Hi,
having read further through this thread one thing comes to mind: The actual length and shear thickness, mostly width of the belt in use.
Some say that this is of quite some importance and actually kept the belt VERY short (pulley under the platter) and beefy, (not thin, round or square) like the flat belts e.g. SME and Linn use.
Then there is the motor controllers, done over a number of times by Linn, and at least once by SME.
Listening to a thin (square) belt of a top Pro-ject table and thin (round) on TransRotor Z3 (sans controller) and then a short (flat) belt with controller on SME gives a pretty clear indication of the difference in transience/attack.
So much so, that on VERY dynamic vinyl the LP seems to 'slip' on the platter (rather then the belt) in the case of the SME -- if the LP is not fixed with the clamp provided.
This raises the question of 'slippage' of the vinyl on the platter in the case high torque direct OR belt drive, and if no platter clamp is used or even available.
Lastly, if the drive is THAT tightly coupled, then the motor controller's performance comes a LOT more to the fore. With the hole drive line becoming so much more unforgiving with next to no measurable 'slip' it is now unable 'paint over' some dynamic problem passages'.
Fix one thing, and it will reveal the next issue for sure.
Too much platter weight (never mind just the main bearing implications) in such a 'tight coupled' scenario may just 'over-load' the motor/controller, running behind torque demand and its reverse, producing delay and over-shoot if the platter's inertia is out of tune with the motor's torque delivery and the controllers feedback loop speed.
All this can explain why a more 'benign' coupling might sound better or more natural, even if slightly less 'dynamic'.
Greetings,