AHHHHHH, now that's better.....fig 4 on page 3 of 22 shows the eccentric rolling roller in the external gear teeth, which is just like you describe the motion of the cart with wobbly wheels running on a road that matches the up and down wobble....very interesting.

It would seem from the patent illustration, re fig 4, that the roller does indeed roll cyclically on the gear teeth and not slide as I first thought.

The fig 4 illustration in the patent, although for a single 2D "slice of the action" very neatly puts the whole principle into perspective.

Keeping it simple, and referring to the fig 4 drawing for a single roller, it would appear that a single straight eccentric roller with rounded tooth form, rolling on a wheel with multi straight rounded teeth, would generate a backlashless forward or backward rotation....reversal of the roller would not mean lost motion.

The disadvantage of the single straight roller would be when it reached the top of the gear wheel tooth, and the gear wheel tooth could then slip past it......but with the phased series of rollers in a helical form and the same for the gear wheel round the gear wheel this would not happen.

This means that the roller could be "screw cut" with a rounded tooth form to give a helix that would mate with the gear teeth, which are more difficult to cut, but not so for a CNC guided tool path.

Perhaps this is the answer to the worm drive replacement.

I wonder if a stacked tooth roller could be made DIY by wire cutting hardened segments with multi tooth elements, staggered to simulate a helical roller, and be made to mesh with a similar stacked tooth gearwheel that also had the teeth staggered round the periphery, as in the original illustration of the EC design.

I don't think many of us could envisage cutting the helical form in both roller and gear in one piece and mating them together successfully, but in phased segmented slabs.....very doable.

CNC wire cutting can give you a very accurate series of "slabs", each one identical and accurate to form, that could be assemble together, and staggered to make the drive, and I would think that for practical purposes the slabs at 6mm thickness and 5 of them would yield gear of 30mm width to match a similar width worm assembly.

I think a worm is the wrong term for the helical roller, even though at first glance it appears to be a worm, as it really is a helical cranked pinion with rounded teeth rolling on the periphery of a helical gear wheel also with rounded teeth.
Ian.