PDA

View Full Version here: : Mirror Cells

alistairsam
23-03-2011, 03:56 PM
Hi,

This should possibly be in the beginners section, but could someone explain the science behind why exactly a mirror cell is required as opposed to say a flat rigid plate supporting the whole mirror?

I've just been reading on mirror cell deformations here http://www.atmpage.org/contrib/Holm/Plop_optimized_cells/compare.html and was wondering if it is worth building say a 3 point mirror cell for an 8" mirror. this is a standard bintel scope with the edge supported mirror. would it make any difference at all?

what mirror size would a mirror cell start making a difference?

if it is to allow cooling, would a refrigerated cooled plate do the job?
which is the best area to place temperature sensors to measure mirror temperature? bottom, sides, upper corner?

dannat
23-03-2011, 04:49 PM
Allistair, I have seen in pics a single truss dob, whereby there was no normal mirror cell, but a bolt glued to the middle of the mirror, next a piece of ply covering the mirror back then a large flat piece of metal screwed o the bolt to keep mirror in place. There were adjusting screws toward the outer edge of the ply.

MrB
23-03-2011, 04:56 PM
I believe the reason a flat plate is not used is due to the different expansion rates of the two materials (glass and wood/metal etc) causing all sorts of unwanted flexure and ruining the figure of the mirror.
The cells do introduce a little flexure of their own, however it is minimal compared.

rmcconachy
23-03-2011, 06:17 PM
Three points define a plane in space so if you pretend for a second that the mirror doesn't sag at all then you can determine where the mirror sits and faces with just three supports. Any extra supports make it likely that our non-sagging mirror will not be in proper contact with all of its support points and will rock just like a chair or table with one leg that is too short or long. A perfectly flat support plate doesn't exist (just like my imaginary non-sagging mirror), what would really happen is that the mirror would rest on a few higher points on the plate and you wouldn't know where these higher points were located. If we start with three support points then we get to choose where they touch the mirror rather than their position being left to chance.

An 8" mirror of `normal' thickness (6:1 in the old days but anything thicker than about an inch will do here) is very tolerant of how it is mounted. You mention the PLOP web page above, try a few different cells in PLOP with different numbers of support points in different places. Three points at just under 40% of the radius is probably a practical optimum for a typical 8" mirror but most anything will work well enough. If the cell induces a tiny fraction of a wave of additional error I'm unlikely to notice in general use. Primary cell design and implementation becomes progressively more important as the diameter of the mirror increases and its relative thickness decreases because this is when the mirror becomes more likely to bend and sag significantly. A 20" mirror of 12:1 diameter:thickness ratio would be horrible on a three point primary cell (it would sag too much away from the support points). Different people choose different tolerances but many agree with the author of PLOP that any cell likely to produce <1/40th wave error on the mirror surface (so <1/20th wave on the wave front) is very likely good enough because this amount of error will be swamped by other sources (seeing usually being the biggest).

Your questions regarding refrigerated cooling are better handled by those who have done it (not me) such as Bird and friends.

Happy observing!

alistairsam
24-03-2011, 01:35 PM
thanks. bit clearer now.

but just as there is no perfectly flat plate, the different points in a mirror cell can't also be guaranteed to form a perfectly flat plane, in addition, the bottom surface of the mirror would also be less than perfectly flat. so do mirror cells compensate for these differences by lowering or raising individual or group of points?

an imperfectly flat plate supporting a mirror, could for example cause the mirror to tilt from the horizontal say by 0.1 degrees, but we're then collimating the three support points on the plate to re-orient the mirror back to zero degrees so our collimating light beam returns correctly.
how different is this from a cell where we still adjust the 9 or 12 point cell at three collimating points to re-orient the mirror to zero degrees?

one major issue with the plate I imagine would be unpredictable surface deformation due to expansion or contraction. this would be minimal in a cell as contact surface area is minimal.
but then a temperature controlled plate could minimize these deflections? it might be more complex than a mirror cell, but woudl reduce mirror sag between cell support points a fair bit.

alistairsam
24-03-2011, 01:39 PM
Hi Daniel,

was this a conical mirror? I saw a thread here with almost exactly the same support you've described but the mirror was conical.

MrB
24-03-2011, 02:57 PM
Three points do indeed 'form'(describe?) a perfectly flat plane, just as two points 'form' a perfectly straight line between them.
But forget the flate plane idea, it is just confusing.

An object on one point is balancing, very unstable.
An object on two points is also balancing, still unstable.
An object on three points is at rest, stable.
Every solid object at rest on another solid object is in direct contact at three points only. Those three points may not necesarrily form an equilateral triangle, they may make a triangle of any shape, but an equilateral is the most stable.
Even if the objects appear to be in contact in more than three places, they are still only in contact in three places. The fourth(fifth, sixth etc) points may appear to be in contact but if you were able to inspect them, they would not be touching.

Thought that was my original reply :shrug: