263 replies to this topic

[dg401]

I'm not sure I agree with him though.  The primary and secondary are well fixed in the OTA, that's true, but that's not necessary to make them coaxial.  Using the tilt adjustments on both mirrors you can point them at one another (make them perfectly coaxial) even if they're not on the mechanical axis of the OTA.

Diagram FTW.

 

[TinySpeck]

Diagram FTW.

 

Here's what I think:  HOM doesn't make your mirrors parallel, it makes them coaxial.  It would not lead to the 2nd figure (parallel but not coaxial).  If they weren't coaxial, they wouldn't look symmetrical on all four vanes.  And HOM will lead to coaxial mirrors whether they are locked to the OTA axis or not.

 

I don't see that the mirrors could be locked close enough to the OTA axis to be good enough to rely on for collimation, and it's not necessary that they be either.

[MikeECha]

 

I'm not sure I agree with him though.  The primary and secondary are well fixed in the OTA, that's true, but that's not necessary to make them coaxial.  Using the tilt adjustments on both mirrors you can point them at one another (make them perfectly coaxial) even if they're not on the mechanical axis of the OTA.

 

 

But geometry disagrees with you.

[TinySpeck]

But geometry disagrees with you.

Your geometry maybe!  If you displace one mirror in a plane perpendicular to the OTA axis, you can always make the mirrors perfectly coaxial by adjusting the tilt of both of them.  That's just a fact.  They won't be coaxial with the OTA, but they will be with each other.  I'm not sure if you're getting at something else, or what.

[MikeECha]

Your geometry maybe!  If you displace one mirror in a plane perpendicular to the OTA axis, you can always make the mirrors perfectly coaxial by adjusting the tilt of both of them.  That's just a fact.  They won't be coaxial with the OTA, but they will be with each other.  I'm not sure if you're getting at something else, or what.

But as dg401 said, this design does not allow for that displacement. The secondary is fix in place by  the center screw and the primary by the center hub. So there is no ability to displace one mirror to chase the other. No in the x-y perpendicular to the tube.

 

There is even a term what you think will happen: "squinted collimation". 

Edited by MikeECha, 09 August 2021 - 05:39 PM.

[nebulachadnezzer]

But as dg401 said, this design does not allow for that displacement. The secondary is fix in place by  the center screw and the primary by the center hub. So there is no ability to displace one mirror to chase the other. No in the x-y perpendicular to the tube.

 

There is even a term what you think will happen: "squinted collimation". 

This is what I was getting out a few days ago. While it's theoretically possible to have coaxial mirrors that aren't aligned to the OTA, it's not with the GSO RCs because the mirrors are centered in the OTA by design (unless you manage to find a way to make them off-center, such as bending or forcing the secondary mount or maybe shimming the primary on the hub -- not even sure tolerances allow for the latter).

There is certainly an argument to be made that the GSO design is imprecise enough to allow for error here but if so it offers no means to correct that error. I don't know whether or not this occurs. But the fact that we can reach a high degree of collimation with at least some examples of the scope must indicate they get it right occasionally.

[nebulachadnezzer]

Here's what I think:  HOM doesn't make your mirrors parallel, it makes them coaxial.  It would not lead to the 2nd figure (parallel but not coaxial).  If they weren't coaxial, they wouldn't look symmetrical on all four vanes.  And HOM will lead to coaxial mirrors whether they are locked to the OTA axis or not.

 

I don't see that the mirrors could be locked close enough to the OTA axis to be good enough to rely on for collimation, and it's not necessary that they be either.

I'm not sure I follow your reasoning. Maybe I'm not thinking about it correctly.

If we assume the GSO design enforces the mirrors being *physically* coaxial (and not adjustable in this manner) AND we give them credit for having managed to figure mirrors with a matching optical and physical center (humor me) then the only way I see you can get a balanced HOM effect is for them to be collimated. Any error in collimation is going to bend the HOM one way or the other. Isn't it?

If the optical centers aren't the physical centers then all kinds of trouble can arise, but with no way to make the mirrors optically coaxial, I'm not sure how you can collimate *hyperbolic* mirrors that are misaligned. If they were spherical then that would be a different story.

I don't know what fraction of a wavelength they'd have to get the mirrors figured incorrectly (off-center) before human eyes could detect an error in the HOM alignment. Even a star test measuring HFD or FWHM has limitations here, but frankly once we're down to near-perfect star tests who cares?

This is really beyond my optical expertise, so perhaps someone can explain to me why my thoughts aren't correct.

[TinySpeck]

But as dg401 said, this design does not allow for that displacement. The secondary is fix in place by  the center screw and the primary by the center hub. So there is no ability to displace one mirror to chase the other. No in the x-y perpendicular to the tube.

 

There is even a term what you think will happen: "squinted collimation". 

YOU said this, and @dg401 agreed.

 

The mirrors are not placed exactly enough on the OTA axis to depend on for collimation.  Fortunately collimation does not depend on that.  The mirrors will be slightly off the OTA axis, but they can be made exactly coaxial with one another by adjusting both their tilts.

 

I don't know what you mean by "squinted collimation".  Imagine moving one mirror WAY off the OTA axis.  You can still make them perfectly coaxial by tilting them both.  There's no "squinted" about it.

 

Please, this is not contributing to this thread.  If you want to continue, contact me by PM.

[MikeECha]

This is what I was getting out a few days ago. While it's theoretically possible to have coaxial mirrors that aren't aligned to the OTA, it's not with the GSO RCs because the mirrors are centered in the OTA by design (unless you manage to find a way to make them off-center, such as bending or forcing the secondary mount or maybe shimming the primary on the hub -- not even sure tolerances allow for the latter).

There is certainly an argument to be made that the GSO design is imprecise enough to allow for error here but if so it offers no means to correct that error. I don't know whether or not this occurs. But the fact that we can reach a high degree of collimation with at least some examples of the scope must indicate they get it right occasionally.

No one would design something like that. It would be nightmare. Adding more degrees of freedom to then turn and have to control them precisely would be crazy. And I am not 100% certain about this but, reading the other day about curved mirror optics I think I read that the tube play the role of the field stop (like in an eye piece). So if tilted, it would project an ellipse on the primary.

 

Think about it. Ideally, the best telescope would be one that is a casted, perfectly machined solid structure where the mirrors sit perfectly collimated and no motion/adjustments required. So in reality, a design with the least amount of degrees of freedom and closest to ideal would be best 

[TinySpeck]

Folks, we're going way into the weeds here.  We have some disagreements about things that have nothing to do with this thread.  Please start another thread.

[nebulachadnezzer]

No one would design something like that. It would be nightmare. Adding more degrees of freedom to then turn and have to control them precisely would be crazy. And I am not 100% certain about this but, reading the other day about curved mirror optics I think I read that the tube play the role of the field stop (like in an eye piece). So if tilted, it would project an ellipse on the primary.

 

Think about it. Ideally, the best telescope would be one that is a casted, perfectly machined solid structure where the mirrors sit perfectly collimated and no motion/adjustments required. So in reality, a design with the least amount of degrees of freedom and closest to ideal would be best 

To be clear I wasn't suggesting they should implement more degrees of freedom. A few days ago I noted that GSO's RC designs have a non-adjustable spider compared to their Newts which are adjustable. I consider the omission of this feature on the RC's a *very* good thing.

I was saying they're not adjustable, and we had better either conclude that they're centered "enough" and coaxial or they aren't and we should just ditch these things and buy real telescopes. In practice, they seem to be centered "enough" or I don't think we could reach good collimation.

[MikeECha]

To be clear I wasn't suggesting they should implement more degrees of freedom. A few days ago I noted that GSO's RC designs have a non-adjustable spider compared to their Newts which are adjustable. I consider the omission of this feature on the RC's a *very* good thing.

I was saying they're not adjustable, and we had better either conclude that they're centered "enough" and coaxial or they aren't and we should just ditch these things and buy real telescopes. In practice, they seem to be centered "enough" or I don't think we could reach good collimation.

I was agreeing with you.

[MikeECha]

YOU said this, and @dg401 agreed.

 

The mirrors are not placed exactly enough on the OTA axis to depend on for collimation.  Fortunately collimation does not depend on that.  The mirrors will be slightly off the OTA axis, but they can be made exactly coaxial with one another by adjusting both their tilts.

 

I don't know what you mean by "squinted collimation".  Imagine moving one mirror WAY off the OTA axis.  You can still make them perfectly coaxial by tilting them both.  There's no "squinted" about it.

 

Please, this is not contributing to this thread.  If you want to continue, contact me by PM.

Search for the term here in the forum.

 

As far as contributing, I corrected something that was wrong and you questioned further. So I responded. 

 

And yes, it is simple geometry: a line between two fixed points can only be drawn in one orientation.

 

Sorry that you feel questioned by a simple principle. 

[nebulachadnezzer]

I was agreeing with you.

There we go again. lol

So to try to rescue this thread for @TinySpeck, I think my whole point here is that the mirrors seem to be *physically* centered in the OTA, and the only way they wind up *optically* coaxial (since "centered" and "coaxial" are not the same thing) is if they are properly collimated.

AND if my (admittedly limited) understanding is correct, the HOM will only be balanced for two hyperbolic mirrors if they are both optically coaxial and collimated. Remember that the HOM not only depends upon the mirror alignment but also the alignment of those to the secondary mount and spider, which are our optical references.

This is a long way of saying: I think achieving a balanced HOM all the way around (or at least at the reference points of all four spider vanes) means the mirrors are collimated, coaxial, and optically centered in the OTA.

I am open to learning how I may have this incorrect.

[dg401]

Like TS said, we may be off into the weeds.

 

I want to make sure I've been using the term "coaxial" correctly.  I've only been at this whole RC collimation thing for 6 weeks, so I'm error prone.  I never much thought about collimation before, because it's never been such a searing pain in my backside before.

 

When I use the term "coaxial" in reference to a mirror pair, I mean (perhaps incorrectly) the following:

 

1.  Concentric in X and Y (the face plane)

2.  Centered on Z, where Z is a line perpendicular to the face plane.

 

For two mirrors of the same diameter, this could be described as the spacial relationship between the two open ends of a regular cylinder.  For difference size mirrors substitute a cone and cut it parallel to the base.

 

If the centers are constrained to the line Z, then the centers (spaced but concentric) can only remain concentric if both centers remain perpendicular to the line Z.

 

[Edit: If what I'm describing does not accurately express the meaning of "coaxial", then would someone please clue me in on the term for what I am describing? "Collimated" comes to mind, but would there be a more general geometric term?]

Edited by dg401, 09 August 2021 - 07:47 PM.

[dg401]

Coaxial:  This

 

When opening a can of worms, it's wise to read the warning label on the can:

"Warning - Contains Worms"

Edited by dg401, 09 August 2021 - 08:04 PM.

[dg401]

If we get too theoretical, we can stray from the practical of what we can do and get caught up in the soupy mire of what we might do if only we could...

 

With that in mind, here are some items to consider:

 

1.  We can talk about axes and planes all we like, but at the end of the day, we're working with 6 adjustment screws and an additional central screw.  I like to tell my customers, "There are 10 ways to do anything and 9 of them are probably wrong".  There are infinite wrong endpoints for these 7 screws and far fewer endpoints that produce good results.

 

2. Terms like "centered" and "HOM" are visual terms.  They have a basis in geometry, but all we're really doing is making something appear a particular way to our eye and hope that the view comports with the geometry behind what we're doing. We don't "do geometry", we turn wrenches and pray we don't drop one down the OTA (Helpful hint:  Attach your hex wrench to a line and clip it to a secondary vane so that if you do drop the wrench into the OTA, it's can't reach the primary mirror and ruin your day.  Also, trash bags make good rain-wear if you don't mind looking like a bum).  Odds are poor we'll ever center a dot in the mechanical sense, but we can center it in a visual sense, and we can 2nd guess our visual centering ad infinitum if we allow ourselves to get caught up in a vicious cycle of obsessive/compulsive perfectionism.  Two different observers can vary in their visual acuity (and the ability of the brain to interpret the visual) such that one observer may be able to judge center better than another.  The two observers will probably argue over who has a better eye for the center (because why not?) and harsh words will be exchanged.  At this point, it's helpful if an impartial 3rd observer happens along to inform them that they're both gormless twits and would you guys please knock off the arguing, it's upsetting the dog.

 

3.  When discussing the Cheshire centering inside the secondary donut, it seems a precise centering won't likely result in optimal on-axis performance.  Rather, we're starting to see that this instead gives us a good reference point for further on-axis optimization.  For this, I've suggested the tri-bahtinov mask.  The TBM is relatively cheap, needs no batteries, and has more diagnostic power than a round chunk of plastic has any right to have.  You can even leave it out in the rain and the TBM will be no worse off for it.  I've got in on my schedule to develop a recipe for using the TBM to bring the 3 equally spaced TBM axes into simultaneous focus via secondary screw adjustments with the TBM installed.  Again.. what we can do, not what we might do if only we could...

 

4.  I don't pay much attention to the whole focuser/primary mirror alignment issue since my focuser/primary happen to be very well aligned.  I choose to use a laser to check this alignment, but others may not.  Not everyone has an aligned laser and anyway they're over-priced and under-useful in many but not all cases.  There are other ways to check/adjust this alignment as TS discusses.  To each his own.

 

5. Once we have completed the iterations of visually centering the Hall of Mirrors using whatever visual cues we prefer (2nd bright reflection of the open end, bla bla bla) and we have achieved best on-axis performance (again, Cheshire probably not centered, but using tri-bahtinov mask at twilight), we have exhausted our ability to further improve the visuals.  The photographic results we get are the results we get.  We might observe, "Left corners look good, upper right looks good, lower right looks blobby", but fixing the lower right without probably screwing up another aspect of the frame will require a whole different procedure and a level of optical understanding that most of us don't possess.  Perhaps someone will come up with a recipe for achieving this, but I suspect we're into such tiny adjustments here that we're in the realm of "improve one aspect, degrade another aspect".   We're working with M5/0.8 metric screws.  One turn will give 800 microns of translation.  The entirely notional idea of "one twentieth of a turn" represents 18 degrees of rotation and 40 microns of translation.  You don't make a twentieth of turn, you bump the hex wrench and pray it makes things better.  I suspect that the optical quality of the mirror system can be made perceptibly better or worse by translations well below 40 microns, which is already a distance that we can only control precisely if we have vivid imaginations.  In short, the "bump" of the wrench that resulted in a "perfect" image was pure luck and we won't reproduce that perfect result with any force other than another bout of pure luck.

 

See:  Diminishing returns.

See:  Leave well enough alone.

See:  Don't mess with success.

See:  Knock it off and image something

 

We didn't buy these things for the "joy" of adjusting them. That's just a necessary slog to complete. Stack some images and process something nice!

Edited by dg401, 10 August 2021 - 02:32 AM.

[quilty]

May I sum up what is the state of the art in my understanding?

 

A balanced Hom makes the mirror axes parallel to at least an extent, other collimation methods do. Any tilt of primary or secondary mirror from perpendicular to the optical axis would be seen. Guess we all can agree

 

We don't know exactly how much HoM can see as well for coaxiality, me at least. So we hope for sufficient machining consistency at GSO to provide enough coaciality which is necessary to not spoil the imaging power of the hyperbolic (parabolic) mirrors. Am I right?

 

And while imaging, do we find this provided?

[dg401]

May I sum up what is the state of the art in my understanding?

 

A balanced Hom makes the mirror axes parallel to at least an extent, other collimation methods do. Any tilt of primary or secondary mirror from perpendicular to the optical axis would be seen. Guess we all can agree

 

We don't know exactly how much HoM can see as well for coaxiality, me at least. So we hope for sufficient machining consistency at GSO to provide enough coaciality which is necessary to not spoil the imaging power of the hyperbolic (parabolic) mirrors. Am I right?

 

And while imaging, do we find this provided?

If you guess we can all agree, you've not read page 8.

 

GSO optics are capable of excellent images, especially for the price.  GSO mechanics are also good enough to produce fine images, but the collimation is challenging.  The OPs procedure attempts to achieve at least a fair collimation and at best an excellent collimation with a minimum of gadgets, on the bench/tabletop, and particularly, without lasers.  As written, the procedure has produced fair to excellent results.  With the procedure tweaked, the fair results have been improved to excellent.

 

A balanced HoM and a centered Cheshire dot approximate a collimated scope.

 

Further tweaking of the Cheshire dot using a tri-bahtinov mask to center all three diffraction spike sets has produced excellent star-fields with 3 of 4 corners showing round (or approximately round) stars (APSC sensor).  "Perfection" on all four corners with an APSC sensor may be possible, but the required ultra-fine tweaks may be beyond our ability to make small enough and precise enough screw adjustments (see my previous comment on screwing up one corner while trying to fix another corner).  If one does achieve "perfection" on all four corners, it may be closer to blind luck that anything that can be reliably replicated.  To extend "perfection" to a full frame sensor will require not only a precise collimation but also a field flattener.  We must distinguish between out of round stars at the corners of an APSC image (astigmatism) versus out of round stars at the corners of an otherwise perfect full frame image (field curvature).
 

Good Lord, I can't sum up anything in a sentence or two, can I?  Let me try again:

 

A balanced HoM is always a good thing.  A centered Cheshire dot may be further tweaked with a Bahtinov mask.  A recipe for this tweak is in the works.

Edited by dg401, 10 August 2021 - 06:24 AM.

[dg401]

Recipe for on-axis optimization with tri-Bahtinov mask.

 

If the Cheshire dot is centered and HoM optimized per this procedure, a tri-bahtinov mask will likely reveal a spike set from the mask out of focus with two sets in focus.  Or you may see one set in focus, one set out of focus, and the last set very nearly in focus.

 

I know I've read about this somewhere, so it's not mine.  Credit to the source or sources where I read this, but I'm not going back and figuring out who it was.

 

Line up the mask segments with the secondary screws.  The mask segments are spaced at 120° as are the secondary adjustment screws.  With an on-axis star or artificial star, focus the mask as precisely as possible.  Adjust the screw corresponding to the worst focused spike pair.   Either tightening or loosening will move the spike pair toward focus.  Don't forget to back off the other pair of screws (I got lazy with this and bottomed out and stripped one of the secondary screws.  Now I've got to tap them out to M6X1, and that's going to make me redo my the focal length math for my scope when it comes to turns of the secondary screws).  Once you get this set of spikes centered, there will probably be another set of spikes just a bit out of focus.  Find the corresponding screw and identify the adjustment direction to center the spike pair.  Adjust the screw corresponding to any out of focus spike pair until all 3 pairs of spikes are perfectly centered.

 

If you recall, I got a really good star-field after adjusting the Cheshire dot (somewhat randomly) to kiss up against the donut ring at the 3:00 position (relative to placing the vixen rail at the bottom).  I began the tri-Bahtinov adjustment with HoM centered and Cheshire dot centered.  It was very interesting that when I checked the Cheshire dot position after focusing the tri-Bahtinov mask on-axis, the dot was located at 4:00 and actually half-inside and half on top of the donut.  This happens to be less than a dot size away from the position that gave me the nice-star-field.

 

Then I started screwing around and stripped the secondary screw.  That put an end to the night outside with the artificial star.  I'll tap them out to M6 tomorrow.  With the way I've turned this poor scope into a test-bed, I'm putting a lot of wear and tear on it, I've already tapped out the screws that attach the vixen saddle to the OTA from M4 to M5 due to stripped threads, and now I've got to tap them to M6 as I've stripped another... and now there's the secondary screws.  Fortunately, I've got more metric screws than I can ever use and the required taps.  Still, it's a pain.  So much for me leaving well enough alone!

[TinySpeck]

Recipe for on-axis optimization with tri-Bahtinov mask.

 

If the Cheshire dot is centered and HoM optimized per this procedure, a tri-bahtinov mask will likely reveal a spike set from the mask out of focus with two sets in focus.  Or you may see one set in focus, one set out of focus, and the last set very nearly in focus.

 

I know I've read about this somewhere, so it's not mine.  Credit to the source or sources where I read this, but I'm not going back and figuring out who it was.

 

Line up the mask segments with the secondary screws.  The mask segments are spaced at 120° as are the secondary adjustment screws.  With an on-axis star or artificial star, focus the mask as precisely as possible.  Adjust the screw corresponding to the worst focused spike pair.   Either tightening or loosening will move the spike pair toward focus.  Don't forget to back off the other pair of screws (I got lazy with this and bottomed out and stripped one of the secondary screws.  Now I've got to tap them out to M6X1, and that's going to make me redo my the focal length math for my scope when it comes to turns of the secondary screws).  Once you get this set of spikes centered, there will probably be another set of spikes just a bit out of focus.  Find the corresponding screw and identify the adjustment direction to center the spike pair.  Adjust the screw corresponding to any out of focus spike pair until all 3 pairs of spikes are perfectly centered.

 

If you recall, I got a really good star-field after adjusting the Cheshire dot (somewhat randomly) to kiss up against the donut ring at the 3:00 position (relative to placing the vixen rail at the bottom).  I began the tri-Bahtinov adjustment with HoM centered and Cheshire dot centered.  It was very interesting that when I checked the Cheshire dot position after focusing the tri-Bahtinov mask on-axis, the dot was located at 4:00 and actually half-inside and half on top of the donut.  This happens to be less than a dot size away from the position that gave me the nice-star-field.

 

Then I started screwing around and stripped the secondary screw.  That put an end to the night outside with the artificial star.  I'll tap them out to M6 tomorrow.  With the way I've turned this poor scope into a test-bed, I'm putting a lot of wear and tear on it, I've already tapped out the screws that attach the vixen saddle to the OTA from M4 to M5 due to stripped threads, and now I've got to tap them to M6 as I've stripped another... and now there's the secondary screws.  Fortunately, I've got more metric screws than I can ever use and the required taps.  Still, it's a pain.  So much for me leaving well enough alone!

Ouch, sorry about having to re-tap those screws!  You've sacrificed your scope to the cause, and we salute you.

 

It seems that your Cheshire dot position is now way off from just kissing the inside of the donut.  Half out of the donut is a long way from being centered, and implies that the original visual method doesn't get you very close.  Am I understanding that correctly?

[dg401]

Ouch, sorry about having to re-tap those screws!  You've sacrificed your scope to the cause, and we salute you.

 

It seems that your Cheshire dot position is now way off from just kissing the inside of the donut.  Half out of the donut is a long way from being centered, and implies that the original visual method doesn't get you very close.  Am I understanding that correctly?

You are understanding that correctly.

 

I've got everything re-tapped so I can proceed.  I won't feel confident that the mask adjustment is reproducible until I've re-focused the diffraction spikes and can show that the Cheshire position returns to 4:00 and half on top the donut ring.  I was able to adjust things so that the on-axis spikes lined up while the Cheshire was very outside the donut.  In this instance though, the corners were awful.  This occurred when trying to optimize the mask from a random position.  So the lesson here is to optimize the mask beginning with the Cheshire dot centered using a minimum of adjustments.

 

I can easily enough adjust the secondary with the scope mounted and pointed at a star/artificial star, but the resulting iterative HoM tweak isn't so simple since that relies on the  focuser being pointed at the ground such that loosening the primary locking screws doesn't cause tilt due to any sideways orientation of the scope.  It's easy enough to take the scope off the mount, but then making the HoM adjustment with the scope in my lap is a bit of a pain.  I think I need a jig laid across a couple saw-horses to make this easier.  By jig, I mean something like a 2 foot square piece of plywood with a ~130 mm hole cut into it to hold the scope with the focuser pointed downward with the adjustment screws accessible.  Much of working this stuff out turns out to be figuring out how to make the performance of various steps less clumsy.  If I'm uncomfortably holding the OTA in my lap, then my adjustments won't be as good as if I can comfortably take my time getting things right.

[TinySpeck]

You are understanding that correctly.

...

Thanks again.  Any ideas why the first tri-Bahtinov experiment was symmetrical when the dot just touched the donut interior but now it requires the dot to be half out of the donut?  That throws a pretty large monkey wrench into this whole procedure.

[dx_ron]

 I think I need a jig laid across a couple saw-horses to make this easier. 

Can you not just use the mount to point the scope straight up? If that makes the top too high, can you shorten your tripod (or hacksaw your pier )?
 

[dg401]

Thanks again.  Any ideas why the first tri-Bahtinov experiment was symmetrical when the dot just touched the donut interior but now it requires the dot to be half out of the donut?  That throws a pretty large monkey wrench into this whole procedure.

If you go back and look at the photo at 3:00, The spikes look pretty good but not perfect. With the mask and the adjustment screws lined up, I was able to tweak it to visual perfection.  I think that the difference between pretty good and perfect, accounts for the difference between 3 o’clock and 4 o’clock. It’s a very small difference.