Non-jamming centrifugal engine starter

The car was started by a small electric motor that had to crank a petrol engine which was, frequently, already turning. The mechanism that did it threw a gear at a spinning flywheel and hoped the teeth would line up.

The problem

The competition car's starter threw a pinion into a spinning flywheel to crank the engine, and about 10% of starts jammed. A jam during a timed run ends the run.

What I did

Replaced the throw-in pinion with a dog gear arrangement, in which a PETG driving gear stays permanently meshed and nothing has to engage under load. Debugged the redesign through 3 successive failure modes before it held.

What happened

A starter that neither jammed nor broke across 1.5 seasons of testing, against roughly 10% of starts before the redesign.

What I learned

Designing a failure out beats defending against it: the event the jam depended on no longer exists, so that mode cannot return. The 3 failure modes that followed were new ones, and testing was the only thing that found them.

The mechanism, and why it jammed

A metal 3D-printed pinion sat on a grooved sleeve press-fitted to the motor shaft. When the motor spun, the pinion's own inertia drove it along the spiral groove to the end of the shaft, where its teeth were thrown into the flywheel teeth in the hope of engaging.

The hope was the problem. The team frequently needed to start the engine while the flywheel was already spinning, and then the teeth did not align. About 10% of starts jammed, and the pinion was slow to replace and cost about $400. A jam during a timed run ends the run - a starter fault of this kind is why a newly designed car body never ran in competition the previous season.

Removing the engagement

If the jam happens at the moment of engagement, the thing to remove is the engagement. I replaced the throw-in pinion with a dog gear arrangement: a PETG driving gear that stays permanently meshed with the metal flywheel, continuously in contact and supported rather than cantilevered, with 3 dog teeth on its rear face. The grooved sleeve stayed, but now it only pushes a dog into those teeth instead of slamming gear teeth together.

Permanent mesh is a wear trade, and it is acceptable here only because the car neither runs long enough nor fast enough for wear to matter.

Before and after

Old mechanism	metal 3D-printed pinion, thrown into the flywheel at engagement
New mechanism	PETG driving gear, permanently meshed, with 3 dog teeth on its rear face
Jams before	roughly 10% of starts
Jams after	none, across 1.5 seasons of testing
Part replaced	cost about $400
Part that replaced it	costs practically nothing, and can be reprinted in-house

Then it broke, three times

Designing the jam out was the easy part. Making the redesign survive took three failures in turn.

The press fit started slipping off the motor shaft under engagement load.
A set screw stopped the slipping and broke the sleeve. It introduced a stress concentration, and the grooved sleeve failed there under load. I machined a flat onto the shaft instead and iterated on press-fit tolerances until a fit held without slipping.
The sleeve then failed in torsion. Others proposed printing it in a stronger resin, which was more brittle and more expensive. I enlarged the sleeve diameter substantially instead, because torsional capacity rises steeply with diameter, and reoriented the print so the layer lines run parallel to the shaft rather than across it.

The decisions I would defend

Geometry over material. A stronger resin was the obvious purchase, and it would have been more brittle, more expensive, and still the same shape. Diameter was both cheaper and less brittle.

Orientation as a design variable. An FDM part is anisotropic, so which way it is printed belongs to the design rather than to the printing. Turning the layer lines to run along the shaft put that anisotropy on the right side of the failure mode.