Baffle tube build.
The 3D printed mandrel did its job quite well. I used Al foil to stop the carbon-epoxy from sticking to it and in the end I think I could have just pushed the tube off it but I started breaking it up with a pair of plyers first. The reason is that I had to make sure it was removable in bits and pieces, as the other tube for the OTA will have aluminium parts bonded to its ends, while still on the spindle, and the mandrel will have to come out in pieces.
Originally this baffle tube was going to have an aluminium reinforcing ring at the front end, but a couple of weeks ago I learned that a local manufacturer was about to start making a carbon fibre reinforced PETG filament for 3D printing. I obtained a sample quantity for trial and I used it to 3D print the front ring.
The tube has 4 layers of carbon cloth and I gave it a textured, slightly matte finish, rather than the smooth and shiny finish of commercial ones.
And don't worry about the pile of leftover plastic as it is biodegradable - made from corn starch.
I had to put this project on hold for the whole of February due to piled up pay work but now I'm getting to a point where I can start moving ahead with it again.
Over the last few days I made some improvements to my Bath interferometer that I intend on using to test the Mangin mirror for astigmatism, before and after removing the central part that has been partially trepanned.
Last night I tested a 6" Bushmaster primary mirror, or rather used it to test the interferometer - as the mirror has already been measured with the previous iteration of the Bath.
You can still see the trefoil distortion caused by the three clamps, although I have slacken them already.
I'm still overloaded with work so I made very little progress on this project.
The grinding of the front lens is finished and, before I start reinventing the wheel, perhaps someone can suggest some good ways of testing it after the polishing.
The front surface is convex with a radius of 6770 and the second one is also convex with a radius of -822. Centre thickness is 14.3
I could use the grinding tools to make test plates but R1 is way too long for that and, if I can, I would like to avoid test plates altogether.
To test a convex surface at 6770 my first thought is a concave test plate, polished spherical and test the two by putting them almost in contact (use 3 pieces of paper as spacers) and counting interference fringes (newtons rings will be visible) under a diffuse monochromatic light source. For that to work though you'll have to partially polish the surface being tested.
Alternatively make an auto-coliimator setup with say an accurate larger parabolic mirror say f/5, and adjusting the position of the light source to create a beam converging on the centre of curvature of the convex surface to be tested. Reflect this off the convex surface back to the mirror and examine the resulting focus.
There will be some spherical aberration in this due to the converging beam and umyou could either calculate the expected amount and figure the lens surface accordingly, or if you want a null test, insert a lens near the focal plane that will give the right amount to correct it. Shouldn't be hard to calculate the required lens, similar to a Ross null test.
I agree that test plates would be the preferred way but I have no idea how to test a 6" sphere with a 7m radius and prefer to avoid the extra work if possible.
However I may do that for the second surface.
I do own a precision 8" flat that will allow me to test the lens in auto collimation with a Ronchi or Ross Null test, but that would show me the combined errors of both surfaces.
So a combination of one test plate and auto collimation with the flat, should do the trick.
Another possibility I'm thinking of is to use the Bath interferometer on both sides of the lens and compare the result to what is expected from ray tracing. The reflecting (back) surface will dominate the seen error and I should be able to correct it iteratively.
Could you use the Enrique Gaviola method?
I used it when making a Cass secondary many years ago....
(See "Advanced Telescope Making Techniques", Vol 1, p63)
Ken
Could you use the Enrique Gaviola method?
I used it when making a Cass secondary many years ago....
(See "Advanced Telescope Making Techniques", Vol 1, p63)
Ken
It should work for R2 but I don't have a suitable lens.
Test R2 with a test plate, then flash polish R1 and make sure your lens focuses starlight (*) where expected (can't remember what glass you used, let's assume BK7, so for given prescription it should focus at 1414.5mm). Make sure you test it reversed (with R2 facing the sky) as this way residual spherical is much lower.
If it focuses as calculated, polish the R1 fully and re-check. If it is still at 1414.5mm or thereabouts, check if wavefront seems smooth (as per Zemax simulation attached). No need to test the lens any further.
It takes more than 20 waves of asphericity to change focus and shape of the Ronchi bands significantly. If surface was this bumpy your full size lap would be dragging and skipping and jumping all over the place.
(of course you don't need to use real starlight, any decent scope will provide good collimated beam)
Thanks Bratislav, but I think auto collimation with my 8" flat would be easier to set up than a star test and would be more sensitive due to the double pass arrangement.
Most of the power and correction comes from R2, in fact R1 has almost no influence. Here's the wavefront that lens generates when eccentricity of R1 is zero (perfect sphere), 20 (EXTREME oblate) and -20 (EXTREME hyperbola).
Spot the difference
Bottom line is R1 doesn't even need to be checked that much.
Most of the power and correction comes from R2, in fact R1 has almost no influence. Here's the wavefront that lens generates when eccentricity of R1 is zero (perfect sphere), 20 (EXTREME oblate) and -20 (EXTREME hyperbola).
Spot the difference
Bottom line is R1 doesn't even need to be checked that much.
Looks like using the Bath on R2 may be the way to go then. I do not intend to do any testing until both surfaces have been fully polished as I'm confident that the shape parameters are accurate enough.
You can definitely test R2 directly through R1 by Foucault and follow the expected longitudinal aberration (see pic #2). This surface when tested through the lens behaves like an oblate with eccentricity of approx 0.5. Problem is Bath interferometry - I have no idea how much Bath reference beam gets affected by the lens ... so no idea what DFTF would report as final eccentricity. I can model a narrow laser beam through R1, reflect it from R2 and back though R1 and tell you how much it will get deformed, but I have no idea how those two fronts will interfere (what pattern the interferogram will look like). Zemax can't do that. Or rather I can't do that in Zemax
Very good point about the reference beam. I haven't thought about that.
It may not be a problem though, as the wave front will remain symmetrical, I think, and it should only affect the amount of spherical aberration seen. A good question is whether it will add or subtract from the 0.5, and I'm assuming that I can hit the vertex of R1 with the reference beam.
And thanks for the ray tracing too - you saved me some time.
I've modelled 2mm laser beam hitting the vertex of R2. It doesn't look good. At 500mm (where Bath beamsplitter/lens would be) it looks like it gets severely affected, nearly 10 waves. See pic.
I think Bath might not be the best way to do this. Use Bath to figure the concave reference and then contact test it. You can still use DFTF or OpenFringe to analyse results ...
Last edited by bratislav; 29-04-2017 at 10:32 AM.
Reason: updated pic
I'm surprised that you got such a massive path difference. Based on a simple calculation I expected less than two waves.
Another thing to investigate through ray racing is whether it is possible to do a Ross null setup using one of my two null lenses. If possible, such a test should be accurate enough for this f ratio. Or would it require a negative null lens?
The problem is, R2 (when set up as reflective) with two passes through the glass lens becomes quite strong in power, with effective focal length of around 250mm. This is short enough to bend laser beam quite a bit before it exits back.
We'll have to sit down and discuss the options as well as results. Perhaps when you come over to pick up the silicone lap mold and some Acculap?