Re: ATM thermal expansion coefficient of plate glass

[3D@taconic.net]

On 4 Jun 2002 at 10:17, jroweapp@postoffice.swbell.net wrote:

> regardless how flat you make plate glass, you will
> still have to support it in some fashion. Your flat plate glass is going to
> bend if its just supported by its edges. How many waves depends on its
> thickness and size. Please look at the efforts that go into making mirror
> cells for glass that is probably way thicker than the kind I imagine you are
> proposing. Consider your whole system.

My calculation purports to show that localized errors on the order of 5 waves/inch will 
cause sub Angstrom blurs in the acquired spectra. I don't know how badly the 
support will bow. Can you quantize it? There are many options for support structures 
that can be analyzed, but theoretical limits might be in the same ballpark as plate 
glass flatness. If so, the calculation is extensible. Have you looked at it?

http://www.drillamerica.com/PDF/plate_calc.pdf

I can put the Mathcad 6 version on line if anyone requests.

I have noted with interest how large mirrors are segmented, and this approach 
applies to my proposal. Since the grating is the same pitch througout, it can be made 
into component modules that rest on individual "tables" each with micro positioning 
adjusters to bring the whole set of gratings into alignment. Remember, once an 
observation begins, the thing is locked down. This gives the design an advantage 
over segmented mirrors which are in a carriage that must move to track the stars as 
they transit. In my design, the earth itself is the only moving part.

Tom