ATM Flex possibilities

["Peter John Smith" <pjifl@bigpond.com.au>]

William Kelley wrote that


>     More recently Alan has produced a program,not yet published,that will
> calculate the appropriate back-shape, and will allow blanks to be cast, or
> CNC lathe generated, that will allow large, fast mirrors to be flexed to a
> paraboloid with a simple bolt glued to the center. These mirrors become
> progressively thicker from the edge to the center, where they rise
abruptly
> to a short, cylindrical stub which is smaller than the shadow of the
> secondary mirror. A  tensioning bolt is epoxied to the stub, and a spring
> maintains the proper tension. A glance at the figure in the Sky&Tel
article
> will make this clear. At this time the concept is not suitable for an
ATMer
> making a mirror of an ordinary flat disk, but proof of the concept should
> lead to commercially available pre-shape blanks. I have just completed the
> first of such mirrors by having Newport cast a bunch of blanks for me, And
> then painstakingly removing casting irregularities. Zonal tests on the 6"
F5
> look great. I'll know just how good when I get a coat of silver on it, and
> set it up for a double-pass null with a flat.


It seems to me that many are assuming ATM's cannot shape the rear to spec.

This is simply another challenge to be solved.

For example, it would be easy to generate an off centre 'sphere' on the back
that rose to a flat stub.  It would be sort of cusp shaped.

Has this been assessed as to flex performance ?

Nearly any odd curve could be impressed onto the cusp portion by using cam
or template control while generating.  Or a series of different R portions
generated as one moves outward.

Maybe giving specific details of an example of this shape would be
productive as it will allow people to apply their ingenuity to its
generation.

I must echo the thoughts of others that an interesting use is Cass
secondaries.  It would be a real test of the technology to flex a convex
secondary to say  K = - 4.  I cannot even imagine what optimum shape the
rear should be.  It would be small enough to experiment easily with
generating the rear shape but conversely, the amount of flex is probably
proportionately far more than for a primary.

Another possibility.  If the technology is accurate and predictable enough,
one could flex a concave testplate into a hyperboloid and use this to
figure a convex surface.  The right calculations may show if this is
feasable.

All very interesting.

Peter Smith.