30 replies to this topic

[Marcel Lancelle]

I looked into focussing masks a bit and noticed that for Bahtinov masks, everyone seems to use a regular grid. This produces a refraction pattern, i.e. a stripe with rainbows and gaps.
I think focusing should be easier when using multiple overlapping frequencies, which should be much closer to a continuous white line.
In particular, this should be an advantage when using a narrow band filter. I couldn't find anything about this topic, has anybody tried or heard of this before?

 

I tested a number of configurations with a simulator 'diffractsim'. The attached image shows some results. The ordering seems to have a small influence, I'm not quite sure why. The second mask seems to have the most continuous line, I also attached the corresponding pdf (svg upload is not allowed?).

 

Unfortunately, I do not own neither the necessary gear for astrophotography nor for fabricating the mask, so I cannot test this myself, but maybe someone here would like to test and give feedback.

Attached Thumbnails

• 

Attached Files

•  bahtinov-lancelle-mask.pdf   1.84KB   94 downloads

[Mike I. Jones]

These are interesting potential alternatives to Bahtinov's regularly spaced gratings.  What's most important is if these new diffraction patterns change with good (or even better) visibility while going through focus.  Can your software tool simulate defocus?  

[Marcel Lancelle]

Yes it can and I ran some tests. As I expected, nothing weird happens. I'm attaching two examples which are out of focus (the left one is more out of focus).
 
It would be great to see someone testing the mask on a real setup and not just a simulation, though

Attached Thumbnails

• 

[kathyastro]

It would be great to see someone testing the mask on a real setup and not just a simulation, though

In a community association I once belonged to, there was an unspoken rule that, if you made a suggestion along the lines of "I'd like to see someone do ...", you were considered to have volunteered for the position.  Just sayin'. 

[Mike I. Jones]

Good, results as expected.  

Idea: experiment with reducing the angles of the slanted gratings.  That would reduce the angle between the slanted diffraction spikes and push the intersections outwards to make the offset more visible?

 

Can't you 3D print one of your masks and try it?  What function are you using to increase the grating spacing?  Linear, quadratic, etc.?

[Marcel Lancelle]

For the slanted gratings I used an angle of 20 degrees which seems to be the usual angle with Bahtinov masks. I'm assuming there's a good reason for that value - does somebody here know? Of course, it can be modified an will influence the angle of the diffraction lines accordingly. But yes, maybe the angle is not optimal with this mask. I'm attaching a version with 10 degrees but I'm not sure which version is better.

 

As I wrote, unfortunately, I neither own a 3D printer nor an astrophotography setup. Well, I have a desktop engraving machine but no workshop - so I can't use it in my living room due to noise and dust, didn't think that through... I also have a cheap Newtonian and a DSLR, but the sensor is too far away so I cannot reach focus.

Attached Thumbnails

• 

[Marcel Lancelle]

I just realized I should have compared the out of focus simulation results, here they are.

Attached Thumbnails

• 

[calypsob]

To me there seems to be alot of central scatter. The original bahtinov has very clean distinctive separation.

 

these designs might be worth exploring too? http://www.goldastro...focus/index.php
 

[Marcel Lancelle]

Oh great, thanks for pointing that out!

The GoldFocus High-Precision Focus mask appears to use 3 different slit widths and a constant thinner bar width. So it's already going into the direction I followed, but still shows rainbow diffraction lines.

I'm attaching the comparison for the same amount of defocus.

Attached Thumbnails

• 

[Marcel Lancelle]

As for the central scattering, I think the simulation uses a medium resolution for the mask which may cause some of it.

I compared to the original Bahtinov mask which I found here (via Wikipedia). Again, both are out of focus by the same amount.

Attached Thumbnails

• 

[luxo II]

I'd suggest using the usual Bahtinov mask (top pic in your post#1) explore what happens if the black bars are narrower, allowing higher transmission. Even to the point they would have to be wires.

[Marcel Lancelle]

@luxo II, could you explain why? Is this for fabrication, or what do you expect from the resulting pattern?

[luxo II]

I expect the same pattern, just brighter.

The Bahtinov works fine, but there are great swathes of sky where bright stars for focusing are few and far between… it would help especially with a small scope (eg 70mm) if the mask let through rather more light.

In addition the spikes could be narrower, like those seen from the vanes of newtonians…

Edited by luxo II, 25 February 2022 - 08:06 PM.

[Arjan]

Marcel,

 

I understand there is an optimum grid constant for Bahtinov masks, that depends on focal length:   f = s * (150..200)

So there is already some spread in this relation, but nevertheless masks should vary considerably between for example shorty refractors and long Schiefs.

 

Did you take this also into account?

[Marcel Lancelle]

luxo II, I understand. The attached simulatons show a regular Bahtinov mask with the same number but thinner bars (center). There is just more light coming in, and your focused star will be brighter (the simulation doesn't show it, it's already saturated).

 

But actually, you want to have brighter diffraction lines, so I think you'd need more edges. The bottom simulation has 3x the thin bars and this seems to be the case.

Warning: the simulation may not be accurate, as it uses a medium resolution, there could be artifacts caused by the thin structures.

 

 

Arjan, I didn't know that. Where can I read more? What does the formula optimize? In my experiments I always used the same focal length and was trying to create more white continuous lines instead of the rainbow interrupted lines. A ratio of 3:1 for thick to thin bars worked better for that goal than a ratio of just 2:1. Yes, I can imagine that the best number of bars depends on the focal length.

Attached Thumbnails

• 

[luxo II]

Of those #2 and #3 would be ones I'd try in practice.

[Arjan]

Marcel, you can find it in the original thread about this topic, which refers to a discussion on a Russian forum with Pavel Bahtinov himself.

 

About the why: The grid works as a diffraction grating, the grating constant determines how far out the maximum is. More slits with the same constant will amplify the maxima. When you vary the constant over the grid you smear out the maxima along the diffraction spike. No idea whether this is good or not, I can imagine that with weak stars you would like to concentrate the light a bit more, but with a bright star you can afford to spread it out and possibly have a better judgment on focus quality...

Edited by Arjan, 01 March 2022 - 11:00 AM.

[calypsob]

Oh great, thanks for pointing that out!

The GoldFocus High-Precision Focus mask appears to use 3 different slit widths and a constant thinner bar width. So it's already going into the direction I followed, but still shows rainbow diffraction lines.

I'm attaching the comparison for the same amount of defocus.

Id like to see more variations of the collimation mask. They all look good to me. 
 

Thin lines give thin bahtinov spikes. Fat lines give fat bahtinov spikes. Just like the spider on a newtonian. 
I feel that on a wide angle lens I get a better focus with thin wirelike lines. 
Maybe thinner lines allow you to focus better, I often print snd pair mine with fast optics.

[Arjan]

"Thin lines give thin bahtinov spikes. Fat lines give fat bahtinov spikes."

Why?

[Mike I. Jones]

"Thin lines give thin bahtinov spikes. Fat lines give fat bahtinov spikes."

Why?

Yes, it's just the opposite, per Fourier Transform scaling theory.  Narrow lines increase diffraction width and vice versa.

[Marcel Lancelle]

Arjan, thanks for the explanation and the link!

 

Just for completeness, Dennis Sakva writes and the end of page 4: "The mask is optimized to work on either first or third order spectrums". The given explanation is to have the bright diffraction parts at a good distance to the central spot.

My idea was to distribute them along the line instead of having spots, thinking that with the lines it would be more precise or better to see, provided there is enough light (or you can expose for long enough).

The formula was apparently found by trial and error. After reading the thread, I guess it's good to try using the formula for maybe the smallest or the average bar/slit width. So yes it appears it makes sense to have different versions for very different focal lengths.

 

calypsob, could you please clarify which variations you'd like to see? Do I understand correctly that you are interested in a version with more and thinner bars/slits?

[Marcel Lancelle]

calypsob, I just realized you wrote about collimation masks, so you mean the Goldstar masks with combined collimation and focussing capabilities?

[luxo II]

Marcel another variable you haven’t played with is the angle of the two panels of slits in the bottom half.

Does anyone has any views about the angle of the “X” in the diffraction pattern ? Why not 60 degrees ? Flatter, or steeper angles ? Does this alter the precision for determining the focus ?

NB when focussing I concentrate on the diffraction lines, not the blobs.

Edited by luxo II, 02 March 2022 - 04:51 AM.

[Marcel Lancelle]

luxo II, see posts #6 and #7 where I compared the usual 20 degrees to 10 degrees.

I guess, if the angle is smaller and lines are thin enough, it may be easier to see an offset.

Edited by Marcel Lancelle, 02 March 2022 - 04:52 AM.

[luxo II]

Ah yes sorry I didn’t pay attention

It’s something I’d like to try in the field though, and compare with one with slits at +/- 30 degrees

Edited by luxo II, 02 March 2022 - 04:55 AM.