22 replies to this topic

[BGRE]

Despite the apparently common misconception that DFTFringe is only suited for testing paraboloids at their CoC, this isn't correct.

Flats ellipsoidal mirrors, hyperboloidal mirrors, even Zernike surfaces etc. and lenses can also be tested interferometrically using DFTFringe to analyse the interferograms.

 

I'll let Dale cover issues such as setting parameters such as waves/fringe etc as well as the simpler cases such as interferometric autocollimation testing of lenses, etc.

Edited by BGRE, 16 April 2025 - 02:52 AM.

[Dale Eason]

Thank you Bruce.  I think it is only MKV that has that misconception.  I keep telling him that is not the case.

 

DFTFringe just analyzes interferograms.  It has optional features that let the user analyze those interferograms as needed.  It has one special feature that many analysis programs lacked back when OpenFringe  and FringeXp was first published.  That is the ability to set a software null that can be used to remove the SA from testing paraboloids at their CoC.  That feature is optional and does not have to be set.  But is usually set as that is one of the most common setups for testing a mirror with an interferometer by ATM's. 

[Gleb1964]

..  That is the ability to set a software null that can be used to remove the SA from testing paraboloids at their CoC. 

.. 

Not only paraboloids but any conic asphere tested from the CoC. More, it is possible to manipulate software null to use it more thinkable situation when you need to remove some amount of SA which related to the test configuration. 

[BGRE]

DFTFringe allows a wavefront to be entered by setting the amplitude of a number Zernike components.

This combined with the ability to subtract wavefronts, allows numeric Nulls for surfaces and setups other than a paraboloid tested at its CoC. Raytracing may be used to find the amplitudes of the Zernike terms for such setups. The Interferometry group wiki has some formulae for the SA Zernike amplitudes seen when testing other conicoidal surfaces at their CoC etc.

 

Another application is to use a sphere to null most of the SA when testing a very large fast paraboloidal or other  conicoidal surface and enter the Zernike term amplitudes of the residual wavefront error as a custom numeric null. Some care has to be taken in setting up the raytrace so that the residual seen is identical to that seen by the interferometer. 

 

Another application using wavefront subtraction allows using say a 1/4 wave flat to measure a test surface with greater accuracy than a 1/4 wave. The residual calibration error of such an interferometer can be measured using a liquid flat for example and subtracted from the result obtained when measuring the test flat. If the residual error is stable such calibration need only be done infrequently.

If the test system axis is horizontal a Ritchey Common setup can be used to calibrate errors in the reference surface in situ.

Such a calibration also uses wavefront subtraction to correct for the residual errors of the interferometer and the Ritchey-Common return sphere.

 

Other substitution methods are possible where a reference surface is compared with a test surface by measuring the test and reference surfaces sequentially with the same interferometer. PDI, Twyman-Green, Fizeau and Bath interferometers can be used for such substitution measurements. Care has to be taken not to disturb interferometer adjustments (including but not limited to camera focus) between sequential tests of the test and reference surfaces.

 

For example, a vertical axis Fizeau may use a PCX lens as a combined comparison flat and collimator which is used to compare a liquid flat with a test flat using the substitution method. A fixed airgap is used so that the test surface to PCX lens flat surface is identical to the liquid flat to PCX lens flat surface.

 

Such a PCX lens with the flat surface of the PCX lens adjacent to the test surface can work well if the PCX lens is slow enough and its diameter small enough without the convex surface of the PCX lens requiring aspherisation, provided that the airgap isn't too large.

Gravitational sag will distort the reference surface, but this is effect is the same when measuring both the test and reference surfaces.

[Gleb1964]

A few examples of how fringe scale factor can be used in reduction of interferometry data:

 

 

 

 

 

I don't see why DFTFringe or other interferogram reduction software would be limited by the testing configuration illustrated above. Users just need to understand what is happening; any limitations are likely only "in their heads." 

[Gleb1964]

Taking here question from another thread:
 

.. 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.                                                                                                                                                                                                                                                                                                                                 

Ok, the main purpose DFTFringe was developed is to support interferograms testing done from the radius. The picture below is illustrating the use of the digital null concept. 

 

Interferogram is obtained from the mirror radius, what would be the wavefront for infinity remote object? 

 

The wavefront for an object at infinity can be achieved by subtracting a certain amount of spherical aberration (SA), known as "digital null". (!)

 

 

(!) - Mention, that concept of the digital null in DFTFringe is not accounting for the change of incidence angle and for high order SA terms, which can be significant for the large fast aspheres.

Edited by Gleb1964, 16 April 2025 - 06:07 PM.

[MKV]

Dear educators :o), please remember that verbosity and complexity is an enemy of learning.

 

Let's just start with the fact that the first prerequisite for using DFTFRimge is to specify  "Mirror configuration." The program cannot run without specifying the "mirror" as the ubiquitous optical system.

 

 

The second prerequisite is an interferogram or just igram, which is an interferometric image of a wavefront. An igram can be real or a simulated. In both cases, it represents the image of an exit pupil wavefront of an optical system, regardless how simple or complex the system is. For those who don't know, this is what an igram looks like. 

 

 

Now, what prompted this thread was my (rather simple, basic) question on another topic how to enter mirror parameters for a simple 6-inch F/8 spherical mirror tested against a star-like source very far away (essentially at infinity), and, as y'all can see -- despite some very interesting posts -- not a single one answers that simple question! Folks, it only takes a minute or less to fill out the "Mirror configuration" box! Why is that a problem? 

 

So, for the willing, :o) one more time, can someone pleas enter the Mirror configuration for a mirror with a diameter of 150 mm, radius of curvature of 2400 mm and a conic constant of 0, with an object at infinity that will give a simulated igram with expected errors of such a mirror tested against a star (single-pass)?  

Edited by MKV, 16 April 2025 - 01:32 PM.

[Dale Eason]

 

 

So, for the willing, :o) one more time, can someone pleas enter the Mirror configuration for a mirror with a diameter of 150 mm, radius of curvature of 2400 mm and a conic constant of 0, with an object at infinity that will give a simulated igram with expected errors of such a mirror tested against a star (single-pass)?  

The desired conic should be -1 not 0 because you will be using the mirror in a telescope without any other correcting optics.  The null should also be enabled for the same reason.  Then in the simulation create a wave front with all 0 values.  That will create a sphere.    

[Gleb1964]

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

Dale, I would take it here.

Star test is just a Fourier transform of the exit Pupil Function.
The Pupil Function is a combination of amplitude and phase distribution across the exit pupil. That is wavefront, calculated by DFTFringe, in another words, plus some phase curvature to make wavefront converging. 

Doesn't matter, if wavefront is calculated by removing the digital null or without in the nulling test configuration. The Star test (=Fourier transform) over converging wavefront doesn't distinguish if it focused from source in infinity, or any other situations. Don't settle on it. 

Recently I have simulated the star test and Ronchi to match the refractor test in the parallel thread. No problem to use DFTFringe for that. 

Everything I done can be verified by using Zemax, so I am responsible for what I'm talking about. 

[MKV]

The desired conic should be -1 not 0 because you will be using the mirror in a telescope without any other correcting optics.  The null should also be enabled for the same reason.  Then in the simulation create a wave front with all 0 values.  That will create a sphere.    

Thank you, Dale. All 0 values result in straight fringes. That's not correct. A sphere tested on a star produces spherical aberration.  I know that I set up my simulation correctly because it checks with raytracing.

 

https://www.cloudyni...les/?p=14084127

 

____________

PS

 

Then, Gleb jumped in and said my simulation was incorrect.  I decided to do a raytrace as a sanity check, and my numbers agree with OSLO as well as with DFTFRimge.  

 

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

_

____________

PSPS

 

And here's my OSLO layout if anyone wants to try.

 

Edited by MKV, 16 April 2025 - 03:58 PM.

[Dale Eason]

MKV,  is your conic column what you entered in the mirror config or the best fit conic displayed by DFTFringe or something else?

[BGRE]

One effect that should be taken into account in raytracing an interferometer setup is the effect of the camera lens.

The camera lens ideally images the test surface onto the detector.

In that case the OPD measured along all possible ray paths from a point on the surface to its image are identical.

In particular the OPD measured along such rays are the same as that of a ray that retraces the path of the incident ray through the camera lens. The effect is that the OPD seen by the interferometer is equal to twice the OPL from the reference to the test surface measured along the incident ray.

Real camera lenses don't always accurately image a curved surface onto a flat surface the effect of this is an additional error term proportional to the object space defocus and the square of the angular error.

Raytracing can be used to include this error.

In most cases this additional error is very small.

The effect of the camera lens and the fact that OPLs should be measured along the ray and not parallel to the axis makes virtually all the results obtained on various websites garbage except for the Seidel term. All higher order terms obtained are incorrect.

One result is that the Numeric null used in DFTFringe for a paraboloid tested at its CoC is useful for larger and faster paraboloids than the naive analysis that calculates OPLs along the axis and not along the ray predicts.

[Gleb1964]

--

Then, Gleb jumped in and said my simulation was incorrect.  I decided to do a raytrace as a sanity check, and my numbers agree with OSLO as well as with DFTFRimge.  

--

 

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.

That was not correct to compare Annulus data with yours.  

Edited by Gleb1964, 16 April 2025 - 06:03 PM.

[Gleb1964]

..question on another topic how to enter mirror parameters for a simple 6-inch F/8 spherical mirror tested against a star-like source very far away ..  

Starting with Mirror configuration:

 

 

Desired conic can be 0 or a small value very close to zero, not important.

 

 

Next producing simulated interferogram with Z8 = 0 and tilt x = 7.5 (for example):

 

 

Nothing wrong yet - all strait fringes for the sphere from the radius. 

 

Calculating wavefront, it is flat with some ripples:

 

 

Looking for the simulation of the star and Ronchi:

 

 

No errors, perfect star test from the radius!

 

 

The same, perfect Ronchi test from the radius, as it should be.

 

 

Now changing desired conic in the Mirror configuration from 0.00..  to -1.0  (as it should be for paraboloid, working at infinity).

 

The calculated wavefront is changed to under-corrected by 0.153 waves SA (632.8nm). Ignoring any demand about wavefront inversion.

 

 

Nothing wrong, wavefront for infinity source should be undercorrected.

 

Next, looking for the star test for infinity object:

 

 

Star test looks like it should be, with different inside and outside defocused images.

 

One problem I found here: in the presence of SA DFTFringe is missing best focus point(!) and defocused images are not symmetrically defocused around best focus.

That is a problem, which make difficult to use the DFTFringe star test with significant amount SA!!!  

 

Now, I want to push DFTFringe to show me Ronchi/Foucault for the infinity source.

I do next steps - go to meny item "saving nulled smoothed wavefront" and saving wavefront and after loading it back. Next changing desired conic in the Mirror configuration to 0 (or small value close to 0.000..).

 

Now looking Ronchi and Foucault for infinity source on spherical mirror (fixed -no, have tested by Zemax, I should use moved source option instead, that should be +/-6mm defocus of the Ronchi screen, Ronchi 130 lpi):

 

 

Verified by Zemax.

 

Done! 

Edited by Gleb1964, 16 April 2025 - 07:18 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.

That was not correct to compare Annulus data with yours.  

 

No problem, Gleb.

[MKV]

Starting with Mirror configuration:... Done! 

Good!

 

MKV,  is your conic column what you entered in the mirror config or the best fit conic displayed by DFTFringe or something else?

I don't remember, Dale. I'll have to check. I believe I just entered 99% in the simulation. Or maybe I entered the AN.

Edited by MKV, 17 April 2025 - 04:30 AM.

[Dale Eason]

 

 

One problem I found here: in the presence of SA DFTFringe is missing best focus point(!) and defocused images are not symmetrically defocused around best focus.

That is a problem, which make difficult to use the DFTFringe star test with significant amount SA!!!  

 

 

That I don't understand.  I think the zernike values are calculcated so as to give one the lowest RMS value when defocus is 0.  But I'm no expert on that.  Whatever will best focus be when RMS is at a minimum?   If not where will best focus be?

[Dale Eason]

The normal setting for Foucault and Ronchi use the current wave front but with the artificial null turned off if it had been enabled in the mirror config.  

 

I have new code in the works that lets one enable auto collimation settings for ronchi and Foucault for the current wave front.  All it has to do is just like it does for star test is to apply the artificial null.  I don't know if the wave front error needs to be doubled but I think it does.  Does that make sense?

[Gleb1964]

There are two options to calculate defocus, from the paraxial focus or from the best focus.

The best focus is when SA is balanced with defocus to get the minimum rms.

 

It seems that DFTFringe simulating defocus counted from the paraxial focus rather than from the best focus. We need Bruce to look at the problem.

Edited by Gleb1964, 16 April 2025 - 08:10 PM.

[Dale Eason]

It uses either side of zernike defocus value of 0.  Years ago we talked about this on the group.  I think that was the solution.

[Dale Eason]

In DFTFringe when defocus is set to 0 that is the minimum value of RMS.  You can see that by using the defocus adjustment and set that value to any none zero value.  It will always create a higher RMS value.  So by what Gleb said about best focus being at lowest RMS then DFTFringe does that as the best focus point.

Edited by Dale Eason, 16 April 2025 - 07:58 PM.

[Gleb1964]

Dale, you can try next, make a wavefront with large enough SA, go to the star test and reduce defocus until you get inside or outside windows in the best focus, while you will see the central image would be not as good focused. That how I proved that star test is not taking defocus from the best focus position.

[Dale Eason]

Dale, you can try next, make a wavefront with large enough SA, go to the star test and reduce defocus until you get inside or outside windows in the best focus, while you will see the central image would be not as good focused. That how I proved that star test is not taking defocus from the best focus position.

How much SA?   The central window is always displayed with defocus set to 0 no matter if you set defocus with some other method.  So I'm puzzled why you think it changes.   It should not and I don't see it changing.