As the secondary gets bigger the support requirements get stiffer (pun). What works fine for a 2.6" quartz secondary isn't necessarily going to work at 8" without good design, similar to supporting an 8" full-thick mirror vs a 20" thin mirror. Recall here (I hope!) that it's not just aspect ratio at work - the effectivness of any multi-point cell goes down roughly as the square of the size for the same aspect ratio - a 200x20 f/4 receives 4x better RMS support than a 400x40 f/4 for the same cell design...

I don't think the CTE of the adhesive makes much difference for flexible bonds like RTV, these are small areas holding to larger pieces. What you want to minimize is the thermal differential stress generated over the much larger area of the secondary and the backing plate, as those will load the secondary through the bond pads. A thicker bond pad takes up more of that stress without loading the secondary, but as the secondary gets larger the bond pads have to not allow much shear in altitude changes, so they have to become proportionally thinner, as Preston explains.

The bottom line is the same as the first sentence: for larger (more massive) secondaries the support has to be better, and CF does the trick for low-expansion substrates. So my 2.6" spider with a stainless "flex" plate is fine, but it will only scale up so far before something more exotic is called for. In the case of the wire spider design, it could indeed be a CF plate mounted to a central pivot.

Best,
Mark