Alternative Reducer Designs

The planetary is what we run. These are the other reducer types we weighed, and the two we're keeping as live alternatives to print and test.

Why the others lost (for this application)

Requirements: ~11:1, small + coaxial, cheap to reprint, backdrivable (manual fallback if the electronics die).

Type	Verdict	Reason
Planetary ✅	in use	coaxial, compact, load shared across planets, backdrivable. Least forgiving of print error. → reducer
Folded compound gear train	live alternative	each mesh is one adjustable pair → far more print-forgiving than the planetary. But two axes (wider) and stacked backlash. → folded compound train
Cycloidal	live alternative (shelved)	highest torque + shock tolerance, and printable. But the eccentric mounts on the same D-flat, and it's a pile of pins + bearings. → cycloidal
Worm	rejected	self-locking → not backdrivable; the wheels lock if power dies.
Harmonic (strain wave)	rejected	needs a spring-steel flexspline you can't print with any strength; overkill ratio.
Belt	rejected	needs a tensioner; low ratio per stage; creeps under a held load.

Torque-capacity ranking

How much output torque a design can hold comes down to how many contacts share the load — it fails at its single most-stressed contact:

cycloidal > planetary > spur / folded train > belt

A cycloidal spreads load across many lobe + pin contacts at once; a planetary splits it across its planets; a spur/folded train runs the full torque through a single tooth pair per mesh. A printed harmonic would invert to last — its thin flexspline can't carry torque in plastic.

So note the tension: the folded train (most print-forgiving) is the weakest on torque, and the cycloidal (shelved) is the strongest. The planetary in use sits in the middle.