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

Dimensions For Ring Spherometer and Foucault ?

ATM mirror making
  • Please log in to reply
80 replies to this topic

#1 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 09:15 PM

Lots of resources online for making spherometers and Foucaults, however pretty much every design assumes the maker does not have access to a well equipped machine shop. I'm getting popsicle sticks and razor blades etc.... Since I am a machinist with his own shop, i think i can do better than most, once I get a good design happening. I have dial indicators and micrometer heads but I don't have a good idea about the dimensions of these instruments.I am going to be grinding my first mirror and am tentatively planning a 14 inch F6..... i am told that would be a nice general dob and i already have a 10 inch. I read that ring spherometers are superior to the kind with 3 legs, but have 3 carbide legs on reserve, just in case someone talks me out of that. 

 

As far as the Foucault goes, I can't even find any professional designs at all. What size is the optical slit ? How much adjustment does the slit need ? Do i put together an X-Y table and if so how much travel do I need on each axis ??

 

Metric or Imperial... makes no difference to me, but I'm leaning metric.

 

Any suggestions ?

 

Thanks gang !!



#2 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 17 October 2019 - 09:31 PM

See
https://www.trioptic...ts/spherometer/

for examples of high accuracy spherometers.

Ring spherometers tend to become somewhat massive as the size increases and require accurately hardened and ground rings.
Up to 6" or so diameter they are manageable.
One drawback is they are not suited to detecting astigmatism whereas this presents no difficulty to a 3 ball spherometer.

The larger the size of spherometer the more accurately you can measure the sagitta.
For a 14" a 12"/300mm diameter spherometer is nice. However a 6' one may be useful for detecting irregularities at the fine grinding stage.
A spherometer head with interchangeable "arms"/disks with various effective diameters (as in the spherotronic) can be useful.
One could either use ruby balls, or even tooling balls.

#3 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 09:46 PM

Well, that was fast !! I'm already leaning more towards the 3 legged spherometer, for the following reasons: A cylindrical spherometer would have to have a perfect 90 degree at it's circumference. I'd want it hardened and ground. That means a hardened sharp edge. Any chance of sliding it while using it would scratch the glass. While these carbide legs are even harder, at least they are rounded. I think that would be much safer. I look forward to hearing other opinions.



#4 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 09:53 PM

One could either use ruby balls, or even tooling balls.

Thanks. Your post almost landed on top of my adjacent update.. I'm against ruby balls, although I have seen some people using them. It's overkill on hardness. You can buy, for very cheap, carbide fingers that are used as replacements on the lever type dial indicators. They are about a buck a piece, so I bought 4. Unfortunately, the threads on them are really tiny so i may be shopping for bigger ones.

 

The biggest issue I have with Ruby is that the harder they come, the easier they crack. I can easily see getting a crack or chip in one and, without knowing, putting it on my nearly finished mirror and putting a nasty scar in the glass. Carbide is hard enough.



#5 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 17 October 2019 - 09:53 PM

A Foucault tester needs an xyz table with travel along the test system axis at least equal to the axial distance between the paraxial and marginal foci of the test surface
For a paraboloid this distance is:

L= (1/4)*D^2/R

for a 350mm diameter F/6

L ~ 7.3mm

With some ingenuity a Foucault/Ronchi tester could be assembled on a carrier plate that mounts to one of the Line-tool 3 axis stages.
The Newport 462 stages are also good.

I have a copy of blueprints for a classical 3 axis optical tester (Foucault/Ronchi/wire) that uses phosphor bronze bearing sliding on flats and V-grooves machined and ground into hardened tool steel plates.
The tungsten halogen light source could be upgraded to a LED

DRO encoders could be used to measure displacements along x y z axes.
This works well with a radius slide should you want to make high accuracy RoC measurements etc.

Edited by BGRE, 18 October 2019 - 03:27 AM.

  • CharlieM likes this

#6 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 09:59 PM

Here's what I bought for legs but, as I said, the threads on them, an M1.6 , belongs in someone's watch. They are just to small and sure to snap off. you think getting a broken screw out of something is hard ....... https://www.ebay.ca/...872.m2749.l2649

 

There are bigger ones.



#7 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 10:09 PM

Here we go !!! Take your pick.

 

https://www.ebay.ca/...-50ySObiVLbYdNA



#8 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 17 October 2019 - 10:12 PM

Vanguard ... If I made say a 6 inch Spherometer (or worse, a 12 inch) with three legs, how does one measure out to the edge ? The closest I would get is 3 inches , or a tad closer, in from that edge .... no ?


Edited by CharlieM, 17 October 2019 - 10:18 PM.


#9 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 17 October 2019 - 10:32 PM

A 6" spherometer with a suitable depth sensor can be used to measure the sagitta at various radial locations to check for uniformity of curvature (constant sagitta) over the surface during grinding.
This isn't feasible with a 12' spherometer. The Heidenhain certo/metro gauges are nice but the readout/display isn't cheap.

How accurately do you want to measure the RoC with a spherometer?

A 350mm F/6 has a sagitta of about 0.67mm when measured with a 150mm/6" spherometer
with a 12"/300mm spherometer the corresponding sagitta is about 2.68mm

For example 0.1% error in RoC requires a depth sensor error below 0.67microns with a 6" spherometer.
For a 12" spherometer the corresponding depth sensor error can be 4 times this or about 2.7 microns.

Once the surface is polished a point source microscope and a radius slide are far more accurate than a spherometer.

Another requirement is a suitably large and accurate calibration flat for the spherometer.

Edited by BGRE, 17 October 2019 - 10:34 PM.

  • MKV and CharlieM like this

#10 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 17 October 2019 - 10:44 PM

Here's a couple of typical line tool 3 axis stages:
https://www.ebay.ca/...agAAOSwQypdO6Uc
https://www.ebay.ca/...QUAAOSwyqhdNKk8
here's one with submicron setting resolution:
https://www.ebay.ca/...PQAAOSwA91Z3VRn

A newport 462 series 3 axis stage:
https://www.ebay.ca/...jEAAOSweN9daDPS

A Newport 562 series 3 axis stage:
https://www.ebay.ca/...fUAAOSw94pdYoES

#11 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 18 October 2019 - 12:19 AM

I have a good flat surface plate ... granite and I can access a bigger one when needed as mine is only about 10 X 12 inches. Those stages are nice but expensive. I can bide my time until one comes up on the cheap. In the meantime I may make one with extra long dowel pin slides and micrometer heads. Wondering if my lathe or vertical mill can be adapted for this ? The bed is 36 inches long and the distance from center to bed is about 6 inches. No doubt not nearly long enough, but I must say, I could probably recreate my bed ways easier than an x-y table and I could lift the saddle, compound rest and all off the lathe and put it on this recreated bed. I do have some cheap chinese DRO's on it. We also have wreckers nearby and I've seen some nice long chrome plated , hardened and ground shafts there. If my rough calculations are correct though, I think I need about 15 feet of length between mirror and tester. I could also head over to my alma mater University and I'm guessing they have a great optics lab... but the back and forth ....

 

I need to bulk up on my studying too as these Youtube videos always leave me with more questions than answers. I bought some muriatic acid today. Painting the basement floor in preparation for this mirror adventure. Coincidentally, the acid can be used to blacken steels and I am thinking that the components for the foucault would probably benefit from a non reflecting coating.

 

I guess grinding Schmidt Cassegrains are out of the question.



#12 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 18 October 2019 - 12:50 AM

I can't answer the question of how accurate do I want to measure the ROC as I have not even ground a mirror yet. This is my novice preparation. I can't see myself not needing a spherometer for frequent checks. I am starting to think up some interesting ideas for the three legs that would avoid too much scraping. I also am worried about too much point pressure from the legs and going with bigger balls just means inaccuracies from the offset of the indicator axis and ball contact point. 

 

What I'm thinking is something quite light weight, say cast aluminum with lots of holes in it. Three dowel pins with small rotatable ball bearing ends that sit on bronze bushing material and a neodymium magnet tucked deep in the pin end, to hold the balls in place. I do have ball end mills so I should be able to put a nice seat in the ends of the dowel pins. Are these legs "set and forget" ? Is there any advantage to having them adjustable ? If not, can you suggest a good length ? I'm thinking a 4  inch diameter for quick measurements of rough grinding progress. I can always build other sizes later. Another article suggested that as an ideal diameter because it keeps the math easy. Something to do with taking the reciprocal. 


Edited by CharlieM, 18 October 2019 - 01:07 AM.


#13 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 01:31 AM

Bigger balls are better, less stress on the test surface. The correction for the ball diameter is part of the spherometer calibration along with the effective diameter.
Since you have to correct for the ball diameter I'd ignore that illusory advantage. With a calculator or suitable spreadsheet having the spherometer constants already entered a so called "odd" diameter isn't a disadvantage.
The accurate calculation of the RoC from the spherometer reading is a bit more complex than a simple reciprocal.

Set and forget is normal followed by calibration which will hold well if there are no adjustments (or wear).
Trioptics bond the balls in place.

I'll see if I can find some images of other professional spherometers.

With a 100mm effective spherometer diameter the sagitta measured with an 350mm diameter F/6 is about
0.3mm which using a standard metric micrometer (0.01mm resolution) would achieve an accuracy of around 6%in RoC.
1% RoC initial accuracy usually suffices for a newtonian primary until you come to mount it when measuring the focal length to within a mm or better is desirable.
0.05% or better measurement of RoC is usually only required for refractor objectives etc.

The flatness of a surface plate is adequate when using a standard micrometer to measure the sagitta.
It will be inadequate if a sagitta measurement error of 1 micron or less is desired.
An optical flat will be required.

#14 Cary

Cary

    Vendor (Optic Wave Labs)

  • ****-
  • Vendors
  • Posts: 960
  • Joined: 07 Jul 2004
  • Loc: Rancho Cordova, CA

Posted 18 October 2019 - 01:36 AM

You're overthinking things.

 

Start grinding a mirror.

 

Use a steel rule and some feeler gauges to determine your sagitta.

 

Once your mirror is ground start polishing it.

 

Print out a Ronchi grating on your laser or inkjet printer on some overhead transparency material. 80 lines per inch.

 

Figure out how to set up the ronchi test.  Take a peek and then forget about it and keep polishing.

 

Polish some more.

 

Polish some more.

 

Polish some more.

 

Polish some more.

 

Polish until you start to cry.

 

Keep polishing.

 

After you are sure that you are crazy then start checking the polish with a red laser pointer.

 

When the laser pointer spot cannot be seen on the polished and clean surface of the glass all the way to the edge then your glass is ready to prepare to figure the parabola.

 

Keep using the ronchi grating test to fix the sphere.  Focus on making sure the edge of the mirror is good ( not turned down) and smooth out any zones.

 

Now you are ready to worry about a knife edge test.  Skip the spherometer.


Edited by Cary, 18 October 2019 - 03:32 AM.

  • PrestonE, Pierre Lemay, dave brock and 4 others like this

#15 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 01:45 AM

Here's a 3 leg spherometer from the 1970's or thereabouts:
Spherometer3Leg.jpg

N.B. each ball tipped rod is clamped in place using a flexure machined into the end of each arm.
These are easily done preferably using a slitting saw and a drill press or equivalent. However a bandsaw or even a hacksaw could be used in the absence of a slitting saw.

Edited by BGRE, 18 October 2019 - 01:53 AM.


#16 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 02:29 AM

For Ronchi or Foucault testing the mirror test stand and the Foucault tester need not sit on the same bench a survey tripod oor similar could be used to mount a compact test head. Some care in the test stand mirror edge support design is probably warranted. I used piano wire flexures with integral over travel stops (quick and easy to do with a lathe).

#17 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 04:15 AM

https://www.trioptic...es-accessories/

shows a few variants of various spherometer mounting arrangements for the 3 ball contacts etc.

#18 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 18 October 2019 - 06:23 AM

Great advice guys. Very cool tips. I'll start working this over the weekend. Good thing I have an indexing head. I do have a depth micrometer but thought indicators were superior, in these applications. Lots of learning happening too.



#19 Mark Harry

Mark Harry

    Cosmos

  • *****
  • Posts: 8281
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 18 October 2019 - 06:55 AM

I have a "truncated" aluminum ring spherometer, where the curve on any particular mild curve such as for mirrors can be tracked right to the nearest mm. I made it over 20 years ago, and it has worked just fine-- for hundreds of mirrors.
Using some sort of straight edge, and feeler gauges, my best was around 1" or so on the same type of curves. Occasionally closer.
Up to you, but the former I mention, we use -EXCLUSIVELY- at the optical shop, and we occasionally hit shorter radii to within a few -MICRONS-, or 10's of microns using a Millimess gauge with testplate, or a digital dial indicator with 1/2 and 1 micron rez. Curves from flats, to hemispheres.
We have no 3-ball sphero fixture in the shop at all-- period. An unnessesary complication.
*****
"....needs an xyz table...."

No it doesn't. We have a fixture that's confined on one axis (the table) on all 3 IFs for testing to micron rez, but  for common sense, it's just not necessary with Newt mirrors here. Proof of that, just deliberately introduce a variant along the mirror axis into a spreadsheet, and observe the change in error measured. The -VAST- majority of the time, the difference is statistically insignificant, if detectable at all.
*****
The biggest thing to help in identifying 'stig that I've seen with KE, is to come up with a good light source, to slit arrangement where the distance from the true axis is within 1/8" (3mm) or less. It can be done, and the old myth that 'stig can't be seen is proven to be hogwash.


  • CharlieM likes this

#20 Cameron_C

Cameron_C

    Mariner 2

  • -----
  • Posts: 241
  • Joined: 05 Feb 2014
  • Loc: Richmond Hill Ontario Canada

Posted 18 October 2019 - 08:14 AM

You're overthinking things.

 

Start grinding a mirror.

 

Use a steel rule and some feeler gauges to determine your sagitta.

 

Once your mirror is ground start polishing it.

 

Print out a Ronchi grating on your laser or inkjet printer on some overhead transparency material. 80 lines per inch.

 

Figure out how to set up the ronchi test.  Take a peek and then forget about it and keep polishing.

 

Polish some more.

 

Polish some more.

 

Polish some more.

 

Polish some more.

 

Polish until you start to cry.

 

Keep polishing.

 

After you are sure that you are crazy then start checking the polish with a red laser pointer.

 

When the laser pointer spot cannot be seen on the polished and clean surface of the glass all the way to the edge then your glass is ready to prepare to figure the parabola.

 

Keep using the ronchi grating test to fix the sphere.  Focus on making sure the edge of the mirror is good ( not turned down) and smooth out any zones.

 

Now you are ready to worry about a knife edge test.  Skip the spherometer.

Cary, I think you may have mistakenly missed the next step which is to "Just Keep Polishing".



#21 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 09:16 AM

Millimess gauge:
https://www.mahr.com...-dial-face--┬Ám/
Mahr also make these for Mitutoyo.
Going rate for a single testplate were well over $1000 (in 2014) per plate (for large test plates ) unless one makes ones own but then an accurate means of calibrating them such as a radius slide and interferometer are required for the highest accurcay.
Zygo also provide multi-axis mounts for holding test parts when measuring these with an interferometer.

Only a small adjustment range is needed as the coarse positioning is easily accomplished by sliding the test head or the test part along the table or equivalent.
If the lateral separation between knife edge and source is d then the field astigmatism exhibited by the test surface is
(1/16)*(d^2)*(D^2/R^3) the acceptable value if d varies with mirror diameter and its f#.
An axial separation d of 9mm results in about 10nm of field astigmatism when testing a 350mm F/6 mirror.
The visibility of any astigmatism depends on the relative orientation of the ke to the astigmatism axes (there are 2 orthogonal axes).
1945 paper on detecting astigmatism with the Foucault test;
http://adsabs.harvar...MNRAS.105..193L
The Foucault test chapter in Malacara's Optical shop testing mentions the rotation of the knife edge shadow with defocus in the presence of astigmatism if the knife edge isn't aligned with one of the axes of astigmatism.

Edited by BGRE, 18 October 2019 - 05:14 PM.


#22 CharlieM

CharlieM

    Explorer 1

  • -----
  • topic starter
  • Posts: 56
  • Joined: 05 Apr 2011
  • Loc: London, Ontario, Canada

Posted 18 October 2019 - 06:32 PM

I have a "truncated" aluminum ring spherometer, where the curve on any particular mild curve such as for mirrors can be tracked right to the nearest mm. I made it over 20 years ago, and it has worked just fine-- for hundreds of mirrors.

Using some sort of straight edge, and feeler gauges, my best was around 1" or so on the same type of curves. Occasionally closer.
Up to you, but the former I mention, we use -EXCLUSIVELY- at the optical shop, and we occasionally hit shorter radii to within a few -MICRONS-, or 10's of microns using a Millimess gauge with testplate, or a digital dial indicator with 1/2 and 1 micron rez. Curves from flats, to hemispheres.
We have no 3-ball sphero fixture in the shop at all-- period. An unnessesary complication.
*****
"....needs an xyz table...."

No it doesn't. We have a fixture that's confined on one axis (the table) on all 3 IFs for testing to micron rez, but  for common sense, it's just not necessary with Newt mirrors here. Proof of that, just deliberately introduce a variant along the mirror axis into a spreadsheet, and observe the change in error measured. The -VAST- majority of the time, the difference is statistically insignificant, if detectable at all.
*****
The biggest thing to help in identifying 'stig that I've seen with KE, is to come up with a good light source, to slit arrangement where the distance from the true axis is within 1/8" (3mm) or less. It can be done, and the old myth that 'stig can't be seen is proven to be hogwash.

I'd hate to trespass on anyone's generousity with their time and advice here, but do you have any photos of the "truncated ring sphero" ?

 

I see the Millimess all over the net but I've noticed that the adequately precise ones, say 5 micron gradients,  have a very limited range of measurement.

 

Sorry your abbreviation... what's an "IF" ?

 

You mentioned a digital indicator. I had read in a couple of places that the digitals are far less useful than the mechanical movements on account, they can't be swiped across surfaces to see any trends.... i guess is the handy word. I could see their superiority on making static measurements though.

 

I wish someone would put up a really thorough tutorial on ATM grinding . The instrumentation options, measurements, calculations. The few that I see on Youtube, seem to be restricted to grinding technique and not tracking progress or metrics.



#23 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 07:11 PM

IF = interferometer, in his case a couple of Zygo Fizeau interferometers with phase shift capability and a selection of transmission spheres etc.

By truncated ring spherometer the easiest way to envisage this is to take a full ring spherometer and cut out sectors leaving something like 3 small ring arcs spaced at 120 degrees around the original circumference. The result being something like the last image I posted with a short circular ring arc replacing each ball foot. replaced by short

#24 Dale Eason

Dale Eason

    Apollo

  • *****
  • Posts: 1321
  • Joined: 24 Nov 2009
  • Loc: Roseville,Mn.

Posted 18 October 2019 - 07:40 PM

 

 

I wish someone would put up a really thorough tutorial on ATM grinding . The instrumentation options, measurements, calculations. The few that I see on Youtube, seem to be restricted to grinding technique and not tracking progress or metrics.

https://stellafane.o.../atm/index.html

 

Like the other person said you are way overthinking this for your first mirror.  Your not going to need a sphereometer for it.  They can be convenient but not necessary in the least.

 

Do a web search of John Dobson how to make a mirror.  I have it on DVD but it is also out there on the web.  

 

The mistake many machinists make when thinking about making their first mirror is the techniques they normally use to force a piece of steel to the shape they want using precise cutting tools.  Mirror making just uses the natural process of grinding using gravity and proper learned technique and the mirror naturally turns into a sphere.  Sort of like magic.  All done without any precision tools other than a normal house hold ruler.

 

Dale 



#25 BGRE

BGRE

    Mercury-Atlas

  • *****
  • Posts: 2505
  • Joined: 21 Mar 2016
  • Loc: New Zealand

Posted 18 October 2019 - 08:00 PM

That may be true for a 6" or 8" mirror where a 10% error in sagitta is relatively inconsequential.
However for the 14" F/6 proposed a 10% error can result in needing to stand on something to access the eyepiece near zenith.
Dobson's (I met him some years ago when he came to NZ) methods can be a bit "slapdash" and can lead to poor results.


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





Also tagged with one or more of these keywords: ATM, mirror making



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics