Bahtinov masks, AF units and when is focus actually focused?

[Craig_]

I pulled out the Esprit 80 for some imaging recently, the first time in a long time since I have used it. I installed a ZWO EAF, and the night started off well pre-meridian flip. As I had not run any AF routine on the scope before, done any backlash calculations, etc, I chose to simply focus manually (electronic control) with a bahtinov mask. I started out imaging in Ha, and my stars were very small and tight indeed. Median FWHM about 1.8px.

After a couple of hours, I needed to meridian flip and refocused at the same time, also switching to Sii. I once again used the bahtinov mask (on a bright star - Rigil Kent) but my star sizes were way higher than pre-flip, eg going from FWHM of ~1.8px on Ha (Chroma 5nm) to more like ~2.5px (or higher) on Sii (Chroma 5nm). This was immediately after refocusing, so I refocused again, same result. Tried again, same result. Ran the AF routine, it was way off (more on that later.)

Gave up and switched to Oiii, once again focusing manually with a mask on Rigil Kent. FWHM down to ~1.7px (Chroma 3nm.) Imaged that filter for a few hours, went back to Sii, again, measurably larger stars than Oiii although better than earlier, maybe 2.2px FWHM or so. I gave up stuffing around with it and just imaged in Sii the rest of the night.

Next clear night, I started on Sii. Once again used the focusing mask, got slightly better results than the prior night on Sii but stars were still bigger than I was getting that prior night on Ha/Oiii (of course, atmospheric conditions change night to night so I didn't stress about this figuring that could be the cause.)

Later I refocused, again with the mask, and FWHM shot up to like 2.5px or so. I decided to try something different, and using the focus tool in the ASIAir (not autofocus) I focused manually, in one direction only, to counter backlash. I would increment my focus bit by bit (step size of 10) watching the HFD star size graph trend, and when it started measurably trending up I knew I had gone too far, and then reversed. Eventually I got to a point where I was sure it was the best focus I could get on that night, and indeed, after taking a few subs my median FWHM was markedly better than it had been all night on Sii. Super confused at this point, I slewed back to Rigil Kent and put the mask on to see how it looked and the diffraction spikes were visibly misaligned. Like, so badly misaligned there's no possible way you could mistake it as being in focus (if the mask is to be believed.) Going by the mask, I was way out of focus and yet average star size was much, much better than I got using the mask.

I've never had an issue with mask focusing before. It's been so long since I used the Esprit 80 that it's possible it's always had this issue with the mask but I've just not picked up on it as I was less discerning of star size back then. But on my Esprit 120 for example, when the bahtinov mask (a different one given different aperture sizes) reports critical focus, my autofocuser agrees. Not so with the Esprit 80! On filter changes on the 2nd night I also noticed the mask was not always accurate with Ha/Oiii either. Not sure why it was on the first night, possible coincidence.

Any theories what's going on here? Why can't I trust the bahtinov mask on the Esprit 80? I understand different filters will result in different star sizes, but at one point or another on all three filters I was finding much better results by ignoring the mask, while at other times the mask was accurate - always focusing on Rigil Kent, never any other star.

Any theories here?

My other issue, probably unrelated, is I have struggled using the ZWO EAF on the Esprit 80 with the ASIAir. It works nicely on the 120, but with the 80 it has yet to run an AF routine that was actually properly focused - it's usually been way off. I've noticed quite a lot of backlash with this combo, I calculated 40-50 steps of backlash, but even adding in that compensation made no difference (actually made it worse, if anything.) Anyone using this combo got any ideas for how to improve?

I'm really happy with the final image, the focus issues were a bit annoying but not insurmountable as I just had to closely monitor my star size and ensure focus was always accurate. But right now I certainly could not trust the autofocus routine with this combo, and need to manually intervene. Final image is attached - 10 hrs on NGC6188.

Cheers

[ChrisD]

is it an earlier Esprit 80 with the crayford or the version with the rack and pinion focuser?

When not using a filter does the bahtinov mask agree with your manual focus?

Chris

[Craig_]

It’s the rack and pinion focuser. I haven’t tried focusing without a filter entirely, this would not be easily done as each position in my EFW is occupied.

[kosborn]

Hi Craig,


I can't really explain your AF problems but it would seem to me that it's better to trust the Bahtinov mask (despite what the numbers might suggest). I don't know the ASIair device but there are a couple of things to consider for autofocus routines. When using NB filters it's important to have a long enough exposure. The other things to consider are number of autofocus steps and step size. Although this video (https://www.youtube.com/watch?v=s8xc1DBbHT0) talks about SGPro autofocus, the theory is the same for all apps and might help.



Kevin

[Craig_]

Hi Kevin,

Even ignoring software analysis of FWHM, doing a flip-between at 100% of the image focused with the mask vs without the mask, there was still a noticeable difference in star size. Not like one was clearly out of focus, but enough of a difference you could tell them apart. My eye agreed with what the software reported on star size.

Thanks for the video link, I will check it out.

[ChrisD]

The older crayford can tend to have a little free play as they age. You should be ok though with a rack and pinion.

With the ZWO EAF on the Asiair, make sure the focus process is completing with the focuser moving in (against gravity). Visually check to confirm, if it's not then use the "reverse" switch on the Asiair focuser setting page to fix.

You should find the best autofocus exposure times for each filter. This blog covers autofocus on the Asiair. It from last year but still relevant.
How to autofocus with asiair (https://eastwindastro.blogspot.com/2021/04/how-to-autofocus-with-asiair.html)

In a previous post I suggested using no filter. Try setting up the autofocus for a Luminance filter (direction, exposure, and step size) and confirm the Bahtinov mask agrees. Then try another filter and find the correct exposure time for autofocus. Set it up one filter at a time.

Chris

[Craig_]

Thanks Chris, some stuff for me to try there :)

I’m out in the field now actually, first sub I took was mask focused on Ha, was extremely particular to ensure it was perfectly focused (according to the mask). FWHM on Ha filter of 2.62. Refocused myself manually using the ASIAir focus tool (not AF, just where it graphs HFD and you can tell if it’s focused based on the graph trend.) First and second sub after doing that - FWHM of 2.19…

Might try a different mask too.

[Stefan Buda]

Craig,

Have you tried rotating the Bahtinov mask 90 degrees after achieving focus to see if in the new orientation there is no focusing error visible?

Bahtinov masks are sensitive to residual astigmatism, if there is any in your optics, so a particular orientation of the mask may not achieve the best focus possible.

[Craig_]

I have made small rotations, with no visible difference, although this was moreso to try and centre the mask in case that was the issue. My optics are pretty good on the Esprit 80, so not sure that will be the issue but happy to try it next clear night as I'll try anything :lol: Cheers

[The_bluester]

I was going to add (Though I don't focus with a mask and never dived in to them enough to know what Stefan wrote about them) was that I would trust measured HFR more than induced diffraction spikes. Smallest stars = best focus.

Very few auto focus routines use a mask, they almost exclusively measure the HFR of a star or multiple stars in the field to determine the best focus point. The only focus assist I have seen use a mask is in APT (Astro Photography Tool) but when I was using APT I didn't have a mask, so I didn't test that function.

[Stefan Buda]

Sounds like you have a centring issue and that may explain the focusing error.
Do you have a picture of your mask attached to the scope? I'd like to know what's going on because I never experienced this problem.

[Craig_]

Thanks, yes, indeed I trust HFR more than the mask in this case, as the results both to my eye and to software analysis have resulted in clearly smaller stars. It's just bizarre as I don't have the same issue on other telescopes.



I've never been too fussed with making the mask perfectly centred on the dew shield before as it's never been an issue. It's something I tried to eliminate it as the cause here, but with no discernable difference. Unfortunately I don't have any photos of the mask on the telescope, but I do have some ASIAir screenshots of supposed focus with the mask which was proven by star analysis to not be perfectly focused. I have zoomed in on the star in these screenshots.

These were all using the Sii filter. The three Sii screenshots were each captured on three different nights. I have tried different exposure times and binning for the focusing shots with the mask, exposure ranging from 3 to 15s. It hasn't made much difference. Obviously given NB, the signal getting through the mask is quite weak; these were focused on Rigil Kent. Also attached (for reference only) is the same mask, on the same star, with the Lum filter - I did not image in Lum so do not know if this was actually correctly focused, or not. A much stronger signal comes through on Lum.

Maybe you can see something in these that I am missing?

[Stefan Buda]

The second image (the one with different orientation) is noticeably out of focus.
Is this a laser cut mask or 3Dprinted? I wonder whether the mask has some sort of geometrical error that generates a bias.

[Craig_]

Funnily enough if I focus using HFD measurements and go back to Rigil Kent, and put the mask back on, the spikes are even less aligned than that second image. Unfortunately I didn't screenshot that.

It's a mask made of thin metal, so I assume laser cut.

[DarkArts]

First off, like Stefan, I'd guess the mask is mis-shapen, mis-centred or at a slight angle.

What does your Airy disk image look like either side of focus? How round are your stars near the edge of the field and is focus/curvature/aberration symmetrical left-right and up-down?

TBH, though I do have a Bahtinov mask for my ED80, I don't use it except sometimes at the start of the night for initial rough setup. I focus by eye these days and before that (when I used a DSLR) I used BYEOS' FWHM measurement.

[Craig_]

I'm not sure about the airy disk side of things. Have not looked at this, but can check it out next clear night.

Here's a PI AbberationInspector of the frame from the first night of imaging (Ha) where I noticed the issue. As mentioned in my OP, although mask focused I was getting very tight stars with Ha, it wasn't until flipping to Sii that I had focus issues. The next night, all three filters had issues focusing with the mask.

Definitely sounds like a replacement mask is worth a shot, even though to my eye this one seems to be in good nick (not bent or anything.)

[Stefan Buda]

So you are using the two focusing methods in two different orientations of the scope (different parts of the sky). Are you sure there's nothing in the optical train that can shift/move when you go from one part of the sky to the other?

[Craig_]

I don’t think it’s movement when slewing around. In that example I gave, critical focus with HFD was made, then I skewed to Rigil Kent, saw the mask focus was way off, went back to my target and focus was unchanged. So between slewing to, and back from, Rigil Kent, the telescope remained in focus although the check of the mask while on Rigil suggested that (if the mask was to be trusted) it was out of focus, FWHM after returning to my target was excellent.

[Stefan Buda]

Yes, I would expect the scope to be in focus for the orientation where focus was achieved even if you moved away temporarily from that position. The question is whether the scope was still focused in the other orientation where the Bahtinov indicated that it was not.

[Craig_]

Ah, got you. I understand.

Well, on the first night I focused with the mask (on Rigil kent) then slewed to my target, this worked with Ha and Oiii but not Sii. Focus was fine for the first two filters, despite the differing orientation. The 2nd night I had issues with all of them.

Next opportunity I will try HFD focusing near Rigil (or another bright star) then pop the mask on and see if it agrees.

[JohnH]

Ummmm....couple of thoughts...isn't Rigel Kent a triple star and might this not prevent a mask giving a clean pattern?



Second thought : autofocus can give you a full frame optimised focus (depends on your software of course) that will be immune to field curvature issues but if you focus on a single star then you have achieved spot focus only stars in that exact position will be in perfect focus, balanced focus will be slightly inside that position I believe...

[Craig_]

Re: Rigil Kent - not sure, I will try another star next time! Thanks

Re: the AF routine, it isn't coming back optimised for any area of the frame, it's just way off (I'm talking 100+ steps out.)