My impression is that the whiffle tree has to float with the mirror supports,
If you want a no-fuss system. It's possible to put them fixed with respect to the mirror box, but then only one position on the bottom and left/right inclination of the primary in the mirror box is optimal. In practice, you collimate using the optimally placed primary playing with the secondary offset and/or focuser orientation to get the optical axis pointing straight up (which allows you to get the whiffle trees left and right close enough to the plane of the mirror's COG).
After this bothersome first collimation that gets the mirror on the right spot, from then on you collimate with one "fixed, do not touch this!" bolt on both the secondary and primary tilt screw sets to avoid straying too far from that optimum when doing axial collimation.
The good thing about a whiffle tree is that small placement errors with respect to the COG plane do not matter as much as with a sling with attachment fixed to the mirror box. With a sling, small errors in the height of the mirror with respect to the fixed attachments where the sling starts will introduce forces normal to the mirror's surface fairly rapidly (hence the need for a solution like that of the Howie Glatter sling, which uses linear bearings to auto-adjust these to the mirror's current position).
With a whiffle tree if two of the roller bearings are a bit lower or higher they do not directly introduce forces normal to the mirror surface (that's why they are roller bearings), just some minor variations in forces on the back support points. With a scope pointed very very very close to the horizon it may matter because if the roller bearing is too low the mirror will roll to the mirror clips on that side when you reach the horizon, but usually we're talking really close to 0° with the horizon in the FOV! And if you get in that situation it's usually acceptable to raise the mirror just a wee bit on all bolts to avoid it.
A couple of millimeters are usually not a problem except if you have a really thin mirror, e.g. a slumped meniscus.
It does mean that adjusting the focal plane placement by moving the mirror forward or back becomes a big no-no, though (I move my mirror up into the tube a couple of cm when I use my binoviewer on my 16" scope -- no such thing on a scope like this). And you do need to check the height of the roller bearings with respect to the mirror every time you collimated, to avoid creeping away from the optimal edge support°.
And you will need to number your trusses too, since large collimation adjustments on the primary will typically skew the mirror otherwise.
°One scope I debugged was damaged in transport --a cell part slightly was slightly bent as a result--, enough to make the position and orientation of the mirror do nasty things to the sling edge support -- we had to redo the initial collimation "from the primary up" in the field under a dark sky. I still weep at the lost photons.
Edited by sixela, 15 October 2021 - 05:22 AM.