Why no one makes hypocycloid gearboxes

Gearboxes can turn a weak fast motor and you need

When you buy a Marginally Clever robot kit, you don’t just get an awesome robot. You push the boundaries of science and engineering by funding my research. It’s a bit like I’m trying to run a personal Skunkworks.

You’ve heard me talk often on this site about my desire to build robot arms for less and make them for everyone. I need a gearbox so I can build the arms with affordable 3d printer electronics and hardware. Gearboxes transform weak but fast motors into slow but strong motors. Not every gearbox is right for robots. A good candidate is a hypocycloid, cousin of the harmonic gearbox.

these video’s aren’t mine. They are here to illustrate the concept.

I spent $3000 on a prototype that doesn’t work. In spite of the incredible tolerances and the top quality machining, we couldn’t get the gearbox to run without binding or jamming. More on that at the link, as well as details on my progress with the robot arm.

On the bright side, I can share my progress and save each of you a small fortune. Plus I haven’t given up. At a recent Bring-a-hack dinner after the 2015 Bay Area Maker Faire I got a number of great tips and suggestions about how to try again. They ranged everywhere from “run a kickstarter to fund more development” to “add more lasers”.

Do you have any ideas why the gearbox doesn’t work? Have you tried to make a hypocycloid or a zero-backlash gearbox? Comment below.

For more up-to-the-minute news about stuff I’m making, follow me on instagram.

Only registered users can comment.

  1. Hi Dan,

    I know I am resurrecting an old post, but speed reducers are my baliwick (I have my own speed reducer startup, basically started because I didn’t like cycloidal or harmonic type speed reducers).

    Here are a couple of papers by a friend of mine, where we tried to build ‘high quality’ cycloids, and it details some of our troubles.

    In a nutshell, it is very hard to get simultaneous multipoint contact going on, at the correct locations. While your cycloid may in theory allow for n=teeth points of contact, in practice you will either be extremely overconstrained, or only running on a subset of teeth.

    The guys at Nabtesco are very good at making cycloids, and it is a bit of black magic to get them working well.

    Feel free to ping me back, if you ever want to talk drive trains.