Again I am in awe of your ATM skills....the nice thing about Zerodur is: you don't get polishing/figuring "blisters" due thermal expansion of the substrate (or so I am told)
I expect the end result will be awesome
Thanks Peter,
Yes, some people may not appreciate that the heat generated during the polishing/figuring process adds to the unpredictability of the result.
I also expect to end up with fewer scratches on the finished surface because of the harder material. My office, where the polishing machine is located, is not exactly a NASA clean room. I even stop polishing on windy days.
I did the casting for the grinding tool substrate out of Ultracal 30. The top is flat - only the lens distortion is making it look curved.
It will be a tile tool.
The casting initially weighed 6.3kg and went down to 5kg as it dried out.
The tiles came from Bunnings.
I glued them on with the shiny face toward the casting, so that the required epoxy layer would be thinner.
I used an improvised angle grinder with a diamond tool to trim the tiles and then used the ring tool to grind off the high spots before I starting on the mirror blank.
I started with 120 carbo and after six wets the two surfaces were well matched and most of the ridges on the tiles were gone. From this stage I started using the spherometer and found that I needed to deepen the curve by about 0.06mm. I made the correction with the help of the ring tool on the blank and by having the tile tool spend more time at the bottom over the next few wets.
The grinding is done. I finished it with 9 micron aluminium oxide.
All hand grinding as there is nothing to gain from machine grinding and I prefer to keep abrasives away from the polishing machine.
I had to stop work on this job for a couple of weeks, but I'm back at it now.
Usually I polish my mirrors with a full diameter lap but my current polishing machine would struggle with a 12" lap, so I'm doing the polishing with an 8".
I'm at 3 hours now with minimum loading and when I get to about 5 hours, I will set up the Ronchi tester to measure the radius of curvature and check the figure.
I had to stop work on this job for a couple of weeks, but I'm back at it now.
Usually I polish my mirrors with a full diameter lap but my current polishing machine would struggle with a 12" lap, so I'm doing the polishing with an 8".
I'm at 3 hours now with minimum loading and when I get to about 5 hours, I will set up the Ronchi tester to measure the radius of curvature and check the figure.
Good day Stefan, how does a machine compare to manual strokes? Do both discs rotate automatically?
The machine I'm using has a single motor driving the platform and the two eccentric mechanism that push and pull the top disc, and because of that, it can't generate real random motion like a human. Three motors with individually adjustable speed controllers would be much better.
Only the bottom disc is driven. The top one rotates automatically.
The machine I'm using has a single motor driving the platform and the two eccentric mechanism that push and pull the top disc, and because of that, it can't generate real random motion like a human. Three motors with individually adjustable speed controllers would be much better.
Only the bottom disc is driven. The top one rotates automatically.
Do you do final figuring by hand then? Or do you use the machine to simulate shorter or longer strokes with the tool offset?
I got to 19 hours with the polishing and I can still see a hint of grey near the the edge, so I'll give it a few more hours because once I start the figuring, the edge will get very little further polishing.
On the other hand, the figure is quite nice and spherical, with a bit of raised edge. I tried to capture a ronchigram, outside of focus, with a handheld camera.
Also I improvised a pitch hardness tester today. Seems to work ok.
I had to stop polishing for about a week because of the low temperatures we were having.
I'm at 22 hours now and started deepening the centre.
Soon I'll have to switch from the Ronchi tester to a Ross Null setup.
I attached a Ross Null lens to my Ronchi tester and carried on the parabolization with the 8" lap until the outer 2 inches or so looked flat. After that I started using a 4" figuring lap so that I don't over deepen the 70% zone, which would be a pain to correct.
After a few hours the lines were straight enough to make test results difficult to interpret. See attached Ronchigram.
So it came time for setting up the Bath interferometer. The first noisy interferogram is attached and the processing shows that my conic constant at the moment is -0.74, and the figure exactly where I expected it to be.
After a few more iterations I should be approaching the diffraction limit.
I made two test lenses one 60mm diameter and one 80mm.
In my experience the Ross Null test is only good enough to get you to about half a wave to your target before you lose sensitivity. Also it is very good at cancelling astigmatism, so you can't test for that.
Well, it took me 7 iterations with the interferometer, each comprising about 10 minutes of polishing, before I reached the diffraction limit at about 0.84 Strehl.
Another half a dozen shorter iterations got me to 0.958 Strehl ratio.
The attached graph shows only the spherical component, without other residuals. If there were no other residuals, the Strehl ratio would be 0.994, which is not a real world number unfortunately.
I'll work a bit more on the residuals and see if I can improve if further.
The problem is that a full evaluation of the surface takes about half a day's hard work because many interferograms need to be captured and processed in order to cancel out air currents, vibrations and instrumental effects.
I'll work a bit more on the residuals and see if I can improve if further.
The problem is that a full evaluation of the surface takes about half a day's hard work because many interferograms need to be captured and processed in order to cancel out air currents, vibrations and instrumental effects.
If I read the results correctly that would be something like Lambda/25 wave-front error even counting up to the extreme edges and less than Lambda/50 for something like 90% of the total area. That's a wonderful and amazing result.