A 3D Printed Cycloidal Gearbox Hits 92% Efficiency in Testing

Cycloidal gearboxes show up in robot arms, humanoid platforms, and quadrupeds because they deliver high torque in a compact package with very low backlash. The catch is that commercial cycloidal drives cost money, and buying one just to test a robot concept is not always practical. Enter Advanced Hobby Lab, who designed a 3D printed cycloidal gearbox for NEMA 17 stepper motors and then put it through real bench tests.

The Build

The design uses two cycloidal gears stacked 180 degrees apart to cancel out vibration. That is a standard approach in industrial drives, and it works just as well in FDM prints. The gears are printed in a high-strength filament and run against metal pins in the housing. A few bearings, a couple of set screws, and the whole assembly bolts together without custom tooling.

Advanced Hobby Lab measured efficiency at 92% during testing. That is a real number, not a simulation. The gearbox also runs quieter than a comparable 3D printed planetary gearbox, which matters if the motor sits inside a robot chassis that carries a microphone or a camera.

What the Test Actually Showed

The 92% efficiency figure comes from measuring actual torque output against the ideal theoretical torque for a 15:1 reduction ratio. The team found that some loss comes from friction in the bearing interfaces, not from the cycloidal geometry itself. That points to a clear upgrade path: better bearings and tighter tolerancing should push the number higher on a second revision.

The print files are available for anyone who wants to replicate the build. That makes this one of the better-documented open gearbox projects on the internet right now. You can also opt for a dual-nested cycloidal design that promises even tighter packaging and fewer parts.

Why It Is Useful

If you are building a small robot and need torque without bulk, a 3D printed cycloidal drive is now a proven option. The 92% efficiency figure is well above the 55% efficiency reported by other hobbyist FDM cycloidal projects in earlier testing. That gap reflects better material choices, improved tolerancing, and the stacked-gear vibration cancellation approach. None of those require industrial equipment. They just require someone to publish their test data.