RE: ATM Dial indicator on mirror

[andydtg@phoenix.net]

Bob Goff wrote:

>I only do this when making Schmidt plates or other fast aspherics and then
>only in the rough stages in the transition between grinding and polishing.
>One needs to develop a sag table ( a table of values, mathematical ).then
>one needs a reliably dial indicator that has no side play ( in the case of
>the 1.5" dia f/0.35 parabola ) or some other means of probing. Then one
>needs to be able to accurately translate ( move ) the probe in " r ".
>
>That scenario is so difficult that, when I rescently made 7 each 6" F/1.5
>parabolas, I decided to just use the good old rose lap and an interferometer
>to test.

A few years ago I was tooling up to grind the 19" dia correcting plate
for my 16" f3.6 Baker corrector; I made up a dial gage jig which
had 2 parallel hardened steel rails, lapped flat to about 1 wave
on top and 3 ball bearings (2 at same end of each rail, one midway
between at other end. A carriage for a .00005" dial gage slid
on 3 balls on the top of the rails with the plunger in between.

The bottom contact balls were differentially lapped so that when the
jig was placed on an optical flat, the gage would stay at the same reading
throughout its travel of about 6". The idea was to place this
jig on the cp with the ends supported on the inside on glass that
would not be ground (6" hole for secondary later), and on the
edge which also would not be ground. Then the cp could be rough
ground on a turntable using a 1-2" dia tool and the profile measured
with the jig to about .0001" accuracy or better, out of a total
max profile depth of about .0015". Fine grinding and polishing &
figuring would not use the jig.


I never got to the cp grinding stage of this project and any now
a bit worried about whether this would work at all - any thoughts?
I have about a year before I resume this project.

Another random thought - the focault test and its variants basically
measure the slope of the surface of the mirror, the individual slope
measurements are combined to deduce the shape of the surface. The
sensitivity of the test is driven by the size of the zone being
measured and the radius of curvature distance - the longer the more sensitive
the test.

Would it not be possible, in principle, to place a small (<1")flat mirror on
the surface of the mirror and measure its 'tilt' at a much longer distance
thus increasing the sensitivity of the test? One would move the test flat
radially on the surface on a 3 point mount & measure its tilt, say, a
100' away. Each type of surface (sphere, parabola, whatever) would
exhibit a predicted off-axis tilt axis intersection at the measurement 
distance that is compared with the actual tilt data (obtained by auto-
collimation on the sliding flat). The principle is the similar to the
caustic test except that this method does not reflect the test beam from
the actual mirror surface. I haven't done the arithmetic yet but would
like some opinions on this idea.




Andy Saulietis / DTG Alt-Az-Fp Drive Systems
12617 Harriet Ln
Santa Fe, TX 77510 USA
409-925-8854 Voice/Fax