Dan
For the kit designs I had two overriding criteria: they had to be as cheap as possible whilst still providing acceptable performance and they had to be almost universally applicable (hence your comment about platters / bearings). Unfortunately this means that they are a long way from optimised for any specific application.
For the bespoke designs I gather as much information as possible, to the extent of getting specific numbers for the rotational moments of inertia of the motors used from the motor manufacturers or in the case of one drive having to measure the numbers myself. I then build a model in a Spice program using some translational analogies and spend a lot of hours doing dynamic modelling.
Depending on the sophistication of the drive, the specific platter and bearing numbers can have some influence on dynamic performance but the most important parameters are the motor and its electrical control. When I am happy that I understand what's going on, we go to prototype.
The results? Well, I think you'll be surprised at what can be achieved with standard AC motors and belts even less compliant than your Mylar when the drive mechanics are understood. Similarly if a manufacturer sends me several samples of a high cost three phase motor and says "do your best then bill me" the results can be pretty good.
The downside is that the controllers end up being quite expensive. I don't know if there's a viable model for producing an aftermarket controller using any of these techniques.
Mark Kelly