Kinda serendipitously funny that you chose today to bump this old thread Tree MD, since the JB weld casted modification to my rope bike's drive capstan was poured last night, and tested at home today successfully!
After two miserable failed field tests in which the 3 wraps on the capstan walked sideways and derailed before I could get 10-15 feet up the rope, it was nothing but sheer stubborn refusal to admit failure that led to this current drive capstan finally tracking true under load.
http://www.youtube.com/watch?v=6iBuYtmw304
I tried a spiraled worm gear capstan design to physically walk that rope in the right direction, and failed. Then just figuring out how to cast the incline plane at the capstan's initial feed point without tearing the unit apart took another week of trial and error, but it does track true at home, tomorrow field test number 3 will tell the real story.
I suspect that an hourglass designed capstan will be the only way to get the unit to track true in both directions.
I can't believe I spent so much time and effort on this particular prototype since my new design spools the rope into a backpack, making whether the rope tracks true or not irrelevant to it's function. I guess it's against my nature to give up before the job's done!
If it works in the field tomorrow, I'll post my success in vid here. If not? It may be another week or two!
jomoco