mirror making
The paraboloidal mirror of a telescope is a high precision optical component, whose surface may deviate only a fraction of 1/10000 mm from its ideal shape.
Surprisingly it is possible to make such a mirror using only the hands and a few simple tools. The quality within reach is at least as good as industrial manufactered mirrors are, many times it is much better.
6 inch/f8
The mirror- kit consisting of a BK7 blank, abrasives and polishing components I bought from VdS material service. The grinding- tool of 6 inch diameter was made from 10mm plate glass which was glued to a 20mm thick iron disk.
Grinding went on quickly and without problems, and working with glass was great fun. During grinding, I found it fascinating to watch a simple piece of glass turn into a telescope mirror with high precission shape and polished surface.
As I know now, I spent far more time in grinding each load of carbo than necessary. It is sufficent to grind until the noise changes clearly from "grinding" to "rushing", which takes at most about 2 - 3 minutes.
Polishing and parabolising required a lot of patience, one reason for that was the minor hardness of the BK7 material compared to Pyrex, Duran or Zerodur. The softer the glass, the more difficult it becomes to polish the correct shape.
10inch/f5
The blank is made from Duran, 42mm thick and had already a f/8 curve when I bought it.
The tool is got 240mm diameter and is made from two plate-glass discs wich are glued to a disc made from three layers of plywood. The one in the middle is got a diameter 1 inch smaller than the outer ones. The result is a groove all around the circumference in which one can grab the tool safely while lifting it from or laying it on the mirror.
For inital grinding of the convex curve I fastened the tool like a pendulum hanging down from the balcony handrail above an electric tile cutting machine. That saved a lot of time during rough grinding.
As I planned to use the mirror together with the one from my 10 inch /f5 GSO in a binocular telescope, it was essential to reach the focal length of the existing mirror as close as possible.
By the beginning of rough grinding, the sagitta was 2,08mm. After 4 hours rough grinding it changed to 3,02mm. The sagitta calculated for the finished mirror was 3,26mm.
The depth of the sagitta was measured in the way shown below.
This method is quite exact if one wants to check the sagitta of the mirror for the determination of focal length because the distance between the points it rests on is maximum, but it fails if you want to check the spherical shape of the surface. If you turn the bar by any angle at all, the results measured may differ one from each other. When I watched this phenomenon the first time after I had finished fine grinding with grid size 400, I was quite shocked. The only explanation seemed to be astigmatism of the mirror.
On the other hand the movements of the tool were absolutely smooth and without any tendence of stopping anywhere.
Another check of the spherical shape was done with the instrument shown above. It rests standing on three points on the surface to check, and there were no differences measureable. The difference checked out with the bar may result from the fact, that the contact points are never exactly the same and the rim of the mirror might have still some irregularities, for example remains from chamfer grinding.
After finishing rough grinding work went on as follows:
K80: 3h MOT/TOT; short to medium central strokes; sagitta 3,22mm
K120: 2,5h MOT/TOT; short to medium central strokes; sagitta 3,22mm
K220: 2h TOT; short central strokes; sagitta 3,25mm
K320: 2h MOT/TOT interchanging, short central strokes; sagitta 3,26mm
Grid size 400,600,800,1200 each 2h TOT short, central strokes, no change of sagitta any more.
After finishing fine grinding one major step on the way to a bino- fit mirror was made: the sagitta measured was exactly the one that was calculated.
Polishing
The 240mm tool was covered with pitch. It was grooved laying inside a bucket under water using a saw blade. Doing this under water prevents the flat from getting polluted by pitch particles cut out.
After polishing for 3 hours with long, central strokes the rim was still grey.
So I decided to return to fine grinding grid size 800 for two hours with very short, central strokes. Then a new pitch lap was made and pressed. After one hour of polishing the surface looked like that:
The edge was clearly higher then, a good assumption for getting the whole surface polished regular.
After 12 hours of polishing the sphere was nearly reached.
Parabolising
Parabolising was started with the 240mm pitch lap doing W-strokes. The first, slight shadow of the paraboloidal curve appeared after one hour of careful parabolising.
Another 5 hours of parabolising left the mirror like shown below:
The paraboloidal shape seems to develop in the right way although the center is still not deep enough. The ronchi- grating is got 6 lines/mm and the light source is on the left.
The foucault- test device with webcam ist very usefull now.
As the center is still not deep enough and parabolising with the 240mm tool needs clearly more time to bring changes to the curve than it did in the beginning, I continue working with a 150mm pitch lap. This led to success and the first useable results by measuring longitudinal aberration with the couder mask. On 05-21-2002 I measured a wavefront error of lambda/6.
At Kurt Schreckling´s site wie checked the mirror in autocollimation with Kurt´s superb 12 inch scope called "quarzmonster".
Kurt judges my mirror to be very slightly undercorrected, there is still that little hill in the center.
I decided to get rid of this raised center with a 120mm diameter lap. After 10 minutes parabolising the analasys of the measurments with figure45 promisses lambda/11.
Very carefully the small lap is used again for only 4 minutes.
The analysis with figure45 says lambda/18, Martin Trittelvitz´ program "parabel" says lambda/17. The surface looks very smooth.
Finally I compared the focal lengths of the two mirrors by means of the foucaulttest. The result was better than I ever dared to hope, both mirrors have a focal length of 1252mm.
Analysis of the finished mirror with figure45. Even if those results are only approximately correct, I am realy satisfied about them.
Conclusion after making the 10 inch- mirror:
- Duran can be easily grinded and polished, much better than BK7 can be.
- The 10 inch- mirror was more comfortable to handle and in spite of its faster f/D ratio easier to finish than the 6 inch mirror was.
- The "last lambda" can be polished much more sensitiv with a small lap (approximately half the diameter of the mirror)
Home
For inital grinding of the convex curve I fastened the tool like a pendulum hanging down from the balcony handrail above an electric tile cutting machine. That saved a lot of time during rough grinding. 
This method is quite exact if one wants to check the sagitta of the mirror for the determination of focal length because the distance between the points it rests on is maximum, but it fails if you want to check the spherical shape of the surface. If you turn the bar by any angle at all, the results measured may differ one from each other. When I watched this phenomenon the first time after I had finished fine grinding with grid size 400, I was quite shocked. The only explanation seemed to be astigmatism of the mirror.
On the other hand the movements of the tool were absolutely smooth and without any tendence of stopping anywhere.
Another check of the spherical shape was done with the instrument shown above. It rests standing on three points on the surface to check, and there were no differences measureable. The difference checked out with the bar may result from the fact, that the contact points are never exactly the same and the rim of the mirror might have still some irregularities, for example remains from chamfer grinding.
After finishing rough grinding work went on as follows:
So I decided to return to fine grinding grid size 800 for two hours with very short, central strokes. Then a new pitch lap was made and pressed. After one hour of polishing the surface looked like that:
The edge was clearly higher then, a good assumption for getting the whole surface polished regular.
After 12 hours of polishing the sphere was nearly reached.
Parabolising
Parabolising was started with the 240mm pitch lap doing W-strokes. The first, slight shadow of the paraboloidal curve appeared after one hour of careful parabolising.
Another 5 hours of parabolising left the mirror like shown below:
The paraboloidal shape seems to develop in the right way although the center is still not deep enough. The ronchi- grating is got 6 lines/mm and the light source is on the left.
Finally I compared the focal lengths of the two mirrors by means of the foucaulttest. The result was better than I ever dared to hope, both mirrors have a focal length of 1252mm.