[ATM] Thoughts about mirror cell design from a stubborn RTV-head

[jonah at eecs.berkeley.edu (Jeff Anderson-Lee)]

Bruce Sayre wrote:

>Jeff Anderson-Lee wrote on 2/18/04:
>
><<|>2) the three bottom whiffletree points are not also used for
>collimation
><<|>
><<|Whether or not you use the 3 bottom points for collimation is
>immaterial 
><<|to the support...
>
>In principle you're right, but as a practical matter, I think it's
>difficult to eliminate lateral slop if the three bottom points are
>integrated with the three collimation bolts. Since the bolts drop
>through holes in the tube bottom, each hole must be larger in diameter
>than the bolt's threads, so there's no guarantee each bolt will rest
>with equal weight in each hole. Especially if the threads wear away.
>Especially if the holes and bolts aren't exactly and precisely located
>and machined with the same center-to-center distances. See those forces
>traveling right up the whiffletree to the mirror? 
>  
>
I suppose it's possible to have vastly differing forces on each 
collimation point, but I'd be surprised.  I'll bow to your experience on 
this point since I think spreading the collimation points is a good idea 
anywaya and simplifies the wiffle tree in some ways.

><<|>4) the whiffletree rests on a rigid, flex-free metal structure
><<|>
><<|See (2) above.  "Rigid" is relative.  Since there are (only) three 
><<|support points, the issues are (a) refocusing and (b) collimation.  
>
>The concern I have here is similar to the above. Typically, two square
>metal tubes span the width of the mirror box and hold the collimation
>bolts; one holds two and the other holds one. Will they always be
>separated exactly by the same amount, so the bolts don't get squeezed?
>See those bad forces traveling right up the whiffletree to attack the
>mirror? 
>  
>
Again I'd be surprised, but again I defer to experience.  Once again, I 
think a back support makes sense, but I don't know that it's manditory, 
but better to error on the side of caution.

><<|>5) the whiffletree's flexible pivots are incapable of lateral
>movement
><<|>
><<|Can this be accomplished without added complexity or stiction.  If
>all 
><<|support points are RTV'd to the mirror this is a moot issue anyways.
>Or
><<|do you mean excessive "slop" or "play" rather than "rotation"?
>
>I wasn't thinking about the small triangles, or "load spreaders" as
>they're called by the engineers, where the mirror rests on the RTV pads.
>It's the mechanical pivots down a level or two that are subject to
>lateral slop.  An 18-point cell has nine of these; a 27-point cell has
>twelve. If these move laterally relative to each other because of play,
>watch out. See those bad forces ... (Hence, Teflon-lined spherical
>bearings.)
>  
>
Where do you get those nifty teflon bearings anyways? ;-)  For smaller 
mirrors I'd just go with "compliant-hinges" made from RTV anyways to 
skip the mechanical headaches.  That probably won't work for a 24x2inch 
mirror though, or at least not all the way down the wiffle tree.

><<|>7) the silicone pads are 1.5mm to 2mm thick
><<|>
><<|Sources suggest 2 to 3 US quarters for RTV pad thickness (or roughly 
><<|1/8in to 3/16in or 3mm to 4mm).
>
>What reasons do the sources give? 
>  
>
Mostly anecdotal evidence from the ATM archives.  (i.e. It worked for me.)

On Nov 7, 2001, Michael Lindner wrote (Re: ATM mounting primary 
mirror):  "...I queried several people who had had good and bad 
experiences with silicone (thanks, guys), and then ran some tests 
varying the size and thickness of the silicone. What seems to cause 
problems with collimation/astigmatism are pads that are too thin or too 
thick, or (to a lesser extent) pads that are too small or too large. I 
settled on pads 1/8" thick and 1" square. The scope holds collimation 
well, with no detectable movement of a laser collimator spot from 
horizon to zenith (which is better than what I experienced with the 
sling)."

On Tue, 31 Mar 1998, "Tom Cathey" <K1JJ@msn.com <mailto:K1JJ@msn.com>> 
wrote: "I have a 22", 2" thick Pyrex mirror in service here that is 
RTVed to a 27 point cell. .. (Each pad of RTV is about the size of a 
penny, and 1/8" high)."

On Tue, 31 Mar 1998, Joe Castoro <astrojoe@erols.com 
<mailto:astrojoe@erols.com>> wrote: "Use a temporary spacer between the 
pad triangle and the mirror so that the GE silcon II does not squish 
down all the way when curing ( about 48 hours ). I use a 3/16 spacer and 
a dab generous enough to equal 1" pads."

On Thu, 29 May 2003,  "Scott Berfield" <sberfield@xbox.com 
<mailto:sberfield@xbox.com>> wrote: For my 8" I used RTV pads. I made a 
small divot in the substrate, stuck a small (2mm) bead in the divot with 
a dab of glue, then applied enough RTV to squish out to around 3/4" in 
diameter."

So perhaps 2mm is enough...
 

>Thanks for the comments, Jeff. (One of these days we'll meet. We're
>probably within 75 miles of each other.)
>  
>
Do you make any of the SVAS star parties?  I'm an occasional attendee.

Jeff Anderson-Lee
Sacramento,CA