so in the case of your scope I would say it’s deffo optical and not tilt…

Some quick updates from last night’s session with SKW:
- Switched to the 1x TSGPU coma corrector
-Tuned out all residual on axis coma using SKW. The process was a breeze compared to anything else I tried 
-Star shapes improved overall but still have squarish or elongated stars near the edges 
-SKW reports severe spherical aberration across the image especially away from the center in the top to bottom direction. Spherical aberration moves from extra-focal to intra-focal from top to bottom.
-Also SKW shows that coma tends to increase off axis but not symmetrically, the pattern of off axis coma seems to be linear top to bottom and not radial. However edge stars still have worse coma even if they are inline with the center of the center in the top to bottom direction. 
-Hocus focus aberration inspector reports a huge .3 deg tilt top to bottom which seems to agree with SKW’s off axis coma and spherical aberration results. If I am to believe hocus focus, I basically need to raise the top of my sensor by 60-80um.

quick conclusion: I believe my camera/EFW combination are tilted relative to the coma corrector mounting point…the camera sensor relative to the mounting flange might not be that bad as shown through my laser test. But the combination of the EFW and camera relative to the mounting point of the corrector are not parallel. In SKW, I could see that I can improve coma/astigmatism in one part of the image and immediately other parts below and above it get worse and in the direction of the supposed tilt…not so much left and right. For example picking a start towards the top or bottom of the image and using that to tune coma/astigmatism/spherical moves the overall aberration pattern in the opposite direction of the chosen star. Hard to tell if the optics of the TSGPU coma corrector are also just not that great and are simply not able to correct for all off axis coma

@andrea tasselli  your suspicions of residual on axis coma and overall tilt seem to be spot on. In retrospect, I should have tested without the TSGPU in place to confirm that coma increases radially while spherical aberrations change in the direction of the supposed tilt…or perhaps I would have seen the apparent tilt disappear with out the corrector since I would have aligned the mirrors purely relative to the sensor plane without having to fight against the effects of the coma corrector. 


I’ll post more details including images when I get the chance.