60 replies to this topic

[elliotwoods]

What you call "waviness" is a 3D rendition of a wavefront map. The waviness is the result of defocusing, which places a reference focus at diffreent zones of the wavefront corresponding to different zones on the aperture When high order and low order spherical aberration are perfectly balanced you have a minimum rms wavefront residual (or best focus), corresponding to a reference focus of the D√0.5 or 70.71% zone.

 

Mladen

Thank you Mladen. 

I could follow along well up to where you discuss the 'reference focus'. I'm having trouble understanding this part.

I'm guessing this means that 'waviness' causes the focal length to vary across different zones of the mirror, hence at the camera image plane you would have a point spread function (in the case of a primary telescope mirror).

 

Particularly the last part "corresponding to a reference focus of the D√0.5 or 70.71% zone."

D here is the diameter or diopter? Sorry if i'm missing some basic knowledge.

[elliotwoods]

I don't want to confuse, (Sorry about that ) my mention of power is the base curve of a flat (it is really a long focus mirror by the true meaning) The waviness I am telling about is the result of production waves. The test we did on DLP mirrors is to look for this and cut the out of spec section.

I attached a paper on the subject used by the mirror / window industry, our parallel optical universe. Their is scale of flatness is several decimals lower than our universe. I had to learn the language of one industry and apply it to OCLI to meet our requirements. I attached more information so you can see how the other industry tests the production process of flat glass. I hope this clears up my definition of waviness for DLP mirrors

(float glass) 

 

If you look for waviness in glass on the web, there are hundreds of sites of companies who make test equipment, there is even

international standards that is required.

 

https://www.glassonw...ns-curved-glass

 

https://www.astm.org...dards/C1652.htm

 

The open chamber I mention is vacuum chamber that has several sections of vacuum. First it was washed, then scrubbed with polishing compound, cleaned, dried, placed in the first chamber, then two more until full vacuum for coating, then reversed, The idea was to run a continuous flow of glass, 24/7/365.  

 

I still believe that 20-30 fringes of flatness (power) will not be seen by the eye, but waviness will be. The eye allows a small

about of defocus as seen. But the eye is very good at seeing small distortions.  That is why waviness was so important to OCLI,

small amounts would result in distortions projected on the TV screen.

 

I think you can open your tolerance more and still have good results.  A prototype using a .75 meter mirror stock should show this.

 

A good mentor should always assist one who seeks knowledge or skill.  Good Luck.

 

Starry Nights

 

https://www.aisglass...ers/AIS-303.pdf

 

http://ianweekes.wee...oller-wave.html

 

https://www.strainop...dge-lift-gauge/

 

https://www.glassglo...-10s-30031.html

 

https://industry.gla...lmayberoll.html

 

https://www.glastory...ms-roller-wave/

Thank you for your instincts on power and waviness.

 

A few of these links highlight a suspicion that I had:

That the tempering process that is commonly applied to architectural glass (especially here in Korea) is a major driver of distortions in the glass

aka "Roller wave distortion"

These seem to lie around the Mode ~10,5 on my scheme above (https://www.cloudyni...2#entry10813826 )

 

Especially this paper you linked is great for highlighting this issue:

https://www.aisglass...ers/AIS-303.pdf

They don't actually say how their proposed system works since they haven't got their patent as time of writing, but I can imagine a few machine vision approaches which would do what they claim.

 

Interestingly they state that the P-V of roller wave distortion can be measured with a gauge 0.025mm (0.001 inch) accuracy (which is shown in more detail in your other links)

And that their system is accurate to 0.01mm (0.0005 inch)

And with the other examples, it seems the P-V of rollerwave distortions are on the order of 100 lambda

 

It's been great to go through all these examples because there are lots of examples of distorted glass and their visual effects.

 

 

All-in-all

The best way to reduce roller-wave distortions is going to be to not temper the front layer of glass.

This will also be necessary if we are to do any grinding or other processing

 

We will meet with our glass supplier next week, and it's great to have all this knowledge so that we know what to ask for.

[Gleb1964]

Thank you Mladen. 

I could follow along well up to where you discuss the 'reference focus'. I'm having trouble understanding this part.

I'm guessing this means that 'waviness' causes the focal length to vary across different zones of the mirror, hence at the camera image plane you would have a point spread function (in the case of a primary telescope mirror).

 

I think discursion has went aside and not relevant to your actual problem. Your mirror does work as field element, not as aperture.  Human eye has a few millimeters opening and when observing remote object trough multiple reflection on your mirror, accepted beam has a miserable footprint compare to mirror diameter. The main problem is distortion caused by local slope change or local tilt of mirror surface. I think that relevant method of control is sort of observing grid distortion after been reflected in the mirror. 

 

Gleb

[MKV]

The main problem is distortion caused by local slope change or local tilt of mirror surface. I think that relevant method of control is sort of observing grid distortion after been reflected in the mirror. 

True, but the OP is talking about making 1.8 meter optical flats presumably by a three-disk method. Doesn't this presuppose that the surfaces will be solids of revolution with  some degree of regularity rather than randomly distorted mass produced sheets of glass? The randomness of the surface may arise from the annealing process of the substrate and the mechanical manner of supporting the flats, but that's an altogether different subject. 

 

Mladen

[MKV]

Thank you Mladen. 

I could follow along well up to where you discuss the 'reference focus'. I'm having trouble understanding this part.

I'm guessing this means that 'waviness' causes the focal length to vary across different zones of the mirror, hence at the camera image plane you would have a point spread function (in the case of a primary telescope mirror).

 

Particularly the last part "corresponding to a reference focus of the D√0.5 or 70.71% zone."

D here is the diameter or diopter? Sorry if i'm missing some basic knowledge.

My apologies, Elliot. I was referring to your 0.1 series of wavefronts. And your guess is correct in principle. I was assuming they represented ground, polished and figured flats, not ordinary sheets of glass.

 

Also, D stands for (clear) aperture diameter (i.e. physical diameter - 2 times the width of the edge bevel). 

 

I'll just keep quiet from now on, as I think things are beginning to be somewhat confusing, as Gleb aptly observed.

 

Mladen

[Oregon-raybender]

Building a prototype scale model I am sure will help here. Float mirror stock (which we used at OCLI) (verses rolled) should do fine.

 

Mounting it is a different issue. I would suggest Silicone (RTV) for bonding with a small plastic ball ( 3-5mm) diameter to maintain thickness of the bond, with many supports. I would so far as suggest thick carpet, like Bob Kestner used on thin mirror polishing support in TM magazine.

UV cement likes to pull on the glass to the metal frame. Paul Yoder's book on mounting optics maybe of help.

 

I agree, don't want to confuse the issue. It comes to point where computer modeling ends and building a proof of concept begins.

 

I think we have presented enough data to support the project. So I will keep quiet too.

 

Starry Nights

[MKV]

I agree, don't want to confuse the issue. It comes to point where computer modeling ends and building a proof of concept begins.

 An empirical, working model is the proof of science even if it doesn't correspond to reality. Case in point: The Ptolemaic navigational system, based on observation science that the sun orbits the earth instead of the other way around. The model still works to this day! :o)

[Pinbout]

Building a prototype scale model I am sure will help here. Float mirror stock (which we used at OCLI) (verses rolled) should do fine.

 

Mounting it is a different issue. I would suggest Silicone (RTV) for bonding with a small plastic ball ( 3-5mm) diameter to maintain thickness of the bond, with many supports. I would so far as suggest thick carpet, like Bob Kestner used on thin mirror polishing support in TM magazine.

UV cement likes to pull on the glass to the metal frame. Paul Yoder's book on mounting optics maybe of help.

 

I agree, don't want to confuse the issue. It comes to point where computer modeling ends and building a proof of concept begins.

 

I think we have presented enough data to support the project. So I will keep quiet too.

 

Starry Nights

The mirror display rests at 90* to gravity and he only has rear support. It needs to be astatic.  If the level of detail (LOD) was magnified, not reduced then he would need to support the edge in some manner like a sling or whiffle tree.

 

It’s so large, I would use plop for fem analysis and use 18points with weights to keep the support forces pushing against the back (astatic) 

 

http://www.astrosurf...taz/astatic.htm

Edited by Pinbout, 20 January 2021 - 08:40 PM.

[elliotwoods]

Thanks all.

Definitely enough to go off here to get started.

 

I've got some glass on the way - so I think I need to start preparing my scaled down fixed post turntable for that (I want to practise the procedure for the larger mirror).

 

Also like Gleb suggested, deflectometry against test patterns will definitely be my favoured method (especially as I have experience with computer vision with 3D geometry). I have some ideas here and will post methods, results and source code.

 

 

Concerning supports - 

* Silicone - I'm planning to use silicone washers to mate the captive steel bolts to the glass (the bolt is captured by the back layer of glass in my glass sandwich)

* UV glue - this would only be for between layers of glass, not between the glass and external structure

* Astatic - very interesting! I've come across the constant force vs constant position problem on other projects also. I think the counterbalance system might be too complicated this time since we need to support multiple orientations. Again, might be something to think about or adapt later.

* 18 points - this is certainly possible. It looks like groups of 3 around each of the 0.6 R positions. I'll have a look into this later.

* Whiffle-tree - looks interesting and feasible. I'll have more of a look into this later when working on the external structure / stewart platform assembly

* Paul Yoder's book has a lot of relevant content (flipping through the contents on Amazon). His other book "Opto Mechanical Systems Design" is available in the national library here and also discussed mounting large mirrors. I'll try to spend some time with that.

 

So to summarise. The sections of work as I see it now are:

 

* Measurement (spherometer gauges, deflectometry with reflected patterns, if required : interferometry methods e.g. Fizeau)

* Glass assembly (FEM, water-jet cutting, glass suppliers, mounting points, UV glue)

* Grinding and polishing (fixed post turntable, grits, 3 flats method, pitch lap, wet room)

* Silvering (Angel Gilding kit, mounting and rotation jig)

* Backing structure (suspension points, whiffle-tree)

[MKV]

Good luck, Elliot! Keep us posted. And keep up your excellent art work! :o)

[davidc135]

An extra thought: If there was the ability to deform the glass surface of the scale models with respect to:

 

A. Overall curvature over the whole diameter

 

B. Astigmatic deformation due to gravity over the whole diameter

 

C. Local surface irregularities of, say, 1/6th mirror diameter

 

Could the behaviour of the full size installation be closely mimicked? Given the right thickness of glass, A would be easily adjusted by either a positive or negative air pressure and B by mechanical means. C too should be manageable.

 

David