Jump to content


Photo

Olso and Ross - Dall null test

  • Please log in to reply
33 replies to this topic

#1 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 03 May 2013 - 01:03 PM

I asked if some one would post an oslo file for this and no one did so here is my take on it. I also asked about sources but have finally found how to get Oslo to take source na and object height in Focal mode.

The file is set up for a Ross test with a 125mm lens. For Dall just remove the 2nd pass though the lens and set the image distance from the mirror to it's radius.

Not much fun trying to get source - lens - mirror distances with sliders but it can be done with Genii. Guess figures based on say 1/2 the lens focal length and run Genii with a high lpm setting. 250 is fine for this mirror. The source na has a slight effect on the result. I chose to slightly over fill the mirror. :smirk: Oslo does odd things with the na over 1.

There is little difference between the 2 tests but Dall seems to have the lead on minimal residual SA and slightly better ray intercept curves. Neither of these tests are really on axis but the Dall test can be via a beam splitter. Neither are perfect - if I have the set up correct. Maybe some one can test the file some how.

:bawling: Both are useless at F3 so guess it's ofner next.

John
-
Whoops Wrong File

Attached Files



#2 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 03 May 2013 - 01:13 PM

I should add that the results usually need the longer distances changing a little to minimise the spot size and maximise the mtf of the set up.

John
-

#3 Mike I. Jones

Mike I. Jones

    Gemini

  • *****
  • Posts: 3257
  • Joined: 02 Jul 2006
  • Loc: Fort Worth TX

Posted 03 May 2013 - 02:24 PM

I tweaked your design in several places. Using "Rim Reference Rays" rather than "Entrance Pupil", and setting the 220mm mirror as a reference surface, makes the light cone go right to the edge of the mirror. I also shifted the test wavelength to 0.585µm, which is a pleasing yellow and better for testing than 0.656µm, the color of red Nyquil and harder to see.

Is the lens you used fictitious or commercial off-the-shelf (COTS)? If COTS, from who?
Mike

Attached Files



#4 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 03 May 2013 - 03:28 PM

It's and off the shelf lens Mike. Wavelength is down to Dall- low dispersion variation -red light - crown. I haven't looked at BK7 in that respect but would suspect it's still true. I have tried a red led with the Dall set up and it's ok but lens problems were very obvious. I should really have changed the wavelength to suite that.

I picked the lens from this list. They do various grades.

http://www.galvoptic...o.uk/lenses.htm

I checked that the focal lengths come out right - backwards. At F3 I tried a range from 80 to 200mm. Not good enough but 120mm was about the best spot wise. Oddly Dall suggests lenses with a focal length of around 1/2 the mirrors diameter. The graph as it comes in ATM III can't be read with sufficient accuracy really so it's good that Genii seems to be able to optimise the set up. I'd guess from the results though that this is really a technique for F5 max and not too large a mirror.

John
-

#5 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 03 May 2013 - 05:01 PM

Problems Mike. :bawling: It wont come up with sensible answers to a Dall set up and I can't see why.

Also very minor problem of not being able to solve for the diameter of the lens used.

No problem setting the wavelength for a red led with the original Ross file.

:confused: Trouble is that I know there is no problem with the Dall test on a 220mm F6 mirror.

Also it's a 125mm FL lens. Sorry about that.

John
-

#6 Mike I. Jones

Mike I. Jones

    Gemini

  • *****
  • Posts: 3257
  • Joined: 02 Jul 2006
  • Loc: Fort Worth TX

Posted 03 May 2013 - 05:46 PM

Post the file, let's see what's what.
Mike

#7 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 04 May 2013 - 03:51 AM

Looks like I should delete the file I posted Mike. The ray intercept plot and LSA plots have a scale that is 4 and 10 times coarser then than yours with the Ross set up. Opic is also giving a better optimisation now. :mad: I've had that sort of problem before and no idea why.

Anyway here is the file changed to Dall.

John
-

Attached Files



#8 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 04 May 2013 - 04:47 AM

It looks like it's an optimisation problem - not finding a true minima so there are likely to be problems plugging any mirror / lens combination into it. Just tried changing the mirror and X data to a 300mm F4.4 mirror. One optimisation may have been ok but a subsequent one was hopeless.

John
-

#9 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2622
  • Joined: 20 Jan 2011

Posted 04 May 2013 - 06:28 AM

It looks like it's an optimisation problem - not finding a true minima so there are likely to be problems plugging any mirror / lens combination into it.

There is no optimization problem, John. The Dall and Ross null tests are theoretically true null tests in monochrome light.

#10 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 04 May 2013 - 07:24 AM

There is a little more info coming up on Google now than last time I looked. I found 2 pages on using Oslo to set it up, one for V6... and another for the earlier version. Both are very similar to the file I posted but only optimise for lens to pin hole spacing and adjust the other manually if needed. This suggests that the conjugates don't matter other than getting the image of the mirror in a suitable place. :smirk: I will try that.

One version of optical shop testing also gives a formulae based purely on the position of the mirror rad centre and the image formed by the lens. It also suggests that there is no difference between Dall and Ross nulls - with a single lens other then less of a possible offset tester problem. It also suggests that it's a pure 3rd order null and similar mirror limitations to the one mentioned in ATM III.

John
-

#11 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 05 May 2013 - 04:44 AM

Mike I tried the reference surface idea but using source NA. It doesn't work out the full projected diameter of the mirror is used.

There doesn't seem to be any real difference between Dall and Ross other than field curvature is much lower on the Ross set up. Ross may leave more residual SA at times but that could be down to optimisation not finding a true minima??? It seems Ross used an aspheric plate to correct the higher order SA.

This has been gone through before on

http://www.atm-works.../dall-null.html
http://www.atm-works...dall-setup.html

Also seems a null is possible with the lens the other way round but I've had no luck.

At 220mmF4 I'm not surprised people use a Ronchi screen. This is the result, I assume as would be seen with a knife edge. Now have to wonder why the file you posted gave such good results. Any mirror changes I make give worse results than I was getting.

John
-

Attached Files



#12 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 05 May 2013 - 05:00 AM

This is the same for 220mmF6 and ties in with when I used it with a knife edge.

John
-

Attached Files



#13 DAVIDG

DAVIDG

    Aurora

  • *****
  • Posts: 4915
  • Joined: 02 Dec 2004
  • Loc: Hockessin, De

Posted 05 May 2013 - 09:38 PM

Ajohn,
Why aren't you use RossNullXp to do the setup ? It also has the option to save the results in a OSLO file. http://www.ceravolo.com/ross_null.pdf
Attached is an example OSLO file it created.

- Dave

Attached Files



#14 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 07 May 2013 - 03:36 AM

Mainly because I am mostly interested in the Dall null test Dave.

The file you posted seems to give an answer to something I cam across. A paper called aberrations in the SMALL lens Dall null test. Could be that there was another using a bigger lens the other way round.

I converted your file to the Dall test and viewed as the links suggest it seems to give better results but I then played with the Ross file and got even better results. Subsequently playing with another Dall set up after optimisation improved the results. On that one a 1/25 wave null on an 220mm F3 mirror rather than 1/20 but it needs a huge lens. The main difference seems to be more LSA on a Dall test and a slower F number during testing.

This is the result of your file and same lens used for a Dall set up. Lower one is the Dall.

One thing I am not clear on is how an Oslo wavefromt interferogram relates to shadows as would be seen with a knife edge during testing. Any one know?

John
-

Attached Files



#15 Mike I. Jones

Mike I. Jones

    Gemini

  • *****
  • Posts: 3257
  • Joined: 02 Jul 2006
  • Loc: Fort Worth TX

Posted 07 May 2013 - 07:45 AM

Just a nit: It's "OSLO", not "OLSO". If you Google "OLSO" trying to find OSLO-EDU you'll just find the "Olson twins"! :vomit:
Mike

#16 DAVIDG

DAVIDG

    Aurora

  • *****
  • Posts: 4915
  • Joined: 02 Dec 2004
  • Loc: Hockessin, De

Posted 07 May 2013 - 08:00 AM

Just a nit: It's "OSLO", not "OLSO". If you Google "OLSO" trying to find OSLO-EDU you'll just find the "Olson twins"! :vomit:
Mike


Sorry Mike, one of these days I'll learn how to type and spell !

- Dave

#17 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 07 May 2013 - 09:33 AM

Wouldn't worry about it Dave. Comments on knife edge shadows against Oslo interferograms would be much appreciated. :( Or even against Ronchi.

I just modified my call back routine to tune null tests up with a slider. With a a lens change I can get around 1/30 wave P/V on a 220mm F3 mirror. Not sure it's practical though. Trying to achieve a magnification of 1 needs distance adjustments in nm or less. Oslo EDU will sadly only allow a min rms auto focus in cmd files and not min opd but iterations and graphics updates work ok.

John
-

#18 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 10 May 2013 - 04:59 AM

RossNullXp is a bit naughty on 2 counts. It sets an image size that would be very difficult to achieve in practice and also always seems to optimise for a mirror on the way to being 10% over the requested size. The image size makes the null look a little better so probably isn't important. The increased size of the mirror can add a significant amount to the null that it gives. If a solve is added to the mirror diameter in the oslo file it produces and the beam radius corrected with a factor formed from the solved rad and the required mirror diameter radius results are mixed but only after refocusing for min opd.

I have sorted out the Oslo problems. I did have the lens to source spacing as the variable in the call back routine with the slider on the mirror to lens distance. On the Ross set up Oslo can get it's nickers in a twist around the point where the magnification of the set up is near unity. If these are reversed it doesn't happen but it still pays to reduce the step size as the source and and return image get near to each other.

I found a null lens that can give a 1/10 pv null in a Ross set up on an F3 220mm mirror. Tols on lens to mirror are 0.25mm and lens to source 0.1mm. The tol on the mirror rad is 0.1mm though. Ouch! This was with a 40mm dia 125mm FL lens.

John
-

#19 DAVIDG

DAVIDG

    Aurora

  • *****
  • Posts: 4915
  • Joined: 02 Dec 2004
  • Loc: Hockessin, De

Posted 10 May 2013 - 09:40 AM

John,
If you never used the Ross or any type of null test, what your going to soon find out is that achieving the null is much more difficult then worrying about if the test is good to 1/10 wave or less, especially with an f/3 mirror. Even with the residue errors in the test and the tight tolerances, if you can achieve a clean null the results will be mirror that is smooth and slightly over or under corrector which is good place to be. One can use other tests, I like to star test to determine what the actual correction is and then fine tune the results. I'm working on 16" f/3 parabola now made from a slumped blank that a group of us at Delmarva Star Gazers made 8 of them. I plan on using the Ross, double pass autocollimation with my 11" x 17" flat, and the star test to cross check the final figure.

- Dave

#20 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 10 May 2013 - 11:01 AM

Dave. With me at the moment it's a case of can I test the mirror with a good chance of success. I intend to try just ordinary Foucault based on some tests shown on youtube but for an F4 mirror from memory. It used a thinish slit like mask which I understand goes back to Ritchie and looks easy to take measurements. I'd guess that the tester used a carefully set slit. I think people are inclined to forget that fast mirrors were made using Foucault and then checked with Hartman (spelling?)

I've looked at Ross and Dall and find that while relatively low nulls can be obtained they don't really leave any scope for error and in my humble opinion aren't that good anyway. More recently I have looked at Ed's Dall conjugate test and find that with a pin hole at 4m from the mirror I can get nulls better than 1/30 wave P/V. That leaves scope for tolerances other than the usual "what is the actual mirror rad". Up the pin hole distance to 24m and the null is to 1/200 P/V. In practice I would wonder if the sky ever offers even 1/30 p/v viewing.

Ed's intended use is via building the scope and using it as a test stand with a led test light 100yds away but it looks like it has a lot more scope than that. Frankly I suspect this test is the only one that really justifies the purchase of a rather expensive 1/20 wave lens intended to be used via the Ross method. :) Not that I would spend that much frankly.

May as well post the 4m null. My advice would be forget Ross and if you must the same lens could be used anyway.

John
-

Attached Files



#21 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 10 May 2013 - 11:13 AM

There are a couple of other points on this test. The paraxial, rms and opd focuses are all close together. RMS and OPD are only 1um apart. The F3 mirror runs at F2.5 in the test so has a significantly better mtf peak than either of the others. The Dall test usually comes out at F5.5 or so, Ross F3.3 or so. :grin: I assume that is an advantage. I'm sure all focuses coming close together is.

John
-

#22 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2622
  • Joined: 20 Jan 2011

Posted 10 May 2013 - 01:08 PM

John, you're absolutely right -- the RossNullXP OSLO file is somewhat of an enigma. If you try to enter it by hand the results don't match, and the only parameter that cannot be manually changed is -- the light source (pinhole) size which, as you noted, is ridiculously small. I would be interesting to find out how the XP alters that number.

In reality, when testing optics, the light source image can be reduced to that of the diffraction limit and no amount of reduction beyond that will make any difference. The test will also be most sensitive at or about the diffraction limit. Anyone who has done the Ritchey Common test for flats will know that. Attaining such a small light source that's bright enough can be achieved by using a well polished steel ball bearing or a microscope objective, using a monochromatic light source (superbright LED plus a narrow-band filter), and the simple lens maker's equation.

However, DAVIDG is right in that getting the error down to 1/10 wave is the least of your priorities when performing a Ross null test. I would personally never trust the Ross results as anything more than a ballpark figure that tells you if you're going in the right direction.

If I have a choice I will not do the Ross test at all, but if you have a large mirror and probably no matching flat for autocllimation purposes, then the Ross is something to consider.

You're also spot on when it comes to Ed Jones' conjugate null test. There was a thread about it not so long ago. Unfotunately not too many have seen the great benefit it offers as an alternative. Of course, the biggest problem are the air currents.

So, for big mirrors, there is still the Foucault test as the only test that can offer qualitative and quantitative results -- but even here we have to thread lightly. The quality of the testing equipment itself, the experience of the tester performing the test, and many other factors play a role in just how reliable the test results will be. It pays to invest in a good professional micrometric stage, for sure, but then you still have to deal with shadows and subjective judgment calls.

besides, the Foucault test is just not best suited for mirrors in the f/3 range. Slitless testers used by many are not very sensitive and this can be demonostrated by its inability to differentiate a spherical mirror from one that departs from true sphericity by a small amount.

That's why, to those who plan on making big mirrors, I say: invest in (or just make) an autoclllimation flat that's bigger than anything you'll ever make -- it will pay itself off many times over, and when you've fulfilled your telescope bucket list you can sell it, recoup the money, or even make a profit. There's always going to be someone who's looking for a suitable optical flat.

#23 DAVIDG

DAVIDG

    Aurora

  • *****
  • Posts: 4915
  • Joined: 02 Dec 2004
  • Loc: Hockessin, De

Posted 10 May 2013 - 01:37 PM

Since this is a fast mirror the focal length is going to be short as well so one could also use oil flat with the telescope point down at the flat. I would use the Ross to get you close as you can then use the oil flat to hone in on the final correction.

- Dave

#24 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2622
  • Joined: 20 Jan 2011

Posted 10 May 2013 - 04:37 PM

Since this is a fast mirror the focal length is going to be short as well so one could also use oil flat with the telescope point down at the flat. I would use the Ross to get you close as you can then use the oil flat to hone in on the final correction.

That is an excellent alternative to the conjugate null test, Dave, and it's as simple as a star test -- no measurements necessary. A 16-inch f/3 tested this way should not take up more than 50 inches of vertical space. BTW, you can also test popular f/10 Celestron and Meade catadioptrics that way.

#25 Ajohn

Ajohn

    Mariner 2

  • *****
  • Posts: 283
  • Joined: 03 Dec 2007

Posted 10 May 2013 - 06:01 PM

XP produces the same order of default source sizes as Oslo does when using a beam diameter rather than an object na or source size.

Air current problems with the source 4m away from the mirror indoors? I don't think so.

Star tests on a real star - In my book the only one that really can show the true state of a mirror is a knife edge. It can be done at times. One of Ed's video shows just how sensitive a knife edge is when he is showing a Ronchi test of a sphere - he uses one of the lines as a knife edge but doesn't mention that. :grin: This might even tempt me to buy a Ronchi screen. I don't think people realise just how delicately a knife edge has to be entered into the beam and that is the biggest problem with them. All down to testers using say a 1/4 unf bolt for adjustment. In real terms all that needs is a rather large say 3in dia head. Similar thoughts on say xM0.5 pitch threads would help with levelling water or oil etc tests too.

Ed refers to the test as the Dall conjugate test. Curiously I came across a paper entitled "aberrations in the small lens Dall null test". I suspect this is the large lens one. As usual as the paper is of some use it isn't freely available.

I have 3 coated flats. 6in, 9in and 19in but have no idea how flat they are. The 9in one is currently awol. Not sure where it is. They do not need to be optically flat really only super smooth and a figure of revolution. The limits for that can be ray traced.

John
-






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics