Found it!
http://www.chiefdelphi.com/forums/sh...ad.php?t=84014
We ran this with 2/3 of a FP gearbox, and the beauty of it -- besides the fact that it worked really well (see the video linked in my original post in this thread) is that there is no chance of it ever stalling, because it never, ever puts any downward force on the ball (aside from the torque to give it the backspin).
I was truly shocked at how many teams seemed to have considered ball control a sort of add-on, last-minute engineering decision. This is one of the first things we built, and we went through several iterative designs before settling on this one. (Good thing every part of our robot is modular, eh?)