Re: [ATM] Mirror cell ramblings re: thin plate glass3/4-inchx24-inch

[Alexis Cousein <al@sgi.com>]

Ted Cohen wrote:
> All -  
> 
> Thank you for that background information and guidance. The reason, I'm
> being somewhat obstinate, is that I've already put about one year into this
> mirror and all of the related setup, so I want to be fairly certain that
> what I was intending isn't feasible. At the risk of sounding obstinate, I
> would just like to clarify a few points that seem to address several of the
> important concerns raised. 
> 
> 1)  The glass/aluminum interface - Several have pointed out the inefficacy
> of mounting the glass on aluminum - I agree 100% - however, ....  In my
> earlier email describing the setup, I mentioned a thin membrane, which will
> come between the glass and aluminum. I am/was intending to create a 2mm
> layer of (some variant of  RTV) directly onto the aluminum cell.

I've had bad experiences with this - especially with a continuous layer,
the RTV cures differently on different spots and *does* pull on the
mirror; it's a fallacy to think that it simply conforms to the
unstrained mirror surface. Bubble wrap is a lot better than RTV for
conforming to a surface without stressing it (though I have no idea
whether it's good enough for your application).

I've seen problems on a lot on scopes on even very small mirrors, in
particular scopes where the secondary is mounted with a ring of RTV
on the outside - you tend to end up with horrible astigmatism.

For secondaries, three PLOP-like support points *do* work for these
secondaries. Make it four points and the secondary bends, because
four points aren't always in a plane. Make the RTV continuous and
you have an infinity of support points, some of which will bend
the secondary.

Conclusion: RTV is fine, but only using a discrete set of contact points
in a true floatation cell - which brings us back full circle.

 > I was intending to mount the aluminum cell using an hexagonal ring
> support at the 70% zone of the aluminum.

And you're only moving the problem: how do you ensure all points of
*that* ring are in a plane? That's exactly the problem wiffle-tree
floatation cells solve, BTW.

 > I do not believe
> that I want a discreet set of support points.

That's a belief not backed up by fact, in my opinion.


-- 
Alexis Cousein                                  al@sgi.com
Senior Systems Engineer/Solutions Architect     SGI/Silicon Graphics
--
<If I have seen further, it is by standing on reference manuals>

_______________________________________________
ATM mailing list http://www.atmlist.net/