When Mice Had Balls

One of the things I've noticed with the rotary encoders I've bought is the low resolution - compared with a purely analog device such as a pot.

Back in the day when mice still had balls ("oh er, sounds a bit rude. Nudge, nudge, wink wink") they were driven by naked rotary encoders with a surprisingly accurate response. This was my first brush with Gray code and I recall being utterly astonished at how simple it was to detect direction.

I'm planning on replacing the nasty pots on my expensive flight yoke with rotary encoders (REs) and that's when I hit an issue: resolution.

The resolution of a potentiometer is theoretically infinite but REs can be quite coarse and that means it's difficult to put small inputs into the design. The same is true of any PC games controller - although curiously some like the XBox use hall effect in the triggers to achieve the same thing. I doodled with that but put it down as a little too fiddly to implement on the existing chassis. I also tried a 9DoF but that proved too complex to work with despite the accuracy. (And 9DoF my hairy white wossisname. These things measure three different things but only one of them is a useful gyroscope.)

Anyhoo.

The resolution problem can be solved with a gearbox - something I picked up from making models as a kid but as things have moved on, I've lost touch with all this stuff and I wonder if anyone is away of something that can be pressed into this sort of service?

I expect I'll need a ratio of at least 10x to drive a lower cost RE but more would be better of course.