Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Reliable mirror quality validation

  • Please log in to reply
24 replies to this topic

#1 Lucullus

Lucullus

    Apollo

  • -----
  • topic starter
  • Posts: 1,059
  • Joined: 14 Feb 2012

Posted 27 January 2021 - 03:13 AM

In order to be confident of a specific mirror quality one is said to cross-check it with different tests and to validate the results from each other.
How certain can one be of the mirror quality if only checking one's Foucault test (slit with light source co-moving with the knife-edge) with a Lyot test? I read somewhere a Lyot test is basically the same as a Foucault test, just with a larger slit width and showing tinier errors with much more contrast, but the same could be achieved with the Foucault test using a camera and longer exposure times to detect the same level of contrast. Is this true and you would need a test which is totally different than the Foucault in order to verify the Foucault test result?
How confident can you be of your mirror quality if only using two tests for verification? What about three tests?

#2 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,373
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 27 January 2021 - 07:34 AM

One way is to calibrate your Foucault against a known optic. If your results match the certified values then you can say your Foucault is pretty good. The problem is where to find a certified optic.

 

A better method is to test your mirror using the autcollimation test. It's almost fail-proof. The down side is that you need a fairly large optical flat.

 

A third method is to test your mirror interferometrically with a simply IF such as a Point Diffraction type or the more popular Bath, and to take many igrams, rotate and average wavefronts using specialized software (which happens to be free).

 

The Lyot test is a test of the surface roughness. If done properly, it can "see" micro  ripple down to the Anstrom level. A Foucault test will show more contrast/detail if the light source  is small, especially if ti approaches the width of an Airy disc (usually a few microns). For that you need a spatial filter otherwise the tiny slit cannot be properly illuminated. The Foucault test can never show roughness to the same extent as the Lyot test.



#3 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 25,435
  • Joined: 22 Feb 2010

Posted 27 January 2021 - 09:57 AM

In order to be confident of a specific mirror quality one is said to cross-check it with different tests and to validate the results from each other.
How certain can one be of the mirror quality if only checking one's Foucault test (slit with light source co-moving with the knife-edge) with a Lyot test? I read somewhere a Lyot test is basically the same as a Foucault test, just with a larger slit width and showing tinier errors with much more contrast, but the same could be achieved with the Foucault test using a camera and longer exposure times to detect the same level of contrast. Is this true and you would need a test which is totally different than the Foucault in order to verify the Foucault test result?
How confident can you be of your mirror quality if only using two tests for verification? What about three tests?

I would not use the lyot test to check a ke test... the quickest test is a ronchi test against a computer rendering of a ronchi.

 

another is testing against a flat

 

or 

 

other nulling optics

 

or 

 

star test... in the field where you actually use the optic

 

a bath if.

 

anything but the lyot. ha ha ha


  • Venetia2004 and Jeff B like this

#4 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 25,435
  • Joined: 22 Feb 2010

Posted 27 January 2021 - 10:23 AM

but really looking at a ronchi 1st and matching against a computer rendering is how I do it.

 

then ke test with zonal measurements

 

then nulling optics

 

then a star test...

 

I can't find my bath if... LOGAN!!! lol.gif


  • Jeff B likes this

#5 sw196060

sw196060

    Viking 1

  • *****
  • Posts: 615
  • Joined: 16 Apr 2020

Posted 27 January 2021 - 01:19 PM

I like testing against a star using my Ronchi grid eyepiece.
  • siriusandthepup, Pinbout and sopticals like this

#6 MitchAlsup

MitchAlsup

    Fly Me to the Moon

  • -----
  • Posts: 5,160
  • Joined: 31 Aug 2009

Posted 27 January 2021 - 03:35 PM

Much of the time you can take the uncoated mirror into a testing telescope (permantely pointing at polaris) and determine if the primary is adequately parabolized, look for zones, ... collimated light from stars is almost as perfect as collimated light gets.


  • SandyHouTex likes this

#7 dan_h

dan_h

    Gemini

  • *****
  • Posts: 3,052
  • Joined: 10 Dec 2007

Posted 28 January 2021 - 10:43 AM

<< How confident can you be of your mirror quality if only using two tests for verification? What about three tests? >> 

 

 

Once you start using multiple tests to validate the mirror surface, you are no longer testing the mirror. You are testing your test methods to see where your errors come from.  

 

Knowing your test equipment/methods including their weaknesses and drawbacks is part of the knowledge base that comes with experience. 

 

dan


  • siriusandthepup and SandyHouTex like this

#8 Lucullus

Lucullus

    Apollo

  • -----
  • topic starter
  • Posts: 1,059
  • Joined: 14 Feb 2012

Posted 29 January 2021 - 03:32 PM

<< How confident can you be of your mirror quality if only using two tests for verification? What about three tests? >>


Once you start using multiple tests to validate the mirror surface, you are no longer testing the mirror. You are testing your test methods to see where your errors come from.

Knowing your test equipment/methods including their weaknesses and drawbacks is part of the knowledge base that comes with experience.

dan


Wonderfully expressed! Of course. I have to clarify. I assume that a specific test reveals only some of the most prominent weaknesses of another test.
If using these tests: Foucault, Ronchi, Lyot, Bath interferometer. What specific weakness do they reveal from the others? And how to you subtract an uncovered error and determine the increased mirror quality without the uncovered error?

#9 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 25,435
  • Joined: 22 Feb 2010

Posted 29 January 2021 - 04:37 PM

<< How confident can you be of your mirror quality if only using two tests for verification? What about three tests? >> 

 

 

Once you start using multiple tests to validate the mirror surface, you are no longer testing the mirror. You are testing your test methods to see where your errors come from.  

 

Knowing your test equipment/methods including their weaknesses and drawbacks is part of the knowledge base that comes with experience. 

 

dan

I completely disagree.

 

I start with a ronchi - see the overall shape, any gross irregularities

 

I move null tests so no measurement just quality...but witth experience I know if the null is good the mirror has a level of completion that would be acceptable.

 

then ke test to verify better than whatever spec i desire.

 

more than 1 test, not testing tests. but if 1 test doesn't agree, something is wrong... we all make mistakes even with simple testing setups.


Edited by Pinbout, 29 January 2021 - 04:37 PM.

  • dan chaffee and sopticals like this

#10 pstarr

pstarr

    Skylab

  • *****
  • Posts: 4,358
  • Joined: 17 Sep 2004
  • Loc: NE Ohio

Posted 30 January 2021 - 05:30 PM

I completely disagree.

 

I start with a ronchi - see the overall shape, any gross irregularities

 

I move null tests so no measurement just quality...but witth experience I know if the null is good the mirror has a level of completion that would be acceptable.

 

then ke test to verify better than whatever spec i desire.

 

more than 1 test, not testing tests. but if 1 test doesn't agree, something is wrong... we all make mistakes even with simple testing setups.

One test on the

Hubble mirror.....what could go wrong?


  • Pinbout likes this

#11 Dale Eason

Dale Eason

    Surveyor 1

  • *****
  • Posts: 1,854
  • Joined: 24 Nov 2009
  • Loc: Roseville,Mn.

Posted 30 January 2021 - 08:53 PM

In order to be confident of a specific mirror quality one is said to cross-check it with different tests and to validate the results from each other.
How certain can one be of the mirror quality if only checking one's Foucault test (slit with light source co-moving with the knife-edge) with a Lyot test? I read somewhere a Lyot test is basically the same as a Foucault test, just with a larger slit width and showing tinier errors with much more contrast, but the same could be achieved with the Foucault test using a camera and longer exposure times to detect the same level of contrast. Is this true and you would need a test which is totally different than the Foucault in order to verify the Foucault test result?
How confident can you be of your mirror quality if only using two tests for verification? What about three tests?

The Lyot test is a poor test to use to check that the mirror has been properly corrected for SA.  In fact it may be almost impossible to do.  However the Lyot test that I know is not made with a slit opening but instead a knife edge made of soot that is progressively thinned.   When at the right density of soot then the small surface features show up.  The usual wisdom is that or 99% atm mirror work Foucault will also see features that matter and can be used to the same purpose.  Lyot can see smaller features still but they usually  will not matter in all be the most stringent application that need the max contrast like a chronograph.

 

All testing is most subject to human error and requires experience.  For the beginner mirror maker it is good to learn the Foucault and Ronchi tests.  They can tell you a lot of what you need to know to make a good telescope mirror.   They have been used for years.  Each has their appeal and areas of best sensitivity.  Learn them both before moving on.  Finally use the star test which can be problematic if your skies are mostly cloudy and/or always have poor seeing.   If you lucky enough to live where that is not the case then focus on that as your second test while trying to calibrate your Foucault or ronchi skills.

 

Dale


Edited by Dale Eason, 31 January 2021 - 05:46 PM.

  • dan_h and Pinbout like this

#12 dan_h

dan_h

    Gemini

  • *****
  • Posts: 3,052
  • Joined: 10 Dec 2007

Posted 31 January 2021 - 09:23 PM

I completely disagree.

 

I start with a ronchi - see the overall shape, any gross irregularities

 

I move null tests so no measurement just quality...but witth experience I know if the null is good the mirror has a level of completion that would be acceptable.

 

then ke test to verify better than whatever spec i desire.

 

more than 1 test, not testing tests. but if 1 test doesn't agree, something is wrong... we all make mistakes even with simple testing setups.

I would have to agree that this is a reasonable way to progress through the finishing of a mirror.   But I think the OP wanted to know what tests were the definitive tests  that you use to claim a mirror as finished.   Foucault and then star test. Done. 

 

Ronchi is a great fast test that is easily used to determine the overall shape and can continue to be used to see the end result of polishing efforts.  ie is the shape improving or no? At some point Ronchi is less effective if you want quantitative results. Ronchi is also difficult to use to determine edge defects due to the diffraction effects that appear to stretch the bands at the edge.

 

Null is great once setup but requires auxillary optics. A DPAC test will null but you need an appropriately sized flat or collimator of suitable aperture. The Dall null test requires a compensating lens instead of a flat. Both tests are not as fast as Ronchi and require more setup and time to reach proper thermal equilibrium in order to benefit from the improved sensitivity.  

 

Finally, the Foucault test to determine proper correction and overall figure. But once at this point, there is no need to return to the Ronchi test to determine overall shape and gross zonal errors.  Gross errors are not what is being examined at this point.  You wouldn't return to a spherometer to verify focal length on a nearly finished mirror.  Why would you back up to Ronchi to look for gross errors and zones when Foucault is more accurate and provides quantitative data.  

 

Sure one may make a mistake. Inside focus, outside focus, bands curved or straight, knife edge from the left or right. There is room for error.  Crap can happen.   But if any one test of three doesn't agree with the others, something with the test is wrong, or with the operator's technique. Either way, you are not testing the mirror.  

 

dan



#13 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 25,435
  • Joined: 22 Feb 2010

Posted 31 January 2021 - 10:52 PM

An oil flat is quick and easy if you can turn you scope/optical tube vertical, pointing down. 
 

https://youtube.com/...hyN0oJICs2L6aP7
 

quick and easy to see it’s junk.


Edited by Pinbout, 31 January 2021 - 10:56 PM.


#14 Lucullus

Lucullus

    Apollo

  • -----
  • topic starter
  • Posts: 1,059
  • Joined: 14 Feb 2012

Posted 02 February 2021 - 08:26 AM

Ok, if I underst

 

I would have to agree that this is a reasonable way to progress through the finishing of a mirror.   But I think the OP wanted to know what tests were the definitive tests  that you use to claim a mirror as finished.   Foucault and then star test. Done. 

 

Ronchi is a great fast test that is easily used to determine the overall shape and can continue to be used to see the end result of polishing efforts.  ie is the shape improving or no? At some point Ronchi is less effective if you want quantitative results. Ronchi is also difficult to use to determine edge defects due to the diffraction effects that appear to stretch the bands at the edge.

 

Null is great once setup but requires auxillary optics. A DPAC test will null but you need an appropriately sized flat or collimator of suitable aperture. The Dall null test requires a compensating lens instead of a flat. Both tests are not as fast as Ronchi and require more setup and time to reach proper thermal equilibrium in order to benefit from the improved sensitivity.  

 

Finally, the Foucault test to determine proper correction and overall figure. But once at this point, there is no need to return to the Ronchi test to determine overall shape and gross zonal errors.  Gross errors are not what is being examined at this point.  You wouldn't return to a spherometer to verify focal length on a nearly finished mirror.  Why would you back up to Ronchi to look for gross errors and zones when Foucault is more accurate and provides quantitative data.  

 

Sure one may make a mistake. Inside focus, outside focus, bands curved or straight, knife edge from the left or right. There is room for error.  Crap can happen.   But if any one test of three doesn't agree with the others, something with the test is wrong, or with the operator's technique. Either way, you are not testing the mirror.  

 

dan

Hmm, from my two telescope building books I believe to know that one is ok to start polishing if the pencil test is successful in that the lines on the mirror and tool over the whole surfaces disappear equally and simultaneously. With this test one is enough spherical accurate to start polishing when reaching below 9micron fine grit powder (many prefer to go to 5 micron or even 3 micron to reduce the polishing time). So the Ronchi test is not necessary at all if the pencil test is successful? What advantages does the Ronchi have over the pencil test?
 



#15 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 25,435
  • Joined: 22 Feb 2010

Posted 02 February 2021 - 10:02 AM

 

So the Ronchi test is not necessary at all if the pencil test is successful? What advantages does the Ronchi have over the pencil test?

Ronchi is after polishing and you pass a reflection test then a laser test on the polish

 

at 12um do a reflection test 

 

https://youtube.com/...21IUi_tyFmSieQr

 

Even after a flash polish with polishing pads 

 

https://youtube.com/...21IUi_tyFmSieQr


Edited by Pinbout, 02 February 2021 - 10:05 AM.


#16 SandyHouTex

SandyHouTex

    Fly Me to the Moon

  • *****
  • Posts: 5,519
  • Joined: 02 Jun 2009
  • Loc: Houston, Texas, USA

Posted 03 February 2021 - 08:20 PM

On the 200 inch they used the Focault test and a Hartmann mask, and it came out virtually perfect.

 

Too many people today want to use lasers with this, or flats with that, and it really doesn’t improve anything if you know what you’re doing in the first place.  Only amatuers constantly “fiddle”.


  • John Lightholder and dan_h like this

#17 BGRE

BGRE

    Gemini

  • *****
  • Posts: 3,221
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 03 February 2021 - 08:48 PM

The NOTT primary was figured (in a man made cave carved out of the bedrock under Turku) using the measurements obtained with an improved (more accurate) version of the Hartmann test.

#18 Lucullus

Lucullus

    Apollo

  • -----
  • topic starter
  • Posts: 1,059
  • Joined: 14 Feb 2012

Posted 04 February 2021 - 10:32 AM

The NOTT primary was figured (in a man made cave carved out of the bedrock under Turku) using the measurements obtained with an improved (more accurate) version of the Hartmann test.

An improved Hartmann mask on a conventional Foucault test you mean? If yes, please elaborate on the mask.
 



#19 duck

duck

    Vostok 1

  • -----
  • Posts: 198
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 04 February 2021 - 04:25 PM

Interferograms no better that Foucault?  Man, did I waste a bunch of money and time on test optics.



#20 Steve Dodds

Steve Dodds

    Owner - Nova Optical

  • *****
  • Vendors
  • Posts: 798
  • Joined: 02 Apr 2007
  • Loc: Utah

Posted 04 February 2021 - 04:53 PM

Interferograms no better that Foucault?  Man, did I waste a bunch of money and time on test optics.

****!!!  Interferometers measure every aberration with thousands of points..  Foucault measures a half dozen points in a straight line across your mirror. And misses most aberrations not related to spherical aberration..



#21 BGRE

BGRE

    Gemini

  • *****
  • Posts: 3,221
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 04 February 2021 - 09:04 PM

An improved Hartmann mask on a conventional Foucault test you mean? If yes, please elaborate on the mask.


They used the interferometric variant of the Hartmann test where the image sensor (without lens) is located in a position where the light from adjacent screen apertures overlap and interfere.
Initially the detector was a photographic plate, the positions of the interference maxima were laboriously measured by hand and before using a computer to calculate the corresponding wavefront error.
There are some papers from VATT that detail the math behind the analysis. Long exposures are used to average out the effects of air turbulence.
The method is about 5x more accurate than other Hartmann techniques.

A point source at the CoC of the test surface is used together with suitable array of apertures in an opaque screen at the test surface.

Its also possible to image the exit pupil onto a smaller aperture array and follow this with another lens.
The errors of the auxiliary optics are calibrated using a point source at the front focus of the lens that images the exit pupil of the system under test onto the aperture array.

#22 Lucullus

Lucullus

    Apollo

  • -----
  • topic starter
  • Posts: 1,059
  • Joined: 14 Feb 2012

Posted 05 February 2021 - 01:31 PM

So the NOTT and VATT were both tested this way and the capital letters are not a typo?

#23 BGRE

BGRE

    Gemini

  • *****
  • Posts: 3,221
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 05 February 2021 - 05:18 PM

Oops that should have been NOT

NOT = Nordic Optical Telescope
NOTT is old Norse for night.

VATT = Vatican Advanced Technology telescope

The VATT wasn't tested that way during figuring however a wavefront sensor using the Interferometric Hartmann test on a projected pupil was used.
Such a wavefront sensor was also used on the NOT and the MMT (although they abandoned it as they needed to modify it to use more apertures to cope with poor seeing).

The interferometric Hartmann test is essentially the 2D version of the Gardner-Bennett test which uses a set of uniformly spaced apertures across a diameter of the test surface.
In principle the dynamic range of the latter can be improved by either using more apertures or by using smaller apertures and only using a single adjacent pair of apertures at a time.

Edited by BGRE, 05 February 2021 - 08:31 PM.


#24 Augustus

Augustus

    Fly Me To The Moon

  • *****
  • Posts: 10,489
  • Joined: 26 Dec 2015
  • Loc: Stamford, Connecticut

Posted 05 February 2021 - 05:27 PM

On the 200 inch they used the Focault test and a Hartmann mask, and it came out virtually perfect.

I've heard it's got on the order of 2 waves of aberrations....



#25 RLK1

RLK1

    Viking 1

  • -----
  • Posts: 808
  • Joined: 19 Apr 2020

Posted 05 February 2021 - 06:58 PM

I've heard it's got on the order of 2 waves of aberrations....

https://www.cloudyni...ve/?hl= palomar




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics