Astronomy and Sky Website of Martin Lewis

Updated 13-7-2020

Mirror 18-Point Support System

When I designed and constructed my 18-inch scope back in the early 90’s there were none of the current analysis tools available to amateur astronomers to help with optimising mirror support systems. Instead I had to rely on designs which were many decades old. I plumped for an 18-pt support arrangement for my mirror from an arrangement derived by Hindle where his original design was modified to make the support triangles as equilateral triangles (see note on page 326 Bk. 2 of Amateur Telescope making by AG Ingalls (1996)).

The arrangement I use for my 460mm, f4.4 mirror, which is 41mm thick, is shown below with three support bars which pivot at their centres and six equilateral triangles which each pivot at the ends of the bars. The mirror sits on 18 hard plastic support pads each 15mm diameter which sit at the corners of each triangle. A 3mm stainless cable edge-sling holds the mirror centrally on the support system and takes the weight when the telescope is tilted away from the vertical.

Dimensions of mirror support for 460mm mirror based on a modified Hindle design which uses equilateral rather than isosceles triangles to simplify the design.

I was curious to apply modern analysis methods to the arrangement I ended up using and to see how it compared to the optimally calculated arrangement. If my design was significantly worse than the optimum then it might be worth thinking about remaking the some or all of the 18-pt support system.

Modern telescope makers can use a CAD design program from David Lewis of Toronto University called PLOP (PLateOPtimiser) which builds on an FEM program called Plate by Toshimi Taki. This will output calculated surface errors in contour map form and also optimise designs and output the dimensions for the parts to make an optimised 18pt support system.

PLOP can also output data to then feed into a 3-D analysis package called Z88 which is built into version 3.0.5 of the program. Z88 further improves the accuracy of the modelling and is also able to output surface error contour maps. You can find out more about PLOP and download the program from here.

As well as being able to calculate errors in the surface of the mirror PLOP can help further improve mirror support designs by taking account of the following;

• Allowing a refocus after the mirror deforms on its support system gives more freedom in the design and results in a lower surface error value.

• The portion of the primary which lies under the shadow of the secondary does not contribute to the final image. Allowing the mirror to deform here and ignoring this deformation again allows the design of systems with a lower error in the rest of the mirror.

PLOP and Z88 were used to calculate the peak to valley and RMS surface errors for my mirror with the current arrangement and also to generate an optimised 18-pt arrangement; refining the design iteratively to minimise the surface errors. In both cases calculations were carried out for the scope pointing vertically upwards so there was no weight on the sling- it was all resting on the 18-pt support system.

I have written a webpage on this site showing you how to use PLOP to generate this data. 

Taking the Z88 values as the more accurate, as they are 3-D calculations, I could more than halve the surface errors arising from the support system by modifying the support components to match the optimised arrangement. However, this would require considerable work and this effort needs to be put into context.

The Cruxis Dobsonian Mirror Edge Support Calculator page says of surface errors’ The computed RMS Surface Error is given in nm (nanometer). The traditional “diffraction-limited” mirror corresponds to an RMS surface error of 20 nm (1/27 wave). The mirror cell should generate errors that are significantly below this value; a value of 5 nm or below corresponds to a virtually perfect edge support.’ Given that my RMS errors are about 1.5nm and all that effort would only improve matters my a minuscule 0.8nm RMS then the effort is not justified. I also can do further Plop calculations showing that my non-optimally supported mirror has the same surface error due to the support system as an optimally supported 24″ f4.4 mirror 50mm thick.

Mirror Edge Support

The very useful Cruxis Dobsonian Mirror Edge Support Calculator page mentioned earlier allows one to calculate the astigmatic surface errors arising from different edge support designs when the scope is pointed towards away from vertical.

After the modifications in late 2014 my scope now follows the best simple arrangement which is a 180° sling (actually it is about 170°) supporting the bottom half of the mirror and placed on the plane of the centre of gravity of the 41mm thick mirror, 18.9mm up from the base. The Cruxis page calculates an astigmatic error due to the support of 0.9nm with the scope pointing horizontally, dropping to half this with the scope at 30° from vertical and 0.7x at 45°.

The page also shows that when the sling is not acting on the line of the centre of gravity then the error increases massively being almost 4x larger if the sling is misplaced by just 2.5mm. In addition to these errors, Nils Olof Carlin showed that further errors can be introduced if the sling is at an angle to the plane of the mirror as it can locally lift or pull down on the mirror, upsetting the weight distribution on the 18-pt support arrangement. It was because of these errors, which can quickly approach the 5nm maximum RMS error budget, that I moved away from the original crude 1cm wide glass fibre reinforced packing strap which was taped approximately at the centreline of the mirror. I’m sure this older arrangement was bound to have introduced some astigmatism in the mirror in the past.

After the recent modification I now use a 3mm, 7×19 strand stainless cable positioned at the centre of gravity line by the fact that it is sitting on foam strip stuck on the edge of the mirror (see my star testing page for pictures). To address Nils’ off-angle concerns, the vee-block clamps which holding the ends of the cable are also carefully positioned so that the cable ends are held in the right in-line position when the mirror is collimated. Howie Glatter came up with an ingenious method of making this always the case, even if the mirror moves up or down, by having the cable ends attached to linear bearings which freely run up and down on steel posts. In my scope, making the bottom collimation bolt fixed prevents the mirror slowly migrating up or down the tube as a result of repeated collimations and so the cable should stay pretty much parallel to the plane of the centre of gravity of the mirror.

Measuring the sling attachment point height with respect to mirror plane by use of a glass disc sitting on the 18-pt support arrangement (which had previously been set at the collimated position with the primary in place).
Close up of sling attachment clamp and mirror side position adjuster cam and top stop. The mirror is in place and has an 8mm edge mask shown to mask off a narrow turned edge.

It is possible to use Z88 (via PLOP) to calculate the expected surface error from different edge support arrangements at different mirror tilts. PLOP does have entry boxes under the Edge Support and Tilt tab for angle of the telescope and the included angle of the sling but PLOP itself will only calculate deformations with the mirror horizontal. If you use the output of PLOP to feed into Z88 and run that then it gives a Z88 surface error contour plot for the tilted mirror. There are, however, issues with this method which mean that the calculated error values are too high. You can read more about this method at the bottom of my Use of PLOP page. Below is an example of the surface error contour plot for a tilted mirror calculated by Z88 and described on that webpage.