217416670000000000000000000000:1 Cycloidal Drive

I'm Shirdel, a student at Victoria Park CI and the current Build and Electronics leader for FRC team 4914.

My entry to the Digital Fabrication Student Design Challenge is a 3D-printed cycloidal speed reducer.

Why?

Why not - ludicrous speed reductions are objectively cool.

This was developed in an effort to make a compact configurable gearbox with a small form factor. Other than that, it has no purpose - it would take longer than the lifespan of the universe to have the output rotate once while spinning the input at 1000rpm.

Fusion 360 online view

Variety of M3 BHCS hardware

6 M3 nuts

M3 spiral-point tap (optional)

~500g filament (can be any material that's not flexible)

~200 4.5mm BB pellets

Grease - optional (I used petroleum jelly, it works fine)

What is a cycloidal drive?

I've played around with making planetary gearboxes before for a different project, and the cycloidal drive is an extension of that concept. It can be thought of as a planetary gear set but with only 1 planet, which is 1 tooth smaller than the ring gear. Because the sizes of the planet and ring gear are so close, an involute profile cannot fit and a cycloidal gear profile must be used.

How is the gearbox constructed?

The largest concern with cycloidal drives is oscillation caused by the rotor moving around the perimeter. To counteract this, the gold part, which is the input, has 2 rotors in each of the 2 stages offset 180° so that the oscillations of one cancel the other, greatly reducing vibrations. The rotors have 35 teeth each, with 36 teeth on the ring gear, giving a -35:1 reduction. Since the gold section has 2 stages, a reduction of 1225:1 is achieved.

All the bearings are printed to save cost. If this were to have a longer service life, then metal bearings probably would have been better. For the most part, I made thrust bearings with a diamond-shaped track which helps the ball bearings ride over printing artifacts. The bearings that are mounted on shafts were printed using files purchased from Cults3D.

After the initial 2 stages, a less complex construction method is used because there's really no point as there's no expectation of high speeds. Initially, I thought that the rotor and output shaft could be combined into 1 piece, which turned out to be very wrong - more in the next step. The idea was that as long as the housing was extended, one could stack more and more rotors to gain larger reductions. Even with this simplification, there was still a large number of parts that took 2 days to print.

Turns out, the output shaft and rotor combination are essential to the cycloidal drive's function, and cannot be consolidated. As a quick fix, I made the rotors half the thickness so that the same number would fit in the same space (6mm to 3mm thickness), and had the output shafts run along with the ring gear teeth, instead of on a bearing, which would've increased complexity and cost.

Assembly is also a challenge, as it's very easy to misalign rotors and cause binding - there is only 1 orientation in which each rotor will fit, and it's completely dependent on the position of each previous stage. Luckily, due to poor tolerances and low expectations for use, it's not actually that important to keep track of this for the orange section of the gearbox. A spray of dry PTFE WD-40 lubricant helped it move a little better, but it's pretty much pointless to apply it, hence why it's left off the BOM.

Congratulations - you know have a monetarily and temporally costly paperweight!

Again, it really is pointless - except that you get to brag about the huge reduction.