Fitting and Constraint
One bearing alone is not enough for fully constraining a gear (see above video). A pair of two bearings gives better constraint, but that also introduces an alignment problem.
Did you notice any misalignment in the CAD? The carbon fiber sheet (the black piece) constrains the bearing in an inaccurate position. The axes of the two bearings are parallel but off by 5 thou. That 5-thou results in a disastrous gear-meshing problem (see following video).
“Off by 5 thou? Just redraw the CAD so that the bearing is constraint with good alignment.”
Sadly, I cannot do that. In fabrication, I am technically not able to measure how much the alignment is off. Maybe it’s off by 1 thou, maybe 10 thou. I don’t know. So, redrawing the CAD will not solve the problem.
The carbon fiber sheet over-constrain the bearing with an interference fit. So, I get rid of the over-constraint by removing the interference fit.
Now, the carbon fiber sheet is in clearance fit with the bearing (see above picture). Once the bearing is placed into the aligned position through an iterative process (see video below), it is fixed in position by both the glue and the clamping force.
The clamping force is provided by the interference fit of the Acetal dowel pin. I brought oversized pins and sand them down to press fit (see left picture). Also for an interference fit, I have to add ~15 thou of tape between the carbon fiber rod and bearing, which used to had a terrible 10 thou clearance fit.
The result is good. Gears mesh well. The video you see is running at ~30% power. I later hang the assembly with ropes and run it at full power for half an hour. Gears still mesh well.
I cannot wait to assemble the whole ornithopter and do the drop test.