Calling out Optical Tolerances

Hey all,

I've got a question about calling out optical tolerances. I'm curious as to what's considered easy, challenging, difficult, or impossible tolerances for optical specifications such as radius of curvature, thicknesses, surface irregularity and the like. For example, a mechanical (not optical) tolerance of +/-0.005" would be considered easy while a tolerance callout of +/-0.00001 (ten millionths) would make the machinist either say you're plumb crazy or ask you if you made a typo (depending on his mood that day).

Cost is a function of optical tolerancing, so this would be useful info. I understand that shop tolerances will vary depending on where you go, but what will typical make them go "oh this is cake" and what will make them go "ouch! how big is your budget?"

Thanks!
Jason

This is very dificult to answer and will bring out lots of fights and discussions. Let me just give you some of my generalizations from an ATM perspective. Commercial perspectives will be different.

If 1/4 wave is considered diffraction limited and each color in the visible spectrum has a different wave lenght which do you use as the standard? I normally use green light because it is about the middle of the spectrum and it is the wavelengths that the eye is best at collecting.

What is this in micro inches well. Green light is about 550 nanometers. This converts to .0000216535 inches. 1/4 of this is .000005413375 inches. So for the mirror to not have any defects in theory that can be distinguished at the eyepiece the surface must be within this tolerance.

Ok so now we have some numbers to look at.

Three main things affect how easy or hard a mirror is to fabricate.

1. It thickness as a ration to its diameter.

2. Its F ratio

3. And your testing equipment and methods.

Lets take the first item. I have been told that it is easy to do a mirror with a span under 200 and hard to do one over 200. Span being defined as mirror thickness ratio times by mirror diameter. I don't have enough experence to prove this but it is a start. I did two six inch mirror one inch thick. This gives a span number of 36. They were pretty easy. I am working on a 16 inch mirror 7/8 inch thick which has a span number of 292 and have not been able to get it finished yet.

The second item F ration. F5 to F10 are pretty easy. Faster than F4 and the lap used to figure the mirror has a hard time changing shape fast enough to correct the mirror so you have to use smaller laps and some exotic technics.

Third the testing. It is pretty easy to introduce an error into your testing results with numbers as small as .000005. Any flexure of the mirror, the mirror support or the test apperatus will taint the results.

Hope this helps
Dave

Hi Jason

I find a tolerance of +/- .002” on center thickness and diameter to be

fairly easy to hold. And the radius tolerance is better specified as a percent

of the radius .1% - .5% a higher % as the radius increases. because a
1" radius +/- .001" on say a 2" dia. part is not to hard to do but a 10" radius
+/- .001 on that same 2" dia. part would be very difficult. I have done some
very difficult lenses in my times ranging from a .040" roc +/- .0005" up to a
25" roc +/- .040". The 25" radius lens was a 2.6" square lens made of sapphire
and it had a center thickness of about .125" +/- .001" It also had to be polished
on the edges and the bevels. I have also done thickness tolerances as tight as
+/- 10 millionths but that was on flat work. One of the most important things is
that the shop you go to has the equipment that can accurately measure what
you are looking for.

Hope this has helped
Ed.................................