When I made the six point cell for my 16 inch mirror (pyrex, 1.5 inches thick), I was assured by analysis from Richard Schwartz that six points were adequate. I followed his advice, but just to be cautious I made the RTV pads rather large...over 1 1/2 inches in diameter. In the four years I've used this mirror, star tested it, and had others star test it...so far nobody has seen signs that the mirror cell was degrading image quality.

Lately I've learned how to use the software from David Lewis called PLOP (Plate Optimzier), and I've been able to answer a question that's been bothering me for years: was the good mirror support in my cell due in large part to the use of wide RTV pads, or to the fact that six points of support was truly adequate for my mirror? It appears the wide pads I used were a help, but not by very much. (In other words, Richard's advice was sound.)

Using PLOP to analyze my mirror, I get the following mirror deformations:

1.6 inch pads - RMS 3.26 e-6 millimeters.

0.8 inch pads - RMS 4.22 e-6 millimeters.

'point' pads - RMS 4.61 e-6 millimeters.


Here are graphic images from the analyses:

1.6 inch pads:


0.8 inch pads:


'point' pads:


Remember, most folks that RTV glue their mirrors to their cell use pads about 3/4 inch in diameter. This means that in my particular case, wide pads are only about 25% better in mirror support than typical pads.


In case you were wondering, here's a depiction of the mesh generated by PLOP to simulate my mirror in finite element analysis. I used many rings in the mesh to ensure that the differences between point, normal, and wide supports were well modeled.


My tentative conclusion is that wide pads in mirror cells are helpful, but only in a small way. Do not expect them to overcome shortcomings in your mirror cell design or implementation.


Instead of making wide support pads, you will get considerably better mirror support (or allow acceptable support of a larger mirror) if you use more support points. Here's an example with 12 points in one ring. (12 point ring - RMS 3.90 e-6 millimeters.) Note that support performance is similar to a six point support using wide or normal pads. (Hey, this is not really better support...didn't I fail to make my point here? Yes, but please read on...)


If you are going to use 12 points, you get better support performance not from one ring, but two. (ring of 4, ring of 8) - RMS 1.47 e-6 millimeters.


Here is the same ring of 4 and ring of 8, but with a change to the rotation angle of one of the rings. In this case support performance is essentially identical.


If you are curious, you may ask 'how does a 12 point support with rings of 3 and 9 compare to a 4/8 cell?' With my mirror dimensions the support performance is essentially identical. (You results may vary, depending on the diameter and thickness of your mirror. But with PLOP it's easy to take the guesswork out of this sort of analysis.)


A comment about PLOP's use (or non-use) of refocusing while analyzing/optimzing mirror cell designs. Cell performance and support point location can chage a great deal with use/non-use of refocusing, depending on how many rings of support in the cell design. With only one ring of supports, use/non-use of refocusing has the greatest change on performance and point location. With three and four rings of supports, refocusing has very little change in support performance, and only small changes in location of support points. I am not advocating one way or the other, but I want you to be aware of the changes it can make in your design analysis, and I recommend you be consistent in your use (or non-use) of refocusing as you investigate various cell designs with PLOP.


For a great example of other support schemes, with 18 or more points, see: http://www.cs.berkeley.edu/~jonah/18plus/

and for analysis of PLOP and good/bad ways to use the software, see: http://www.cs.berkeley.edu/~jonah/18plus/p18.html


Another good reference on mirror cell designs is from Mark Holm at: http://users.telerama.com/~mdholm/atm/cells/index.html


All feedback is encouraged!

email: t-k-r-a-j-c-i-@-s-a-n-.-o-s-d-.-m-i-l (remove the dashes)

Last update: 29 Mar 2002