92 replies to this topic

[Gleb1964]

I used DFTFRinge for a D500 F3 mirror and simulated a perfect 100% paraboloid, one which was only 99% corrected, and one 98%.

 

These are the DFTFRinge results:

 

Conic  Strehl   RMS    PV

 

-1.000  0.999  0.004  λ/71

-0.990  0.930  0.043  λ/7

-0.980  0.752  0.085  λ/3

..

Mladen,

 

you probably did something wrong.

 

Simulating a single pass star test for D500 F3 mirror (at 632.8nm), I am getting:

 

Conic  Strehl   RMS    PV

 

-1.000  1.000  0.000  λ/∞

-0.990  0.959  0.035  λ/12.7

-0.980  0.845  0.071  λ/6.3

 

The difference in the simulated star images is quite visible. Here are star test images with realistic 30% central obscuration shadow, defocus 8 waves:

 

[MKV]

Mladen,

 

you probably did something wrong.

 

Simulating a single pass star test for D500 F3 mirror (at 632.8nm), I am getting:

 

Conic  Strehl   RMS    PV

 

-1.000  1.000  0.000  λ/∞

-0.990  0.959  0.035  λ/12.7

-0.980  0.845  0.071  λ/6.3

 

The difference in the simulated star images is quite visible. Here are star test images with realistic 30% central obscuration shadow, defocus 8 waves:

Thanks, Gleb. Great job I don't use DFTFRinge very often and have no clue how to simulate single pass. I thought DFTFRinge defaults to testing at ROC.

 

Can you post your mirror configuration that shows single pass setup, please? Thanks!

 

Anyway, I'm glad you decided to do this because the numbers I got seemed a little severe. Your figures look better, but still show how touchy these mirrors are. The patterns from conics -0.99 and -0.98 look almost identical, and those in-between, such as -0.981, -0.984, etc. would be a continuum that would be practically impossible to differentiate.

 

Seems like the assessment would have to say the conic is between -0.98 and -0.99 and the PV error will be between 13 and 6 waves, yeah and your Strehl anywhere from 0.845 and 0.959. -- that's a lot of uncertainty, imo.

 

So, I would say, the test is a "generous" approximation but nothing even close to "spot-on," as some are reporting. Maybe in smaller and slower mirrors this a little better. 

 

And let's not forget that simulation is as good as it gets -- theoretically speaking. Reality will always fall short.

 

So, in the end, instead of clarifying how did Suiter come up with numbers as small as λ/80 on sensitivity, I guess more research is needed to discover where this originated and why.

[Gleb1964]

"Single pass" is in my head, I don't see any special setting in DFTFringe. DFTFringe is a flexible tool, you can use it to generate star images in a single or double pass (or any multiple pass), with central obstruction or without.

To generate a double pass star image I would keep in mind that 0.99 conic in the double pass would by twice worse, I should generate 0.98 conic instead. 

 

Here is an example of star images with mirror conics 1.00, 0.99 and 0.98 how it would be double pass, where the wavefront conic would be 1.00, 0.98 and 0.96 instead. No central obscuration, defocus +/-3.6 waves. 

I can mention that I am not quite happy how DFTFringe is choosing defocus offset, because it seems referring to the paraxial focus point, while it should use the best focus (at minimum rms focus).

 

Edited by Gleb1964, 15 April 2025 - 08:10 AM.

[MKV]

Thank you Gleb. DFTFRinge was designed to test mirrors tested at the radius of curvature  -- for mirror makers. It doesn't let you place the source at infinity.                                                                                                                                                                                                                                                                                                                                 

[Gleb1964]

DFTFringe does generate star images, Foucault and Ronchi test for converging wavefronts.

 

The wavefront is either calculated from an interferogram or generated by simulating Zernike terms.

 

The wavefront can be corrected for some amount of spherical aberration, known as an "artificial null", which accounts for testing aspheres from the center of curvature.

Choose of correction for "artificial null" is a tool to simulate ether remote source ether source at any position. 

 

I don't see the point in limiting DFTFringe to testing mirrors only at the radius of curvature.

Edited by Gleb1964, 15 April 2025 - 08:06 AM.

[Dale Eason]

MKV, Since DFTFringe is a program to analyze interferograms  testing at other than ROC is done by changing the waves per fringe value and turning on/off the artificial null as necessary.  It was designed to test at ROC and for any usual setup to null the mirror like auto collimation or ROS null.  

 

Edited by Dale Eason, 15 April 2025 - 12:17 PM.

[Dale Eason]

Thank you Gleb. DFTFRinge was designed to test mirrors tested at the radius of curvature  -- for mirror makers. It doesn't let you place the source at infinity.                                                                                                                                                                                                                                                                                                                                 

Actually DFTFringe star test simulation automatically sets the source at infinity.  

[MKV]

 

MKV, Since DFTFringe is a program to analyze interferograms  testing at other than ROC is done by changing the waves per fringe value and turning on/off the artificial null as necessary. 

What if it's a lens? There's only one reflection, hence waves per fringe number should be 1.

 

Actually DFTFringe star test simulation automatically sets the source at infinity. 

Can you show me a mirror configuration) of  a ≈ 6" F8 (D150 mm R 2400 mm) sphere cc = 0), and object at infinity? This should yield a PV of  ≈ 0.26, RMS of ≈0.08 and a Strehl of ≈0.76, while the best fitting conic should remain 0.

Edited by MKV, 15 April 2025 - 11:22 PM.

[MKV]

I don't see the point in limiting DFTFringe to testing mirrors only at the radius of curvature.

That's how it was designed -- for mirror makers, to test the figure when making mirrors.

[Dale Eason]

What if it's a lens? There's only one reflection, hence waves per fringe number should be 1.

 

Can you show me a mirror configuration) of  a ≈ 6" F8 (D150 mm R 2400 mm) sphere cc = 0), and object at infinity? This should yield a PV of  ≈ 0.26, RMS of ≈0.08 and a Strehl of ≈0.76, while the best fitting conic should remain 0.

You can set the waves per fringe to whatever needs it to be.

Not in this thread.  You start another thread and I will.  

[Gleb1964]

In DFTFringe, the scale factor is only involved in data reduction from an interferogram fringes to the wavefront. The user can set the scale factor arbitrarily, considering the testing configuration or other factors that user have in mind.

 

"Star source at infinity" is only an initial conceptual idea.

For example, testing a spherical mirror from the radius, gives a nulled wavefront. But I keep in mind, that the Star test (and Ronchi/Foucault) would show the result for source located at the Center of curvature, not at infinity.

Edited by Gleb1964, 16 April 2025 - 05:48 AM.

[Dale Eason]

 But I keep in mind, that the Star test (and Ronchi/Foucault) would show the result for source located at the Center of curvature, not at infinity.

DFTFringe Star test shows source located at infinity.  Ronchi/Foucault is at Center of curvature.

[MKV]

Mladen,

 

you probably did something wrong.

 

Simulating a single pass star test for D500 F3 mirror (at 632.8nm), I am getting:

 

Conic  Strehl   RMS    PV

 

-1.000  1.000  0.000  λ/∞

-0.990  0.959  0.035  λ/12.7

-0.980  0.845  0.071  λ/6.3

Gleb, I looked at your results and decided to do a single-pass raytrace at infinity and compare both sets of numbers. For simplicity, I did so only for the conic of -0.99. I also noticed that you used the HeNe wavelength of 632.8 nm and I originally used 546.1 nm (because DFTFringe defaults to that), so I repeated the trace in 632.8 nm as well.  Here's the comparison table

 

 Conic     Strehl       RMS       PV       λ/x         λ nm         Poster         S/W ________________________________________________________________________________________

-0.990      .930       0.043     .151      λ/7         546.1         MKV       DFTFRinge
-0.990      .922       0.045     .145      λ/7         632.8         MKV          OSLO

-0.990      .959       0.033     .079      λ/13        632.8       Gleb1964    DFTFringe 

  

A couple of things here stand out. (1) Difference between DFTFRinge and OSLO, and (2) Conic. 

 

My trace in both wavelengths, made in both OSLO and DFTFRinge agree. Yours doesn't agree with either by a significant margin. 

[duck]

is there a re-focus issue?

[MKV]

is there a re-focus issue?

I am sure Gleb did take that into account. For such a large and fast mirror, refocus to (minimum RMS OPD) is a must. Otherwise the Strehl drops to 0.22!

[BGRE]

One of the default merit functions in Zemax minimises the rms OPD.

This merit function is the one usually used.

In unusual cases such as optimising Fizeau collimators, minimising the ray slope error is required.

The default option of minimising the focal spot diameter accomplishes this if the collimator efl is fixed.

[Gleb1964]

I am sure Gleb did take that into account. For such a large and fast mirror, refocus to (minimum RMS OPD) is a must. Otherwise the Strehl drops to 0.22!

Sorry for confusion, Mladen.

 

I was using Annulus Zernike to produce an example of star test with the 30% central obstruction. I realize, that was twice reducing the amplitude and rms, but all simulation data are correct otherwise. 

 

Second time I have generate data without central obstruction, so the data also correct.

 

What I did wrong is that I have compared my obstructed data with yours without obstruction, that was incorrect comparison, sorry for that.  

Edited by Gleb1964, 16 April 2025 - 06:04 PM.

[MKV]

Sorry for confusion, Mladen.

 

I was using Annulus Zernike to produce an example of star test with the 30% central obstruction. I realize, that was twice reducing the amplitude and rms, but all simulation data are correct otherwise. 

 

Second time I have generate data without central obstruction, so the data also correct.

 

What I did wrong is that I have compared my obstructed data with yours without obstruction, that was incorrect comparison, sorry for that.  

It's all good, Gleb. No problem. :o)