316 replies to this topic

[SandyHouTex]

I think Suiter touches on this, but to the group:

 

What would the seeing have to be in order for optics better than diffraction limited be apparent? My recollection is that Suiter makes the point that it would have to be darn good, and people hand-wring far too much about their optics not being good enough. Unless you have adaptive optics corrections in your imaging train, none of this may matter. Likely the distinction between 1/5 wave and 1/10 wave is undetectable, and for sure 1/10 wave and 1/50-100 wave is not unless tested in the lab.

 

Bob

As I recall, when they did the test at Stellafane with a bunch of Newts figured to 1 wave, 1/2 wave, 1/4 wave, and 1/8 wave, people could see the improvement between 1/4 wave and 1/8 wave.

[Bob Campbell]

As I recall, when they did the test at Stellafane with a bunch of Newts figured to 1 wave, 1/2 wave, 1/4 wave, and 1/8 wave, people could see the improvement between 1/4 wave and 1/8 wave.

Thanks for the info.

 

This post might shed additional light on the subject:

 

https://www.cloudyni...tics/?p=1815065

 

Bob

[SandyHouTex]

Thanks for the info.

 

This post might shed additional light on the subject:

 

https://www.cloudyni...tics/?p=1815065

 

Bob

That's it.  Thanks for finding it.

[davidc135]

I added to the setup in post 46 by placing a flat at Focus B as in the pihoto below so that the ke tests with and without the 2ndry can be directly compared. The binocular objective was swpped for a Criterion finder doublet with a longer focal length making the first focus A 31ins from the objective.

 

On arriving at B the rays switch sides, top and bottom and side to side and then return exactly co-axially to A and so it's not possible to do this test without a beam splitter. Luckily I have a 20mm cube from Surplus Shed.

 

I found out that the 2ndry is under-corrected by 1/4 to 1/3 wave and also that the surface has slight grooves around 2/3 r which suggests that it could have been hand figured or re-touched as mentioned earlier. In spite of this roughness the Meade sct performs very well. Both 50mm doublets had good quality smooth optics.

 

 

On another note I tested the corrector plate by interference with my 12'' flat and found that the neutral zone was around the 86% radius mark which came as a surprise. Both sides had been aspherised.

 

David

Edited by davidc135, 15 June 2022 - 02:13 AM.

[MKV]

What would the seeing have to be in order for optics better than diffraction limited be apparent? My recollection is that Suiter makes the point that it would have to be darn good,...

Sure, but that would also depend to a large extent on the type of object observed (extended vs point source), the type on instrument and accessories used (obstructed, unobstructed, type of eyepiece, etc.), the visual acuity of the observer (i.e. age, ophthalmic correction, and so on.), but I think one thing will be a dead giveaway in favor of the 1/10 wave (exist wavefront) optics: the snap-to-focus.  

[davidc135]

This thread has wandered off topic.  David

[dan chaffee]

Sure, but that would also depend to a large extent on the type of object observed (extended vs point source), the type on instrument and accessories used (obstructed, unobstructed, type of eyepiece, etc.), the visual acuity of the observer (i.e. age, ophthalmic correction, and so on.), but I think one thing will be a dead giveaway in favor of the 1/10 wave (exist wavefront) optics: the snap-to-focus.  

I think this raises a question about the perception of how close

to perfection the optics are in the presence of a large obstruction.

Typically, the commercial SCT's are using secondaries of about

33% of the aperture. In all but the higher spatial frequencies, the MTF

of a perfectly corrected instrument so obstructed is very similar to

unobstructed .25wavefront 3rd order spherical aberration.   It doesn't manifest

itself so much for questions of unsharpness as it does to contrast

on bright extended objects. One look at the planetary images by

Ed Grafton or Theirry Legault , using 14 inch SCT's is evidence

that contrast lost in visual observation can be recovered handsomely

in processing of ccds. However, I suspect these are very well

figured instruments well under .25 wavefront SA and would in fact,

snap to focus. But I wonder, how bad would they have to be for

errors to be perceived in the presence of a 33% obstruction, aside

from star testing, of course and in good seeing?

Edited by dan chaffee, 15 June 2022 - 02:49 AM.

[Bob Campbell]

"But I wonder, how bad would they have to be for

errors to be perceived in the presence of a 33% obstruction, aside

from star testing, of course and in good seeing?"

 

 

 

Exactly. The sct design is 'flawed' from the beginning with respect to central obstruction/contrast, so hand wringing wrt wavefront error is kind of a lost cause.

 

The original post was about aspheric secondaries, which did/did not correct for coma; we've departed from that considerably here, not that bothers me particularly.

 

good discussions, nevertheless,

 

Bob

 

Edited by Bob Campbell, 15 June 2022 - 09:02 AM.

[MKV]

The Cassegrain calculator told me that the 2ndry RoC was 10ins so, in the setup below, all I had to do was adjust the testing lens' focii to take account of that. The longer focus turned out to be 28ins. Light from the precision pinhole travels through the doublet and is reflected off the convex mirror to retrace its steps to be tested by knife edge at A. The errors seen at A will be twice those of the doublet plus any contribution by the 2ndry. A spherical convex surface will not add any error.

 

Your diagram/test set up is incorrect.

 

https://www.cloudyni...tach_id=2070637

 

The binocular lens should be turned around, so the signal reaches the flat (back) surface first. Your signal should come from lens' focal conjugate.

 

 

edit: OP's link inserted

 

PS That test is a test for measuring the radius of curvature of a convex surface.

Edited by MKV, 15 June 2022 - 12:52 PM.

[DAVIDG]

"On another note I tested the corrector plate by interference with my 12'' flat and found that the neutral zone was around the 86% radius mark which came as a surprise. Both sides had been aspherised."

   

 If this is in reference to a Meade corrector plate then yes they aspherized both sides. Meade corrector plates are thicker then Celestron  and with aspherizing on both sides that indicates to me that they were using the classic vacuum pan method vs the Master Block method that Celestron used. 

    If you  have made correctors  you would understand that the thicker plate reduces the problem of astigmatism when the plate is deflected under vacuum but since it is thicker it can not be deflected as much as a thinner  plate. since it would break.  This means you can't put the full correction on one side  as Celestron did but need to put 1/2 of the correction on both sides. Also putting the correction on both sides remove the problem  that the "flat" side of the corrector which Celestron  has, if  left untouched may not be very optically smooth.   As for the neutral zone being at 86% that gives the best achromatic correction so no surprise that anyone that understand the design would do this. It doesn't cost any more and the optical performance is better. 

 

    BTW Celestron did put the correction on both side of the correctors they made for their Schmidt cameras thou.  With the faster primary in the camera the corrector require a deeper curve  on the corrector  and even with the Master Block method and thin glass , the. glass would break under the  deflective required to pull it against the Master surface. So they split the correction in 1/2 and put it on both sides.

 

                      - Dave 

[Starman1]

If I recall correctly, Celestron made the F/6.3 flattener reducer/ corrector that was meant for correcting the coma as well as the curvature.

 

In the beginning, people did not have 31 Naglers and 41Panoptics to really see the curvature and coma.  I just wonder..  I think Glenn stated the traditional view which I seconded. Glenn has a lot more experience in optics than I do as he was an optical guy with Ceravolo until Peter moved west.

 

Jon

I can tell you from personal experience that it flattens the field and reduces the focal length but does NOT eliminate coma.

[davidc135]

Your diagram/test set up is incorrect.

 

https://www.cloudyni...tach_id=2070637

 

The binocular lens should be turned around, so the signal reaches the flat (back) surface first. Your signal should come from lens' focal conjugate.

 

cvx conj test.jpg

 

edit: OP's link inserted

 

PS That test is a test for measuring the radius of curvature of a convex surface.

I think my diagram is correct. The crown convex faces the longer focus. You have the plane facing the shorter conjugate which is also correct. However, it may not matter much for this test where the same error due to the objective is compared with and without the 2ndry. But I'll reverse the lens and see what effect that has.

 

David

Edited by davidc135, 15 June 2022 - 01:31 PM.

[davidc135]

Your diagram/test set up is incorrect.

 

https://www.cloudyni...tach_id=2070637

 

The binocular lens should be turned around, so the signal reaches the flat (back) surface first. Your signal should come from lens' focal conjugate.

 

cvx conj test.jpg

 

edit: OP's link inserted

 

PS That test is a test for measuring the radius of curvature of a convex surface.

 

"On another note I tested the corrector plate by interference with my 12'' flat and found that the neutral zone was around the 86% radius mark which came as a surprise. Both sides had been aspherised."

   

 If this is in reference to a Meade corrector plate then yes they aspherized both sides. Meade corrector plates are thicker then Celestron  and with aspherizing on both sides that indicates to me that they were using the classic vacuum pan method vs the Master Block method that Celestron used. 

    If you  have made correctors  you would understand that the thicker plate reduces the problem of astigmatism when the plate is deflected under vacuum but since it is thicker it can not be deflected as much as a thinner  plate. since it would break.  This means you can't put the full correction on one side  as Celestron did but need to put 1/2 of the correction on both sides. Also putting the correction on both sides remove the problem  that the "flat" side of the corrector which Celestron  has, if  left untouched may not be very optically smooth.   As for the neutral zone being at 86% that gives the best achromatic correction so no surprise that anyone that understand the design would do this. It doesn't cost any more and the optical performance is better. 

 

    BTW Celestron did put the correction on both side of the correctors they made for their Schmidt cameras thou.  With the faster primary in the camera the corrector require a deeper curve  on the corrector  and even with the Master Block method and thin glass , the. glass would break under the  deflective required to pull it against the Master surface. So they split the correction in 1/2 and put it on both sides.

 

                      - Dave 

Yes, agreed on everything. The only reason I was surprised was that the chromatic improvement gained by the 86% nz is at best marginal on the usual 8'' sct. But if it's no extra work I suppose why not.   David

[davidc135]

Well, it matters a lot which way the objective is facing. On turning it round SA increased 6 fold!. Partly due to the change in conjugate focal lengths.

 

One thing I may have got wrong is estimating the SA error. I hadn't taken into account the fact that the light source was stationary. Therefore the amount of ke travel was doubled comparing paraxial and marginal nulls so I could have doubled the actual errors.

 

I wonder if there is a handy, back of an envelope way of calculating error due to reducing object distance from an achromat.

 

David

Edited by davidc135, 15 June 2022 - 03:26 PM.

[MKV]

Well, it matters a lot which way the objective is facing. On turning it round SA increased 6 fold!. Partly due to the change in conjugate focal lengths.

 

One thing I may have got wrong is estimating the SA error. I hadn't taken into account the fact that the light source was stationary. Therefore the amount of ke travel was doubled comparing paraxial and marginal nulls so I could have doubled the actual errors.

 

I wonder if there is a handy, back of an envelope way of calculating error due to reducing object distance from an achromat.

Your setup can be raytraced. If you give me the focal length of the 50 mm binocular lens and the secondary radius of curvature, as well as the wavelength used. I will assume the achromat is the standard BK7/F2 combo. They are usually designed in the neighborhood of 550 and 587.6 nm. to work at or near infinity focus

 

Where did you see this test and what's the logic behind it? I know of this arrangement only as far as the test of a convex surface radius of curvature is concerned using an autostigmatic microscope, not the SA, conic constant or similar parameters.

 

But I will be happy to do a raytrace analysis for you.

 

PS When an object is brought  closer to a lens or a mirror  designed to work at infinity, the wavefront will become overcorrected. 

[luxo II]

Well, it matters a lot which way the objective is facing. On turning it round SA increased 6 fold!. Partly due to the change in conjugate focal lengths.

 

...

 

I wonder if there is a handy, back of an envelope way of calculating error due to reducing object distance from an achromat.

The mistake is assuming the lens on its own when used with finite conjugates has zero SA - it won't, and, you're using a  binocular objective (which does not have zero SA - by design), and you don't know the exact design of the doublet (ie glass types, radii, thicknesses), so raytracing is not an option.

 

The second issue - and its probably a showstopper for your test - is that the SA produced by a mirror also depends on the position of the conjugates. For example, a spherical mirror tested with a conjugate at the centre of curvature will produce zero SA, yet with an infinite conjugate it produces lots of SA, as we all know.

 

The alternative - which I think is what your test layout is trying to suggest - is the Dall or Ross null test - where, given the desired shape of a convex mirror the SA is computed for the test configuration, and a lens is designed that compensates for exactly that SA. Any remaining SA measured then indicates a defect. And this lens may have to be ground, polished and tested first to confirm it is as expected, otherwise your just introducing yet another unknown quantity.

 

https://www.telescop...s_null_test.htm

Edited by luxo II, 15 June 2022 - 07:54 PM.

[MKV]

The alternative - which I think is what your test layout is trying to suggest - is the Dall or Ross null test - where, given the desired shape of a convex mirror the SA is computed for the test configuration, and a lens is designed that compensates for exactly that SA.

 

https://www.telescop...s_null_test.htm

There is NO Dall or Ross null test for convex mirrors.

 

A Raytrace David's setup is very much possible by using a "Pefect Lens" of a given focal length option so that any resulting errors are attributable only to test optics and the method used.

 

Mladen

[davidc135]

The mistake is assuming the lens on its own when used with finite conjugates has zero SA - it won't, and, you're using a  binocular objective (which does not have zero SA - by design), and you don't know the exact design of the doublet (ie glass types, radii, thicknesses), so raytracing is not an option.

 

I realise that the binocular objective has SA with finite conjugates. The test accomodates the induced doublet SA as it is common to both setups, one including the tested convex mirror and the reference arrangement with a plane mirror at the shorter conjugate. The point of this test is to show whether or not the secondary is sufficiently aspheric to even begin to correct for coma and it's obvious that this 2ndry is a sphere apart froma trivial 1/4 wave or so.

My ray tracing query was to get to a ball park figure- am I looking at 3 waves or is it really 11/2. Not essential for my aim but I'd like to know.

 

The second issue - and its probably a showstopper for your test - is that the SA produced by a mirror also depends on the position of the conjugates. For example, a spherical mirror tested with a conjugate at the centre of curvature will produce zero SA, yet with an infinite conjugate it produces lots of SA, as we all know.

 

The test is arranged and the doublet's conjugates are adjusted such that the convex mirror's CoC is coincident with the doublet's shorter conjugate. Therefore, as you say, a spherical mirror contributes no SA. These conjugates are determined (with the 2ndry in place) when the source at A and it's reflected image are coincident.

 

A couple of practical advantages are that the slow cone of rays at A is easily tested, the correct positions for the convex or plane mirrors are easily found and and that all components are easily aligned. Virtually all of the 2ndry's aperture is covered. The 50mm F/4 or F/4.5 doublets lend themselves to the task very well.

 

The price paid is the nuisance of the inherent test SA but it's not really that relevant to the intended aim. 

 

The alternative - which I think is what your test layout is trying to suggest - is the Dall or Ross null test - where, given the desired shape of a convex mirror the SA is computed for the test configuration, and a lens is designed that compensates for exactly that SA. Any remaining SA measured then indicates a defect. And this lens may have to be ground, polished and tested first to confirm it is as expected, otherwise your just introducing yet another unknown quantity.

 

Putting the Dall or Ross test into the mix is getting too complicated.

 

Everything is common to the two optical trains in my test except for the possible asphericity of the convex mirror.

 

The test could be polished up and my 'exactly how much inherent test SA' question cleared up if the tester wants to get down to 1/10 wave accuracy but the 8'' Meade ACF' s 2ndry SA of around 3 waves wf would absolutely jump out if it was tested in this way.

 

David

 

https://www.telescop...s_null_test.htm

Edited by davidc135, 16 June 2022 - 03:09 AM.

[luxo II]

There is NO Dall or Ross null test for convex mirrors.

Yup, of course. I was about to get to that but wanted to see how much he knew.

 

I’d measure the actual SA of the lens first though in the intended configuration rather than rely on a raytrace as the lens details are not known.

 

@David - one aspect of this is that when trying to assess a test for a complete system to see if it’s better than 0.25 wave the errors (deviation from the design target) from each element are compounded. The resolution of the test itself has to be significantly better than the quality criterion you’re aiming at. Even if the test can discriminate 0.1, it still implies the complete scope could be anywhere from 0.15 (good) to 0.35 wave which is visually poor - especially if the worst errors are in the outermost 30% of the mirror.

Edited by luxo II, 16 June 2022 - 04:10 AM.

[MKV]

Yup, of course. I was about to get to that but wanted to see how much he knew

_________________

 

I’d measure the actual SA of the lens first though in the intended configuration rather than rely on a raytrace as the lens details are not known.

I think it's much easier and proper to ask, perhaps via PM. Just my 2-cents' worth.

___________________

 

Binocular objectives are pretty much standard commercial quality 1/2 wave optics because they are low power telescope and do not need to be better than that. 

 

I normally use Thorlabs or Melles-Griot and similar off-the shelf items. You find they are pretty much the same across the board, probably  even a notch better than your average mass-produced overseas lenses,  but that wold have to be established objectively. Here are the specs

 

https://www.thorlabs...ectgroup_id=120

 

And here's the lens I used (the specs are given with the previous link)

 

https://www.thorlabs...er=ACT508-200-A

 

Good enough for a raytrace analysis, maybe even too good. 

 

The configuration Dave gives is not a null test, unless the aperture is severely stopped down. A Hindle sphere is the tool to use unless he can have it Zygo tested at a nearby shop for pretty $$. 

 

The other option is to make a matching concave test plate, polish it and figure it dead-on spherical (nulled) with a knife-edge, and test the convex surface by contact interference (Newton's rings). For ATMs, convex surfaces are a very difficult subject to tackle, and require a lot of extra work and tooling. 

[davidc135]

Yup, of course. I was about to get to that but wanted to see how much he knew.

 

I'm familiar enough with them.

 

I’d measure the actual SA of the lens first though in the intended configuration rather than rely on a raytrace as the lens details are not known.

 

The knife edge travel after double pass with plane mirror in the correct configuration can be converted into SA with a simple formula:

 

SA = 28.4 ket/f² 

 

So, with long conjugate focal ratio of 15.5 and k.e.t of 26.5mm that becomes

 

SA = 28.4x26.5/240.25

      =3.13 waves double pass

 

Or half that per pass

 

But I don't know if the stationary source affects those figures. Ie half them or not. I don't think so but I'd like to know. That's where even an approx raytrace of a 'typical' doublet of the same spec would help.

 

 

@David - one aspect of this is that when trying to assess a test for a complete system to see if it’s better than 0.25 wave the errors (deviation from the design target) from each element are compounded. The resolution of the test itself has to be significantly better than the quality criterion you’re aiming at. Even if the test can discriminate 0.1, it still implies the complete scope could be anywhere from 0.15 (good) to 0.35 wave which is visually poor - especially if the worst errors are in the outermost 30% of the mirror.

 

As I said before a secondary that has been aspherised to correct coma will contribute of the order of 2.5 to 3 waves SA on reflection so only moderate accuracy is required for the test to give an answer. A roughly carried out Foucault test will not be very acurate. It's not the fault of the test. In the same way care and attention to detail will reduce errors in this test.

 

David

[davidc135]

I think it's much easier and proper to ask, perhaps via PM. Just my 2-cents' worth.

 

I'm sorry, Mladen. I hadn't read your post 65 properly with your kind offer to do a ray trace.

___________________

 

Binocular objectives are pretty much standard commercial quality 1/2 wave optics because they are low power telescope and do not need to be better than that. 

 

I normally use Thorlabs or Melles-Griot and similar off-the shelf items. You find they are pretty much the same across the board, probably  even a notch better than your average mass-produced overseas lenses,  but that wold have to be established objectively. Here are the specs

 

https://www.thorlabs...ectgroup_id=120

 

And here's the lens I used (the specs are given with the previous link)

 

https://www.thorlabs...er=ACT508-200-A

 

Good enough for a raytrace analysis, maybe even too good. 

 

The configuration Dave gives is not a null test, unless the aperture is severely stopped down. A Hindle sphere is the tool to use unless he can have it Zygo tested at a nearby shop for pretty $$. 

 

It doesn't need to be a null to do it's job. At F/15 the knife edge and a ruler is pretty sensitive. Still, it's true that the outer part of the aperture is difficult to measure.

 

The other option is to make a matching concave test plate, polish it and figure it dead-on spherical (nulled) with a knife-edge, and test the convex surface by contact interference (Newton's rings). For ATMs, convex surfaces are a very difficult subject to tackle, and require a lot of extra work and tooling. 

 

The matching concave plate would work well but it's a fair bit of effort. For the simple aim of this thread just chuck a few common components onto the optical bench and in 20 minutes you have the answer. Many reading this thread will have a beam splitting cube but, if not, they're inexpensive.

 

David

Edited by davidc135, 16 June 2022 - 09:05 AM.

[Gleb1964]

Whilst this experiment featured a Meade secondary I doubt if Celestron's version would show anything different. It's easy to do so why not?
 
David
 
PS What's this test called?

David.

You have done everything correct. Doublet lens has some spherical aberration, but it is common in both configurations and it doesn't matter much if you are comparing configurations.
Well done.

Gleb

[davidc135]

Your setup can be raytraced. If you give me the focal length of the 50 mm binocular lens and the secondary radius of curvature, as well as the wavelength used. I will assume the achromat is the standard BK7/F2 combo. They are usually designed in the neighborhood of 550 and 587.6 nm. to work at or near infinity focus

 

Where did you see this test and what's the logic behind it? I know of this arrangement only as far as the test of a convex surface radius of curvature is concerned using an autostigmatic microscope, not the SA, conic constant or similar parameters.

 

But I will be happy to do a raytrace analysis for you.

 

PS When an object is brought  closer to a lens or a mirror  designed to work at infinity, the wavefront will become overcorrected. 

Somehow I missed your offer earlier and caught up a bit late. Thank you. Pm sent.  David

Edited by davidc135, 16 June 2022 - 02:05 PM.

[davidc135]

The SA can cause trouble in estimating Foucault ke positions especially for the outer part of the aperture as mentioned above and so I wondered if there was a better way to measure the marginal null.

 

In the poor man's caustic test longitudinal measurements are 3x the Foucault ke travel varying as 3r²/2R for a paraboloid at its centre of curvature or proportional to 3r² for any smooth conic.

 

 As the knife edge first contacts and crosses the caustic a small shadow forms and grows with a corresponding dot of light lingering on the opposite side. I took the required position to be (sq.rt 0.33).r or 0.57x25.4 =14.45mm from centre and marked the position with a sharpie.

 

The length of the long conjugate A increased slightly to 28.5ins, F/14.5. Using the sharpie mark as a guide I found ke travel figures of 34mm with or without the 2ndry in place which corresponded to a greater SA of 4.75 waves under-correction in double pass.

 

I feel these results are more accurate than the first set but this amount of SA  makes any kind of accurate testing difficult. 2 objectives placed face to face of suitable diameter and F ratios with A and B as their infinity foci would have little or no SA and would be more accurate although trickier to set up.

 

David