Experiments with a tri-Bahtinov mask. Input appreciated.

[wimvb]

A while ago, I designed a tri-Bahtinov mask to aid in the collimation of my Mak-Newt. I even cnc-ed the secions of the mask. Since my cnc machine is limited to a 30 x 18 cm work area, and the telescope has a 20 cm tube, I had to make 3 sections of the mask. Before I had an opportunity to assemble the sections, other things got in the way, and I forgot all about it. (I wasn't happy with the final outcome, so forgetting was quite easy, and age helps.)

A week or so ago, I came across the sections and finally glued them together. On a partially clear night, I put the mask on my MN190 and tested on Arcturus. For the first test, I only did a coarse focus and took RGB images. For the second test, I focused better and took an image with the L filter in place. Here are the results.

When I did the second test, I found the mask not that sensitive to defocus, so I wonder how sensitive it will be as a collimation tool. Does anyone here have experience of using such a mask for collimation?

Here is my design. Each third of the mask holds one section of the three zones, as described in this link. The mask is cnc-ed from 2 mm black acrylic. The slots are 4 mm wide and spaced 4 mm.

The sections are glued together with epoxy (butt jointed). I have since designed supports to make the joints more stable, but haven't yet cut those.

https://satakagi.github.io/tribahtinovWebApps/reports/improveSensitivity.html

[almcl]

Hi Wim

I had a go at making a tri-bahtinov out of dark card a couple of years ago and its been kicking around the workshop gathering dust ever since. 

I can't now remember if I got any results with it - apparently I got bored with cutting out the slots and may have tried to use it in its partially cut-out state - but can't find any images of the result.  If I did use it, it would have been on the 190 MN.

But looking at your images, how are they interpreted?  In the 2nd image, I see that the 9 o'clock - 3 o'clock spike is reasonably symmetrical, while the  1 o'clock - 7 o'clock and 11 o'clock - 5 o'clock are much less so.  Does this show a focus problem, a collimation issue or something else altogether?

 

[wimvb]

@almcl With perfect collimation, if you defocus, the central spike of the 11 and 1 o'clock patterns move towards each other, and the 3 o'clock central spike moves up. The 5, 7 and 9 o'clock patterns are the same as those that differ by 6 hours (opposite of the central star). If the 3 o'clock spike moves up, so does the 9 o'clock spike. A tri Bahtinov mask produces three ordinary Bahtinov patterns. The idea is that with miscollimation, the three patterns are focused at different focuser positions. Collimation equalises the focus positions. It's like a three legged stool, where you want the seat to be absolutely level.

What I found in my second test was that I could move focus position quite a distance before I could see a clear shifting of the central spike. That is why I questioned its applicability. Otoh, it may very well be that the coupler between the focus motor and the focuser shaft has come loose during the winter, causing some play in the focuser. The large temperature fluctuations that we've experienced during winter, can do that.

I just put my gear in summer storage until mid August (no astro darkness left after the coming full moon), but I plan to do some more tests when I set up again. I will have to check collimation then anyway, so I will be doing a side by side comparison (barlowed laser/cheshire vs tri-Bahtinov mask) then.

[almcl]

Thanks for that, WIm.  Much appreciated.  

We still have some astro dark left so if the clouds clear I'll use a freshly made mask (with all the slots cut out this time!) and see how my in-focus and defocussed spikes look.  I'm strictly a manual focusser using a conventional Bahtinov and APT's Bahtinov assistant.  My collimation is done by eye with Concenter and Cheshire, so it will be interesting to see what the tri-Bahtinov makes of this.

[david_taurus83]

I got one today in the post. A nice 3D printed tri bahtinov mask from Ebay seller astrobyorlando. Some testing on my C8 Edge tonight. I've never owned an SCT before now but I got damn as near perfect collimation within a few minutes with this mask. I wasn't far out to begin with but it looks like it does the trick? It may be different with a newt as there are different angles involved?

 

[wimvb]

From what I've read/seen so far, I understand that the principle behind the tri-Bahtinov mask is that collimation (tilted mirror) results in a tilted focal plane, which spreads best focus for a star in different directions. But so, in a way, does tilt in the focuser or camera. So I wonder if it's even possible to distinguish one from the other. I will test more when astro darkness returns in August. Atm my telescope is in summer storage.

Edited April 25 by wimvb

[almcl]

Short unexpectedly clear evening gave me a chance to test my 'cut-out-of-thin-card' Tri-Bahtinov mask last evening.

Set up using my normal method of conventional Bahtinov and APT's Bahtinov aid, got us down to a very small error, shown in the first image.

I then substituted the Tri-bahtinov and got image 2. (not quite right but not too bad)

Racked the focusser out about 0.15 mm (approx 10th turn of fine focus knob on the Moonlight focuser) and got image 3.  The spikes have shown some movement.

Then reacked the focuser a similar distance back past the original position to get image 4. Again the spikes show some change. 

Lastly I tried returning to the in-focus point - image 5.

From these images it appears that collimation isn't too far off, although I would appreciate any comments/observations on this.

Two things struck me while doing these tests, one was how much camera tilt we had and how this would affect the result.  (ASTAP gave a figure 8% tilt on an image shortly after the mask was removed.)

The other was how easy adjusting the collimation screws to correct any perceived mis-collimation would be in the dark. Perhaps a Tri-Bahtinov grabber (think I've readof this somwhere on CN) would help.

For the moment I think I'll stick to Concenter and Cheshire in daylight - which doesn't seem to have put collimation too far out?

Edited April 26 by almcl

[wimvb]

@almcl your findings and concerns are as mine. Judging from image 2, I would say that collimation is good, and the concentre/cheshire method of collimation may be just as accurate as tri-Bahtinov collimation. It has the added advantage (for the former) that it can be done at any time, so won't interfere with observing or imaging time. The advantage of the tri-Bahtinov is that it is easy to do with imaging gear in place. This makes it an easier method to check collimation during imaging or during a season. No need to remove the camera, so no need to retake flats.

Edited April 26 by wimvb

[almcl]

Just as you say, Wim.  I wonder if all I would do using the Tri-B mask would be to collimate out the camera tilt which is sometimes quite large according to ASTAP (I mis-typed above - it's ASTAP I use for SQM and tilt).

I found a thread on Cloudy Nights (here) which referenced a Tri-Bahtinov grabber on Github (here) which, although a bit quirky, seems to give understandable results.  

Here's what it made of image 5 in the sequence above:

 

I haven't worked out yet how to relate the channels to collimation screws (there is an assist diagram available but not sure how it works yet). Also not sure if I want to collimate for tilt at set-up every time (= lost imaging time) or whether 'close enough' is sufficient?

Perhaps something to have a play with during the next two months of Astro twilight?

Edited April 26 by almcl

[Vroobel]

Am I right if I assume that any sensor tilt adjustment makes sense only if the camera is permanently attached to a rest of the setup and is never touched, particularly never rotated? Of course, I don't mean a case when the sensor isn't parallel to a flange and it 'accidentally' passed a QC which hopefully doesn't happen to more expensive cameras. 

[almcl]

Not sure of the answer to that, Vroobel, but as regards camera sensor tilt, there's a fairly long thread here which goes into detail about how to measure chip tilt in the camera, independently of scope/lens &c amd also how to correct it.  

Might take a bit of reading as it's grown to 7 pages now,

[wimvb]

8 hours ago, Vroobel said:

Am I right if I assume that any sensor tilt adjustment makes sense only if the camera is permanently attached to a rest of the setup and is never touched, particularly never rotated? Of course, I don't mean a case when the sensor isn't parallel to a flange and it 'accidentally' passed a QC which hopefully doesn't happen to more expensive cameras. 

Everything with two surfaces can have tilt. In theory you'd first make sure that the camera itself doesn't have tilt. The link in @almcl's latest post is a thread here on sgl about a cheap rig and laser for that. ZWO cameras have a tilt adjutment plate. Second is tilt in the focuser. Then misalignment in the optical system. If you think about this too much, your brain will hurt.

Edited April 26 by wimvb

[wimvb]

9 hours ago, almcl said:

Just as you say, Wim.  I wonder if all I would do using the Tri-B mask would be to collimate out the camera tilt which is sometimes quite large according to ASTAP (I mis-typed above - it's ASTAP I use for SQM and tilt).

I found a thread on Cloudy Nights (here) which referenced a Tri-Bahtinov grabber on Github (here) which, although a bit quirky, seems to give understandable results.  

Here's what it made of image 5 in the sequence above:

 

I haven't worked out yet how to relate the channels to collimation screws (there is an assist diagram available but not sure how it works yet). Also not sure if I want to collimate for tilt at set-up every time (= lost imaging time) or whether 'close enough' is sufficient?

Perhaps something to have a play with during the next two months of Astro twilight?

I'll check those links. My scope is already in storage because I want to avoid birch pollen. Regarding orientation, if you rotate the mask such that the inner grid aligns with the collimation screws, the diffraction spikes are aligned with the screws. In the end, you'll probably need to test which is which anyway.

[gorann]

I used to worry a lot about tilt, eggy stars,  and how to fix it, and then BlurXT2 came along and I stopped worrying😅

Edited April 26 by gorann

[almcl]

1 hour ago, wimvb said:

if you rotate the mask such that the inner grid aligns with the collimation screws, the diffraction spikes are aligned with the screws. In the end, you'll probably need to test which is which anyway.

Thanks, Wim. 

I suspect there will be some trial and probably quite a lot of error while getting to grips with this.  On the positive side my 190MN has proved remarkably stable as far as collimation goes, must be all those nice baffles and the corrector plate keeping everything in position 🙂so once an accurate collimation is dialled in it should hold it for a while.

 

 

[wimvb]

11 hours ago, almcl said:

190MN has proved remarkably stable as far as collimation goes

That's my experience as well. No flimsy spider vanes to carry the secondary assembly. I typically only need to check and perhaps tweak collimation when I put the scope back on the mount in August.