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

Who wants to make a Ross Null Lens?

  • Please log in to reply
174 replies to this topic

#51 Mark Harry

Mark Harry

    Cosmos

  • *****
  • Posts: 8,798
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 18 August 2020 - 06:42 AM

"But in practice, a lens size that
produces a residual wavefront error of 1/10th wave or better will work very well. This set
up will produce a 1/20th wave surface if the mirror is figured perfectly"
*******
Absolutely not. Common sense. This is the myth. If you want an optic to have a 1/10th wave error when finished, the test apparatus capability -HAS- to be significantly better.
And the other fly in the ointment, the SPACING between optic and reference has to be set accurately, especially with fast, larger mirrors---not by using a tape rule, either.


  • MKV likes this

#52 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 18 August 2020 - 07:48 AM

And the other fly in the ointment, the SPACING between optic and reference has to be set accurately, especially with fast, larger mirrors---not by using a tape rule, either.

waytogo.gifbow.gif  

 

____

 

PS that also goes for alignment...


Edited by MKV, 18 August 2020 - 07:49 AM.


#53 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 24,885
  • Joined: 22 Feb 2010

Posted 18 August 2020 - 07:58 AM

"But in practice, a lens size that
produces a residual wavefront error of 1/10th wave or better will work very well. This set
up will produce a 1/20th wave surface if the mirror is figured perfectly"
*******
Absolutely not. Common sense. This is the myth. If you want an optic to have a 1/10th wave error when finished, the test apparatus capability -HAS- to be significantly better.
And the other fly in the ointment, the SPACING between optic and reference has to be set accurately, especially with fast, larger mirrors---not by using a tape rule, either.

There’s tolerances for lens spacing 

 

the faster are obviously tighter  

 

Not a bad star test for a f3.5 

https://youtu.be/qdRqmnMR1Sg



#54 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 18 August 2020 - 10:19 AM

There’s tolerances for lens spacing 

the faster are obviously tighter  

No argument there in principle. However, for long RoCs there is always some sagging using a tape measure. A tape measure sually can not reach both elements very accurately. It's often eyeballed.

 

Sagging and eyeballing can take up a few mm, plus or minus. The arc is longer than the segment. Another issue is that tape measures are not certified. Their interna uncertainty error ie never specified, we just assume that they are absolutely accurate, which f course is a fallacy. Tape measures are not precision measuring devices, at leats not when it comes to optics. 

 

Third, the distance has to be measured from one surface to anther surface. Sometimes that's not easy. If you assume the distance is inside the lens you have to take into account the path length through the glass, vs. air.

 

The optical path length (OPL) through the medium of refractive index compared to the geometric path length s  will be OPL = n time s. hence if s =10, and n = 1.5, then OPL = 10*1.5 = 15. Problems with this arise if the glass area used is not sufficiently homogenous (i.e. of equal density) different parts of the wavefront will exhibit different path lengths leading to wavefront distortion and error at the focus.

 

Typically, only the most expensive lenses have high degree of homogeneity. Commercial-grade lenses are highly inhomogenous even in the central portion of the lens (the central portion is often cited by ATMs as being of higher optical quality than the outer portions...which of course is not necessarily true without interferometrically testing the lens and looking for any significant phase shift changes in various of the working diameter, and why such lenses are not recommended for testing puproses.

 

Here is a Schott company's information on substrate homogeneity with FAQ answers.

 

Another potential source of error is axial misalignment. Professional labs use a device called an autocollimator. This device allows alignment of optical axes to a few seconds of arc. This can not be achieved by hand or plane eyesight.

 

None of these and other sources of errors (such as incorrect wavelgnth, or bandwidth) is ever accounted for in ATM Ross null testing. The argument is that it's not needed because its not detectable. That's because, contrary to popular belief, ATM testing methods are usually not all that accurate.


Edited by MKV, 18 August 2020 - 10:20 AM.

  • Oregon-raybender likes this

#55 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 24,885
  • Joined: 22 Feb 2010

Posted 18 August 2020 - 10:33 AM

 

Third, the distance has to be measured from one surface to anther surface. Sometimes that's not easy.

make the lens holder have a surface that can be measured too.  you touch the center of the mirror with the tape.

 

the tape can be supported over its distance, not a problem.



#56 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 18 August 2020 - 11:37 AM

you touch the center of the mirror with the tape.

Tape measures (usually 1" wide) have square ends and cannot touch the center. The gap can be large for deeper curves. Has to be accounted for.

 

tape measure uncertainty.jpg

 

All these little exceptions can and do add up. The lens should be touched at the center as well. Autostigmatic microsope can measure the distance to a micron level, on-axis. 


Edited by MKV, 18 August 2020 - 11:41 AM.


#57 Mark Harry

Mark Harry

    Cosmos

  • *****
  • Posts: 8,798
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 21 August 2020 - 03:47 AM

No issues at Pioneer for testing GLS's lens. Some are on vacay, so it isn't quite so crowded/busy, and I'll have plenty of opportunity to do the test.


  • LarsMalmgren and MKV like this

#58 duck

duck

    Sputnik

  • -----
  • Posts: 43
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 21 August 2020 - 10:41 AM

I feel like I need to defend Peter Ceravolo after all the trash talking against garage made optics.  If you test a mirror with ROC of 8 feet, and reduce the fringes in an IF to 3 or 4, theory and practice diverge significantly.  Your going to need hundreds of igrams to reduce vibration and seeing in order to plunk your fringes into a computer program so that "random" errors don't swamp the true wavefront errors.  Then you have the biases, which aren't reduced by averaging in a single setup.  For an ATM, this is a waste of time.  If you want to produce certified 1/10 wave optics commercially, a significant fraction of the price will be to recover the costs of the testing, unless your producing small spherical optics which will fit on a rail on top of a granite surface block and can be analyzed with one shot in 3 seconds.  For me, watching the video output of the IF for a few minutes allows the brain to average out vibration and seeing.

 

I certainly believe I can make a 6" dia ross null lens to 1/10 wave accuracy using the same techniques i used to make a couple cass secondaries.

The BK7 element of a 6" Fraunhofer objective looks like a ROSS null candidate for a 24" f/4 paraboloid.  A question for OSLO proficient.  Can the null be improved by aspherizing one surface of the lens?  The Lerch/George provided s/w for Ross null setup requires spherical surfaces on the lens.

 

Also, since I'm somewhat of a Luddite, how does one embed an image in a post. 


  • Pinbout likes this

#59 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 21 August 2020 - 12:19 PM

May I suggest that you refrain from characterizing other people's opinions as "trashing"? No one trashed Peter Ceravolo, either personally or his work as garage-made.

 

As for your objection to IF testing. Maybe you're not familiar with common path interferometers (Point-Diffraction, the Bath, etc.) which are pretty much immune from vibration issues. Your biggest problem will be air current which can be minimized by running a fan. More importantly, your igrams will tell you if there is any vibration: you will observe soemthing known as "print trough" or some firnges will be dimmer or fuzzy. If you observe these with your common-path IFs, your setup is at fault or you have heavy traffic next to your faiclity, etc. Try testing late at night, when there's less traffic.

 

You don't need hundreds of igrams to get an idea where you are correction-wise. You take 4-5 for each rotation of the four to five rotations, a total of 20 to 25 igrams in all, not hundreds of them. It takes 10-15 minutes max. If you're testing a finished lens/mirror you may take a few extra igrams, but it's usually not required. The rotations of the test optic (0, 45, 90, 180, and 275 degrees) also take care of the " biases" (such as test stand astigmatism, etc.) concurrently there's no need to acocunt for these with separate tests. 

 

Professional labs test optics in a manner of minutes and this includes a setup to submicron and arc second levels (something ATMs can only dream of!), so the time is not a significant factor in the price unless you have hundreds of items to test. There are enough videos to prove that.

 

This one is less than 3 minutes long, 

 

https://www.youtube....h?v=5r79RAoZsCM

 

and this one is just under 2 min.

 

https://www.youtube....h?v=Wq7BgrdtM0c

 

If you believe you can make a 6-inch Ross null lens with surface accuracy of 0.1 wave on both surfaces, please do so and prove it. More power to you; no one will stop you. Make sure, however, that all other tolerances are also met -- such as inhomogeneity of the glass, radius accuracy of 0.01% or angular accurace to a few ar secs. Just to illustrate what this means, fort a 300 mm radius of curvature, the permissible error comes to ±0.03 mm. I presume that you have a suitable autostigmatic microscope to measure radii of curvature to that accuracy. 

 

For someone who speaks with authority, I am surprised you don't seem to have optical analysis proficiency, or understanding how conjugate tests work. A congjugate null test will null an aspheric wavefront by creating an aspheric wavefront of opposite sign and equal in magnitude. For that you need spherical surfaces, for obvious reasons.  

 

The key to a conjugate Ross null test is a proper separation of the light source to the lens and lens to mirror, within established tolerance and a measuring method used. The Lerch/George calculator (RossNullXP) provides the setup for the minimum rms OPD solution and one that also ensures equal path length (phase) to which, by the way, the Ronchi test is completely blind.

 

The information how to embed images is right at your finger tips. Go to the lower right corner of your screen and you shall find it. 

 

Cheers!


  • Oregon-raybender likes this

#60 duck

duck

    Sputnik

  • -----
  • Posts: 43
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 21 August 2020 - 04:46 PM

I gather that the null cannot be improved by aspherizing one the the lens surfaces.



#61 Pinbout

Pinbout

    ISS

  • *****
  • Posts: 24,885
  • Joined: 22 Feb 2010

Posted 21 August 2020 - 06:28 PM

I gather that the null cannot be improved by aspherizing one the the lens surfaces.

for a 24in f4 the null is fine enough, even with a ronchi, fine tune over/under with the ke.



#62 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 21 August 2020 - 07:22 PM

If you believe you can make a 6-inch Ross null lens with surface accuracy of 0.1 wave on both surfaces, please do so and prove it. More power to you; no one will stop you. Make sure, however, that all other tolerances are also met -- such as inhomogeneity of the glass, radius accuracy of 0.01% or angular accurace to a few ar secs. Just to illustrate what this means, fort a 300 mm radius of curvature, the permissible error comes to ±0.03 mm. I presume that you have a suitable autostigmatic microscope to measure radii of curvature to that accuracy.

Mladen,

how about to calculate tolerances for Ross null? You would be surprised, how loose they are. Especially in case of lens with long focal distance. The most critical is angular alignment of lens, next to it is decentering of lens. Lens radiuses and thickness are quite free. Spacing tolerances between elements also.
Here is example about Macleod's Ross null (from parallel thread). Residual wavefront rms of ideal null is 0.024 waves (I was calculation on 550nm wavelength). Here is a table of tolerances, which double wavefront rms to 0.048 waves:
                  
 
                        units             Value   Criterion   Change
lens, flat surface   fringes of power     ±31.87     0.048     0.024
lens, convex radius        mm             ±0.422     0.048     0.024
mirror radius              mm             ±0.555     0.048     0.024
lens thickness             mm             ±1.000     0.043     0.019
distance lens-mirror       mm             ±0.509     0.048     0.024
decentering lens (X,Y)     mm             ±0.295     0.048     0.024
tilt lens (X-,Y-axis)      deg            ±0.036     0.048     0.024


Macleod Ross null tolerances.png
  • Oregon-raybender and MKV like this

#63 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 21 August 2020 - 07:33 PM

I gather that the null cannot be improved by aspherizing one the the lens surfaces.

Right,

the residual wavefront after nulling by lens is not absolutely flat, but does represent combination of balanced low and high order spherical aberrations. That is for ideal parabolic mirror. See picture of example of residual wavefront I have posted above.
Now, is it possible to polish that curve on lens to remove it? Not practical. Depending of parameters of nulled mirror, that residual curve changing amplitude and size.

Gleb

#64 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 22 August 2020 - 01:59 AM

Mladen, fringes of power is suitable for flat surface, because with infinity radius it's tolerances tends to infinity. It is a standard option to chose.

If test configuration don't use all lens aperture, means beam footprint is less than diameter, that is a release factor. Next factor is a residual wavefront quality of null, the best theoretical quality that can be reached. You can see, it is quite poor in this example, thus it is difficult to detect small wavelength errors on significant residual background. Tolerance analysis is taking that in account.

Gleb



#65 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 22 August 2020 - 11:12 AM

Mladen,

how about to calculate tolerances for Ross null? You would be surprised, how loose they are. Especially in case of lens with long focal distance. The most critical is angular alignment of lens, next to it is decentering of lens. Lens radiuses and thickness are quite free. Spacing tolerances between elements also.
Here is example about Macleod's Ross null (from parallel thread). Residual wavefront rms of ideal null is 0.024 waves (I was calculation on 550nm wavelength). Here is a table of tolerances, which double wavefront rms to 0.048 waves:

Hi Gleb,

 

That was an excellent idea! Please disregard my reply last night (which I removed so as to avoid any confusion). It was late and I should have waited until the morning   before replying; lesson learned -- again :o(

 

As far as power tolerance for flat surfaces expressed in fringes, that's standard as you say, and no different than for AC flats.  A surface that has effectively insignificant refraction is specified in fringes of power. What surprised me was the tolerance for the convex surface of the lens.

 

As is the case with the Offner nulling compensator, the Ross null test is in effect a double-pass test for the lens, doubling its error at each surface. For this reason, the tolerances for such lenses are supposed to be tight, both optically and mechanically.

 

Your results provide interesting answers but also raise more questions. What's surprising is that the results suggest this isn't so, and that only the decenter and angular alignment (tip/tilt) are critical. The lens, according to the results you posted, doesn't even have to be commercial grade, worse than catalog lenses!* So, when Ceravolo insists on using a precision lens (which requires 0.01% on the radius of curvature*, etc.) it flies in the face of your results. He even went as far as to manufacture such lenses for a hefty price for the Ross null test. In his write up on the Ross null test, not surprisingly, he never mentions any lenses of lesser quality as adequate for the test. Nothing wrong with that if one takes into consideration that the lens error contribution is effectively doubled because the lens is passed twice, thus that its total error contribution to the wavefront must be smaller, significantly smaller, than that of the test optic.

 

Of course, after reading this I had to try SYNOPSYS, and I obtained similar results as yours. But, I wish to double check because it was late and it's a good thing to repeat when fiddling with numbers in the state of "prefainting euphoria" :o).

 

And, by the way, our friend Mcleod is interested only in getting the mirror down to 1/4 wave, not 1/20 wave! So, he can probably get a plastic condenser lens in a school stationary store and he should be good to go, lol!

 

Anyway, very interesting, to say the least, and thanks again!

 

Mladen

 

PS

By the way your results list the wavelength at 633 nm, not 550 nm.

 

__________________

 

*Industrial standards for precision lenses on radii of curvature are very clear: 0.01% of the radius.

 

radii_errors_tolerances.jpg


  • Oregon-raybender likes this

#66 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 22 August 2020 - 11:22 AM

Mladen, fringes of power is suitable for flat surface, because with infinity radius it's tolerances tends to infinity. It is a standard option to chose.

If test configuration don't use all lens aperture, means beam footprint is less than diameter, that is a release factor. Next factor is a residual wavefront quality of null, the best theoretical quality that can be reached. You can see, it is quite poor in this example, thus it is difficult to detect small wavelength errors on significant residual background. Tolerance analysis is taking that in account.

Gleb, when dealing with the Ronchi test, the most common test method used, it's pointless to speak of rms OPD. The difference between the PV focus and the RMS focus is often a few microns. No Ronchi test will show the difference. But a knife-edge will. It's the PV residual error the Ronchi method is showing, and if it's anything smaller than 1/4, possibly 1/5 wave PV, it's usually the extent to which you can see any difference (bowing) of Ronchi bands. 



#67 duck

duck

    Sputnik

  • -----
  • Posts: 43
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 22 August 2020 - 02:57 PM

Thanks MKV and Gleb.  The tolerance analysis does show one thing that is pretty difficult.  The ROC of the lens and mirror need to be known to within 1/2 mm.  Maybe not too hard for the lens because if homemade will need a matching concave test plate anyway.  But for a primary with ROC = 192", more difficult.  Need to make an extensible rod.  In the past have clamped 1x2" wood strips together.

 

Such fun!



#68 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 22 August 2020 - 06:00 PM

But for a primary with ROC = 192", more difficult.  Need to make an extensible rod.  In the past have clamped 1x2" wood strips together.

Yup, a radius rod. You've got it! :o)

 

ross_bar_LR (2).jpg

 

With two 10-32 nylon pan head screws on each end easily adjustable 0.1 mm (1/8 turn of the screw).. 

 

Getting axial alignment and tip/tilt to tolerances is a bit more challenging. A good autocollimator helps.

 

AC_1_lr_3.jpg



#69 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 22 August 2020 - 06:07 PM

The ROC of the lens and mirror need to be known to within 1/2 mm.  Maybe not too hard for the lens because if homemade will need a matching concave test plate anyway.  But for a primary with ROC = 192", more difficult.  Need to make an extensible rod.

Actually not, if correct method to set lens position is used. If you set lens by mirror-to-lens distance, that way is incorporate mirror's radius error into lens position error. If instead you set lens using distance from center of curvature, error of mirror radius is not included, it is mostly self compensated. I haven't used that compensation in tolerance analysis, otherwise that would significantly release mirror radius tolerance, may be by factor x6..x8.

  

 

By the way your results list the wavelength at 633 nm, not 550 nm.

 

Mladen, wavelength 633nm assume only fringes of power, like lens surface control during production with HeNe laser. All other calculations, opd, rms is used 550nm.



#70 duck

duck

    Sputnik

  • -----
  • Posts: 43
  • Joined: 11 Jun 2020
  • Loc: madera ca

Posted 22 August 2020 - 08:11 PM

Didn't think of that, Gleb.  MKV - I scratched my $350 (circa 2000) Ceravolo lens by not putting the nylon screws into the wood.  BK7 is soft?

 

Tip/tilt and centering the lens is a bugaboo in my setup.  Next improvement necessary is a rail and optomechanical mount for the lens.

 

An improvement in my work environment would be helpful.  I once reduced the tilt to 1 fringe (aforementioned 8 feet ROC).  It was a light show.  Vibration +  seeing about 1/2 wave, I suspect.  For me to get a surface map of the mirror that reduced seeing and vibration effects to 1/20 wave would take 100 igrams.



#71 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 22 August 2020 - 08:57 PM

Didn't think of that, Gleb.  MKV - I scratched my $350 (circa 2000) Ceravolo lens by not putting the nylon screws into the wood.  BK7 is soft?

That's a shame. Today it costs more than twice as much. The wood may have had a shard of metal from a saw or a file. You never know.

 

Nylon is hard and doesn't compress much, especially if it's tightened by hand. I place a piece of celaphane between the nylon screw and one of the optical sufraces and tighten the screws until I can just pull out the celophane without tugging. The celophane is 0.05 mm (0.005" inches) thick, so the error is insignificant.


Edited by MKV, 22 August 2020 - 08:58 PM.


#72 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 23 August 2020 - 07:32 AM

Gleb,

 

I ran a ±0.025 wave tolerance in Synopsys for Mcleod's 300 mm f/3 Ross null test. I changed the layout of test parameters a little to reflect more the reality of an actual setup by restricting all decimals to two significant figures (in mm). The wavlength used for the trace and analysis was 632.8 nm. The layout and the summary of the results are shown below.

 

 

As you can see the only significant difference between your data and mine appears to be the number of fringes of error for the flat lens surface (abut 9-fold higher in your case) and the lens central thickness (8-fold). The lens convex surface tol. (yours/mine) is 0.422/0.361, mirror is 0.555/0.592, lens to mirror dist. 0.509/0.116, decenter X,Y 0.295/0.117, and tip/tilt 0.036/0.049 deg (or 2.16/2.92 arc min). 

 

Here's a summary for all three optical surfaces, plus a complete layout along with tolerances. The results are in mm unless otherwise stated.

 

mcleod 330 F4_5 TOL ross lens.png

 

mcleod 330 F4_5 TOL mirror.png

 

mcleod 330 F4_5 TOL ota.png

 

My results suggest tighter tolerances, but I don't think I would call your results "wide open" for tolerances -- except for the flat side of the lens. I don;t think they're easily or commonly met by ATMs. I didn't take into account inhomogeneity, coatings, and surface irregularities, which could also affect the results. 

 

The best way to assess how good or bad a specific Ross null test is to use an interfeormeter to assess the mirror by itself and in combination with the available Ross lens and a specific setup. Technically, the reuslts of both tests should be about the same as far as the mirror quality is concerned. In other words, if an IF test shows the mirror Strehl is 0.98, the Ross null test should provide the same numbers. Any difference is due to the RN test (setup, lens quality, or both)

 

Mladen

 

 

 

 

 

 

Attached Thumbnails

  • mcleod 330 F3 TOL data.png


#73 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 23 August 2020 - 09:16 AM

Mladen,

 

First, I want to be sure we exploring the same optical layout. But I see your PCX lens diameter is 66mm (should be 100), and mirror is 305mm (should be 330).

Second - light does propagate forward and back trough the same system. I mean that tolerances randomly change system on a way forward and when light does propagate back, it is going trough exactly the same changes, not a randomly changed once more on way back. In Zemax layout I have done, all element's parameters on back propagation is connected by "pick up" option with correspondent parameter of propagation forward.

Third - focus compensation is used. That means, with any change of system parameters wavefront analysis is done from the best focus with minimum rms. Knife would be placed exactly in best focus getting nulling shadows.

Fourth -  "a ±0.025 wave tolerance" means change of rms, yes? Like residual rms 0.025 ±0.025.



#74 MKV

MKV

    Cosmos

  • *****
  • Posts: 9,065
  • Joined: 20 Jan 2011
  • Loc: Southeast, USA

Posted 23 August 2020 - 04:54 PM

Mladen,

 

First, I want to be sure we exploring the same optical layout. But I see your PCX lens diameter is 66mm (should be 100), and mirror is 305mm (should be 330).

Second - light does propagate forward and back trough the same system. I mean that tolerances randomly change system on a way forward and when light does propagate back, it is going trough exactly the same changes, not a randomly changed once more on way back. In Zemax layout I have done, all element's parameters on back propagation is connected by "pick up" option with correspondent parameter of propagation forward.

Third - focus compensation is used. That means, with any change of system parameters wavefront analysis is done from the best focus with minimum rms. Knife would be placed exactly in best focus getting nulling shadows.

Fourth -  "a ±0.025 wave tolerance" means change of rms, yes? Like residual rms 0.025 ±0.025.

Gleb,

It is the same optical layout as yours -- the data pertain to Mcleod's 330 mm f/3 mirror and a Ross null lens of 100 mm diameter, 10 mm central thickness, and convex radius of curvature oif 206.7 mm, BK7/N/BK7 no matter what a label or the file name may say. The reason my lens size is 66 mm is because that's the working diameter of the lens. Same with the mirror. I should have used a slightly large Y1 value on surface #1 for a given object distance. I'm sure this is pretty irrelevant in the overall assessment of tolerances. Are you suggesting the lens errors cancel on the way back? Then why does literature for Ross and Offner, etc. tests say the lens should be precision optic?

 

My tolerancing command was simple

 

mcleod 330 F3 TOL setup.png

 

where TOL WAVE 0.025 simply means "maximum permissible value of the wavefront variance" which to me means ±0.025. I suppose I could have used DEGRADE WAVE 0.1 which to me means "fractional change permitted in the wavefront variance," waveronft opd±a fraction thereof (i.e. 0.1 for 10% or 0.01 for 1%, etc.)

 

One can also use BTOL 2 or BTOL 3 where numbers = standard deviation (sigma) value.

 

Mladen



#75 Gleb1964

Gleb1964

    Vostok 1

  • -----
  • Posts: 136
  • Joined: 13 Jul 2016
  • Loc: Sweden

Posted 23 August 2020 - 05:14 PM

Not, I am not suggesting that errors canceled on the way back. What I am talking about, that light travel forward and back the same elements (lens), but formal prescription is done like light passing lens1, mirror and another lens2. If tolerance algoritm randomly changing some parameters of lens1, it should do the same change with lens2, because in reality it is the same lens lens1=lens2. But if algorithm has no such instructions, it may change lens1 and lens2 differently and that would affect result of tolerancing.

Regarding diameter of lens - number of fringes (or sag) is in square depending of diameter. 


Edited by Gleb1964, 23 August 2020 - 05:17 PM.



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