22 replies to this topic

[duck]

I believe I've found bad guidance in Texereau.  He writes that frequent flipping of positions (tot/mot) will better match ROC's of the test plate and optic.  What I've found is that there is more benefit to completely working in one orientation at each grit.  When you flip frequently, you change where the action is occurring to fast.  Frequent flipping may work OK for flats.

 

Would like to hear opinions...  

[PrestonE]

Hola Duck,

 

When we are making a Test Plate to fit an Optic, we use specific tooling for each piece.

 

Example, the Concave Test Plate has an Convex Tool and the Convex Optic has a Concave Tool.

 

We work these TOT and MOT back and forth once the desired ROC is achieved.

 

We do Not Try to use the Test Plate to Produce the Optic directly...as they will never fit

properly in our experience.

 

If we are not alternating TOT and MOT back and forth the result is continually Increasing or Decreasing ROC

depending on what one is doing.

 

We find that with the Convex on Top, that works much Faster than when the Concave is on Top, ie the Convex on Top

will Flatten the surface Much Faster than the Concave on top will Deepen the Surface.

 

We can hit the ROC within a 0.00005" with both the Convex and Concave surfaces, such that when given a 

quick polish they are within a couple of waves correction and thus are finished up fairly quickly.

 

Does that answer your question?

 

Regards,

 

Preston

[hamishbarker]

For the two convex mirrors I have made, I found that the mirror roc would increase (go flat) much faster on top than it would decrease when on the bottom. It was always a battle to keep the mirror roc short enough. Maybe bad actions on my part, I don't know.

I did better when i made a separate concave tool and pushed it to a bit shorter roc. Then I was able to get such a good match that at one point I had perfectly straight fringes. Thankfully Dale's dftfringe software can analyse curved fringes also, so perfect roc matching isn't mandatory.

[duck]

No.  My current attempt with cast iron laps cannot come close to an ROC mismatch of 1 micron.  (Have you confused delta_ROC with delta-sag?) I get 5 microns delta-sag (calculated from the measured ROCs of each surface), which is 10 waves of light, if they were able to be tested in contact IF.  This is with 60 micron alumina.  I presume as I get to finer grades, the delta-sag will tend toward zero.

 

What my post was about was that frequent flipping of MOT/TOT does not result in better conformance of the surfaces.  I got the best results (from spherometer measurements) when leaving the orientation fixed.  And YES - I am getting ZERO ROC change with concave on top with these cast iron laps.

 

Now ready to verify the ROC optically.   

[Oregon-raybender]

The mismatch is equal to the thickness of the grit as it grinds down.

1 micron is close enough, even 5 is fine. The ROC will change again

as you start to polish. I ground the part towards the flat side of the curve.

That way I know it will hit the edge first. Now to test you radius to

the testplate is simple. Polish the part with wax, place a small amount

of wax on the surface and buff with a cloth until it shines. Then check the

number of fringes using the testplate. Also, I don't change the process

by flipping. I correct the tool using a fine stone to get the radius I want.

For a convex tool, to shorten the radius I stone the edge, for flatting

the curve by stoning the center. This is nice thing about using cast iron tooling.

 

Getting within 10 fringes or less is close enough, I can change

it during polishing.

 

Starry Nights

[starspangled]

I made a lot of test plate pairs  years ago when I worked at a government research facility optics lab.  

 

We had precision lapped matching template guages ( perhaps 3mm thick ? )  made for the curve by CNC which were used during grinding . The concave mirror was tested and fine tuned for ROC via flash polishing and tested for precise ROC on an optical bench with pinhole source and eyepiece with a precision graduated scale , and polish finished at the design ROC , but not fine  figured . 

 

A convex plate was then manufactured and adjusted for ROC by flash polishing and interference test .Once the ROC was good both surfaces were figured on their own polishers and tested against each other interference for zones , by moving one surface sideways against the other and noting any bends , until the bands were dead straight and no curve . Some of the largest plate diameters were 200mm and ROC as long as 10 metres .

Edited by starspangled, 27 March 2025 - 06:14 PM.

[ThomasM]

 

We can hit the ROC within a 0.00005" with both the Convex and Concave surfaces, such that when given a 

quick polish they are within a couple of waves correction and thus are finished up fairly quickly.

 

Preston, this is really amazing, so you can hit the ROC within 13 micron? You can distinguish between 200 mm and 200.0013 mm? How do you measure this?

 

Thomas

[duck]

as usual for me, this is a tedious process.  To optically measure the ROC of the concave surface (cast iron, 60micron alumina grind), I polished in some black shoe polish.  Then I wetted the surface with water.  I put the lap on the Foucault test stand and opened the source wide.  I never found the return image.  The best I could do was focus on a lamp, with some provision for shielding to see the image.  Very rough and not really worth the effort.

 

The glass will arrive soon.

 

I think the RayBender has said how to effect ROC changes on the cast iron laps....put them on the table and stone them.  Suppose I want a shorter ROC...I would put the convex lap on the table (rotating very slowly) and stone the outer zones until I got the sag I need.  Then lap the two (concave on top) together to get conformance.

Edited by duck, 28 March 2025 - 12:26 PM.

[PrestonE]

Preston, this is really amazing, so you can hit the ROC within 13 micron? You can distinguish between 200 mm and 200.0013 mm? How do you measure this?

 

Thomas

Hola Thomas, The below link gives the specs to our instrument.

 

https://shop.mitutoy...-10/index.xhtml

 

Regards,

 

Preston

[Oregon-raybender]

You can check the ROC of the tool using what you mention above.

Focus the lamp is good. One thing you can try is using two dots,

or two LEDs close to each other. Adjust the image plate until they

cross over. I used measuring bars or stick with the length of the ROC.

The only other best thing is microscope with a projected source,

on a rail. Focus in the center on the surface and move back towards

the focus point. Not sure how close you will get, it depends on many

factors of all the parts.

 

Preston, did you use the digital gage for the radius sag, ?

In setting the gage, I used gage blocks on the feet to correct

height for CC, and for the CX, gage blocks for the center point.

I would measure each block set with a digital mic to be sure

of the values.

 

This maybe of help ?

 

Starry Nights

 

https://wp.optics.ar...nses-I.pptx.pdf

[PrestonE]

"Preston, did you use the digital gage for the radius sag, ?

In setting the gage, I used gage blocks on the feet to correct

height for CC, and for the CX, gage blocks for the center point.

I would measure each block set with a digital mic to be sure

of the values."

 

We have a Full Metrology lab, and Cross check these gages with K Grade Calibration

blocks and our Tesa-MicronHITE 0.00001" scale Electronic Height Gage.

 

One can Breath on it and see a difference instantly...

 

This is kept in a room at 25deg C +/- 0.2 for most of the year.

 

As well we have other Direct Measuring Gages that can measure 2 millionths with +/-

1 as well.

 

So yes, we understand fine measurements.

 

We also have built an ASM with a Digital Scale that measures to 0.00005 inches, though

being able to determine the Sweet Spot down to that level has not been possible for us.

 

We generally can see a difference with a really good spherical surface of about +/- 0.001 inches 

on an ROC of >24" F4.

 

Regards,

 

Preston

[Oregon-raybender]

OK, understood. I had the same type of lab set-up.

I spent money buying right equipment.

ASM was handy to have. 

 

It was fun demoing the sensitive of the equipment engineers, 

hand warming or warm breath, watch the values grow.

The best was using the ZYGO, showing the effects of heat

waves from the hand in the optical path when you tilt

the fringes.

 

I forgot to mention, to other folks. You ZERO the gage

when using gage blocks. Doing so allows you measure

a tight tolerance with out using the length (travel) of the indicator.

(which can have slight errors)

 

Starry Nights

[ThomasM]

Hola Thomas, The below link gives the specs to our instrument.

 

https://shop.mitutoy...-10/index.xhtml

 

Regards,

 

Preston

Preston,

thank you. so the acuracy of the digital indicator is specifed with 0.002 mm, this is amazing.But this doesn't translate 1:1 into the precision of the radius of curvature. An example, if we take a lens or mirror with 200 mm radius of curvature, with 100 mm diameter, this is a very fast system, the sag is ~ 5.9 mm. Now if we can measure the sag with 0..002 mm precision the ROC will be 200 mm +/- 0.035 mm. This is still very impressive. If the system is slower the error is larger.  I hope that I didn't make a mistake.

 

Thomas
 

[MKV]

Now if we can measure the sag with 0..002 mm precision the ROC will be 200 mm +/- 0.035 mm. This is still very impressive. If the system is slower the error is larger.  I hope that I didn't make a mistake.

An autostigmatic microscope (ASM) doesn't work that way (i.e. measure sagitta). Here's how it works (in this case for concave surfaces; for convex it requires a relay lens).

 

 

 

Your starting position is the reading you get when the MO forms a sharp Airy disk image in monochrome light. At this point the x-y table micrometers are zeroed out, and the ASM carriage is translated along the optical axis (Z) until you see a second Airy disc formed by the mirror  or lens . The horizontal distance is read to a full mm using a vernier scale on the ASM base, and the XY table is used the image to a sharp focus. The exact position then is the axial distance from the first setting down to a full mm + the distance it took to move ASM with micrometers to a snap focus. 

[ThomasM]

An autostigmatic microscope (ASM) doesn't work that way (i.e. measure sagitta). Here's how it works (in this case for concave surfaces; for convex it requires a relay lens).

 

asm setup for ROC_1 (2023_10_30 21_12_06 UTC).jpg

 

asm_vernier.jpg

 

Your starting position is the reading you get when the MO forms a sharp Airy disk image in monochrome light. At this point the x-y table micrometers are zeroed out, and the ASM carriage is translated along the optical axis (Z) until you see a second Airy disc formed by the mirror  or lens . The horizontal distance is read to a full mm using a vernier scale on the ASM base, and the XY table is used the image to a sharp focus. The exact position then is the axial distance from the first setting down to a full mm + the distance it took to move ASM with micrometers to a snap focus. 

 

Thanks, so the radius is the difference between d1 and d2. So you need to measure this with high precision, the Mitotoyo mentioned in post #9  can measure up to maybe 10mm, with 0.002mm, but I would expect the accuracy of the radius of curvature to be limited by the accuracy of measuring the long distance with the ASM. What is the accuracy of this measurement? Or am I missing something?
 

Edited by ThomasM, 29 March 2025 - 03:54 AM.

[BGRE]

In post #11 Preston states that a digital scale with a resolution of 50 microinches (1.27 microns) is used to measure the difference positions of the ASM between the cateye and confocal positions. The setting accuracy depends on the test surface NA (as viewed from its RoC, provided the test surface is fully illuminated) the microscope objective NA and residual aberrations of the microscope objective etc. 1 wave of defocus at F/8 is equivalent to an axial distance from true focus of 40 microns.

 

The same type of RoC measurement can be made with a spherical wave interferometer with somewhat better accuracy as the interferometer can actually measure residual defocus error etc. The cateye position has the added complication that the return beam is clocked by 180 degrees with respect to the incident beam.

Edited by BGRE, 29 March 2025 - 04:43 AM.

[MKV]

Thanks, so the radius is the difference between d1 and d2. So you need to measure this with high precision, the Mitotoyo mentioned in post #9  can measure up to maybe 10mm, with 0.002mm, but I would expect the accuracy of the radius of curvature to be limited by the accuracy of measuring the long distance with the ASM. What is the accuracy of this measurement? Or am I missing something?

It depends on the level of accuracy desired. For best (mechanical) precision, a radius bar is configured with interlocking segments which are cut to size and measured individually with calipers (which should be calibrated). Nobody here needs or can achieve an ROC to the precision mentioned earlier, but every device needs to be calibrated so that we know what we are measuring and how much of the number we can trust, just so we don't get too carried away with too many significant figures. :o)

 

For precision optics (Mak correctors, etc), radius of curvature tolerances are in the range of +/-0.1%. For optics beyond ATM dreams, even as low as +/-0.01%! But for us (CN crowd) rarely will we need anything better than a few 10ths of a mm accuracy. For example, an ROC of 1 meter (1000 mm) 0.1% = 1 mm. For a small lens with an ROC of 100 mmm 0.1% = 0.1 mm. 

 

We should keep in mind that a telescope consists of an objective and auxiliary optics, each of which has it's own flaws, plus of mechanical supports and devices, which are habitually never accounted for. Then there is alignment, temperature, atmosphere, and so on to disrupt the wavefront. The only way to know how good our telescopes are is to test the exit pupil wavefront under actual observing conditions. This is usually done visually, which is highly subjective and also susceptible to visual acuity due to age, diseases, genetics, etc., but almost never done interferometrically (as it should be). 

 

 

[duck]

proceeding as if everything is fine...channeled the laps with the 1/8" ball end mill.  Made square 1 1/2" facets, only 0.131" max depth.  Took 3 hours on the mill.  Stoned the channels.  Close examination of surfaces is intriguing.  Seems like it's quite grainy. 

Edited by duck, 29 March 2025 - 08:52 PM.

[MKV]

Here is a very informative summary of what needs to be met to make precision lenses and mirrors, such as attempted by Duck and Preston. Highly recommend that everyone interested in high-ended optical production know this. From the University of Arizona, a summary table of most important tolerances to meet:

 

ref: https://wp.optics.ar...-components.pdf

 

Or you can actually complete a course 415/515

 

This will be very helpful in calibrating spherometers and centering optics, etc.

Edited by MKV, 30 March 2025 - 02:20 PM.

[Arjan]

Duck, indeed when switching mot and tot too quickly you may end up with a rough surface. When changing RoC this is usually not happening equally everywhere, so you need to wait for spherical equilibrium before the switch.

E.g. when deepening the curve mot, this will start in the center. Subsequent tot counteraction lowers the edge, so you may get a high zone halfway as a result. Shooting for a specific RoC needs to be done carefully.

[duck]

agreed!  It's possible to switch too frequently.  The ROC I'm shooting for is 48".  What I have measured occasionally is an ROC mismatch between convex and concave laps of about 0.2"  Therefore, the sagitta error for these 6" diameter laps is 3^2/(2*48^2) * 0.2" = 0.00039" = 17.8 waves!  That's 9 waves on each surface.  But that's when I take no extra precaution (with 60 micron alumina) to get them closer.  With very gently stroking, I've had the ROC mismatch below 0.1".  Most of the effort on the laps has been to take out tooling marks left when I generated the curves on the mill.  I've got to get a better technique, because it was chattter at the end of a pass with the cutter that left some 5-6 thousandth's gouges.  I should have made another cut, but I didn't think it would be that time consuming to get them out.  WRONG!  

 

 

The laps are channeled and set aside.  Next step in this 24" Cass build is to grind and polish the secondary's back side.  Got the flat lap on the grinding machine and am waiting for glass delivery. 

 

I watched the S&T video with StellarVue again.  Vic Maris made a point of telling Dennis di Dicco that StellarVue makes their own cast iron laps.  I get the impression that making the laps is a critical step in production of a new lens.  Just what is the testing protocol for a 3 element apo lens?  I bet the laps are at the exact ROC of the prescription.  The glass is ground against the lap and a nearly worthless spherometer reading is taken.  The glass is polished and this is where I don't know exactly what testing on each individual surface is performed.  Does StellarVue hand figure each surface into a perfect sphere...using a test plate for the convex surfaces?  Or does StellarVue go directly to a complete objective before any testing and figuring?  This is not relevant to my mundane optical proficiency, but interesting! 

Edited by duck, 31 March 2025 - 11:48 AM.

[Oregon-raybender]

If this is the reading you are getting with 60 micron is no

concern. As you go to smaller the grit size the ROC of both

will match quite well. By the time you use 9 micron you should be

really close.

 

Yes the laps are key to production. I would make 3 cast iron tooling

sets. The lens is generated to ROC, as I noted slightly flat to the 

requirement. One tool for rough (25 micron grit), one for 12 Micron and last one

for 9 and 5 micron. Each one would be corrected as I noted above in

the method. I still would use the sphereometer as I went along for a single 

lens (50mm and larger) 

 

With Stellarvue they would do the same. I would guess they would use a test plate

and get within a 1-3 fringes (Polish as needed to get the performance

they want.)  It is only if the assembly test shows some error that

they would work on one or two of the surfaces. By now they would

know what surfaces to correct. The question is for them, they may have

a CNC grinder / polisher. If so, not need for a testplate, since the machine

would profile the ROC after each run.

 

 

Starry Nights

 

https://www.youtube....h?v=gz2Pm3DBOds

[MKV]

What I have measured occasionally is an ROC mismatch between convex and concave laps of about 0.2"  Therefore, the sagitta error for these 6" diameter laps is 3^2/(2*48^2) * 0.2" = 0.00039" = 17.8 waves!  That's 9 waves on each surface.

...

 

The glass is ground against the lap and a nearly worthless spherometer reading is taken.  

I presume you use different equations to determine the ROC/sag of concave and convex surfaces.

 

You can find the equations in Advanced Telescope Making  -- Mechanical, on spherometers.

 

There are high accuracy spherometers that are precisely calibrated. The same has to be done with DIY spherometers if exact equal but opposite ROCs are required. Spherometers can also be certified with an autostigmatic microscope ROC readings, or simply by using a spherometer on a known precision surface.

 

Calibration of your spherometer is covered also covered in Advanced Telescope Making -- Mechanical, by William Browne.

 

__

edit Advances --> Advanced

Edited by MKV, 31 March 2025 - 08:18 PM.