Yeah, I think one of the nice things about making it a "clean" number (either an integer or a rational with a small integer denominator) is that you can easily validate it without needing high-precision measurement equipment: put a mark on both gears (maybe even embedded in the 3D print), line up the marks, rotate the large gear 1 full rotation, and count the number of rotations the smaller gear makes. Check to see if the marks line up perfectly after those rotations.

▲

nullc 2 days ago | parent [-]

His capstan reduction can't go all the way around even once.

▲

hinkley 2 days ago | parent [-]

It could though. I don’t think the thought has occurred to him yet. He could make the stack twice as high and go around twice as far. If you moved the worm gear you could go farther, but I don’t know how he would do that with his drive.

He could also go with narrower rope, and spread the load over more windings, which would give him more throw.

▲

mlhpdx 2 days ago | parent [-]

What are the loads on a drive in this use case? 35lbs robot bouncing on one foot would be 350lbs-ish of dynamic load? That can be handled with 2mm Dyneema.

	▲	hinkley 20 hours ago \| parent [-]
		At 2:02 he shows a working model with a smaller diameter cord than his earlier ones. Looked like he’s already using 1/8 or smaller. Notice that the fastener bearing the weight of the cord is wrapped around two bends. If I recall my physics properly, each of those direction changes behaves like a pulley, halving the strain on the bight (dyneema is difficult to fasten though. Slippery bugger.) No wait, he is already using 2mm: https://www.aaedmusa.com/projects/cara but it’s 2 ropes per leg, so maybe he could use 1/16” rope (1.6mm).