In the original cell, the mirror, 250 mm (10") and only 25 mm (1") thick, was placed on six cork pads that were glued to an aluminum frame. There were 3 nylon screws on the side, and three retainers lined with cork. For a mirror with this dimensions classic rules require a 9 point floating support. The nylon screws on the side were OK, but the retainers where so tight, that the cork was squashed to a third of its original thickness.
I decided to modify the original cell, and make a classical 9 point floating support.
The original cell consists of two parts, the mirror holder and the part that is screwed to the tube (called "plate" hereafter). These two are connected by three collimation screws. There are springs on the collimation screws, and three separate bolts to fix the position of the mirror holder after collimation.
To center the mirror in the tube, I had to drill three new holes in the plate. There was enough space available to do so.
Since the three retainers have a limited range, it was important not to increase the distance between mirror and mirror holder. The triangles for the floating support thus had to go in between the plate and the holder. By a lucky coincidence (?) the three collimation screws are in exactly the right position for the center of the triangles. So these screws now hold the triangles, but still serve as collimation screws (Note: in the original design, these screws were loose on the holder side, for this application they were locked into place by a nut below the holder. Also I replaced the original screws by slightly longer ones). The springs were put on the fixation bolts.
As can be seen above, the mirror support points are bolts in the corners of the triangles. The height of the bolts is such that the mirror is about 1.5 to 2 mm above the holder. The retainers thus had enough travel to be reused. A piece of cork is glued on the top of the bolts, so the mirror does not touch the metal of the bolts. This picture also shows the three springs that are attached to a ring. This construction keeps the triangles properly orientated. The center holes are just visible.
The above picture shows how the triangles are put on the collimation bolts (that are fixed to the mirror holder. The mirror holder is show from below. The plate is missing in this picture. The tree springs in the middle are now slightly stretched. The shadows of the collimation bolts show that there is enough place left for connection to the plate.
Here you see the mirror holder and the triangles as seen from above. The three cork pads on the mirror holder are leftovers from the original support points, They don't touch the mirror. They were left in place to have some extra protection against the mirror bumping on the metal of the holder. On careful examination it can be seen that for each of the three inner support points I had make a slight "dent" (with a file) in the inner circle of the holder.
These last two pictures show the assembled cell, from the top (left) and bottom (right). The bottom view also shows the old mounting holes for the collimation screws.
After inserting the mirror, the nylon screws on the side were adjusted so that they just did not touch the mirror. This can easily be checked by inserting a thin piece of paper between the mirror and the nylon screws. The retainers were fastened in the same fashion. There is thus a minute amount of sideways movement possible for the mirror (about 0.05 mm; the thickness of the used paper sheet).
This page just describes the general idea, if you like more information please feel free to mail me.