Re: ATM thermal expansion coefficient of plate glass

[3D@taconic.net]

Bob May writes:

> You'll find that the expansion coefficient runs between 4 and 16 x 10^-6
> /
> deg. C.  Common optical glasses tend to run around 7 and is probably
> about
> the figure that I'd use for plate glass.

Is that 4-16 x 10^-6 meters, i.e. 4-16 microns expansion per meter 
per degree C? 


> I might note that while plate glass may be used in gratings, it's not
> done
> on the as produced surfaces that normally come from plate glass but
> rather
> the surface is smoothed up to much better tolerances.

That's not clear from Optometrics grating catalog. Perhaps they do 
lavish the extra attention on plate glass before putting it to use, but 
that would seem like a selling point that they would mention in their 
literature. They certainly do use better than float glass for masters. 
Plate glass is used strictly for replicas. However, if that is the case, 
the reasons may have more to do with their techniques of realizing 
the master rather than concern about the  performance of the 
grating once it is made. What would be the point of making a good 
master and lousy replicas?

I made a pilgrimage to Optometrics years ago to see their facility 
near Boston and was treated to watching them scratch out gratings 
mechanically. The grating was cut into a shiny soft metal that had 
been evaporated onto a glass substrate. They do not cut into the 
glass itself. Most of their gratings are much smaller than your typical 
telescope mirror.

Their small sizes do not address the questions about very large 
gratings, as I propose.

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

However, I think when it comes to flatness, my calculation

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

is extensible to any length once a value is set for localized surface 
irregularities. Once I get the thermal expansion ratio straight 
(microns over meters?) I'll repeat the analysis. The expansion 
parameter would affect the wave front phase change along the 
receiving angle quite significantly, because my proposed device 
uses grazing incidence for the receiver. If there is a phase delay in 
grazing incidence mode on a very long grating, it could have a 
profound effect and prove my concept faulty. 

If there is something wrong with my prior calculation for flatness, 
please stop me before I kill again.

Tom Ditto
with apologies to Ken Hunter for my continued babbling "off topic" 
and sincere appreciation for Bob May's thoughtful assistance.