A note from Alan adds more data about troubleshooting problems with the classic Kensington Expert Mouse trackball scroll ring:
I have two comments and a question: first I made the mistake of purchasing 4 used expert mice on ebay etc and each had a different problem but 3 of 4 also had faulty scroll rings. 2nd: one of them was dated 2020 (a wireless version). so they definitely haven’t fixed this issue and it’s very wide spread (or maybe why shady sellers decide to part ways with their trackballs).
question: from reading across your quotes it’s not clear but it seems like there is no real consistent fix to this issue nor a really strong conclusion as to what causes it? My futzing with a couple of these does seem to suggest that alignment of the ring makes a difference but not a lasting one.
As far as the alignment non-fix goes, tweaking the detector position just changes the amount of light passing through the wrong side of the reversed IC, without solving the problem. That’s what we’ve all done, with essentially the same results: feels good, doesn’t last.
Kensington (whoever they are these days) may have fixed the problem with a different quadrature detector oriented in the proper direction, but that’s not something we civilians can accomplish.
It should be possible to unsolder the reversed detector (if, indeed, it is), aim the lens (if that’s what it is) at the emitter, then somehow resolder the leads to the same pads. Perhaps flip it to put the leads on the top, away from the PCB, secure it with a generous blob of hot-melt glue, and connect jumpers from pads to leads?
So far, the two new-ish units on my desks continue to work well, depriving me of sufficient motivation to dig into my junkers.
If anybody is willing to hack their defunct trackball, please let us all know what happened!
Because you may be reading this in our future, comments on this particular post will probably have been disabled to reduce the attack surface for spammers. Send me an email / use the comment form (linky over on the right), or comment on the post of the day and I’ll sort it out. Thanks!
The Smell of Molten Projects in the Morning 
You might remember that a decade (or more?) ago I once posted my scrollring broke within a few years. Well, I never stopped using that mouse, but one day it just started working again. It’s been two or three years since, still going strong.
Spontaneous remission is a thing!
I’d still like to see any single explanation for all the evidence.
The thing that is so sad here is millions of ball mice use(d) optical encoders reliably. All my old ball mice failed because of mechanical issues with the rollers not making good contact. Never from the optical encoder not picking up movement. And yet here Kensington can’t get it right despite having all the space in the world (relative to the inside of a mouse, anyway) and at least a decade of time to redesign it.
I personally feel like it’s more likely to be a poor software implementation of the optical tracking algorithm than a hardware failure since the hardware involved is pretty simple. Unless there’s more going on in the quadrature detector than I imagine (I’m assuming it’s just two photodetectors in a single blob of plastic that keeps them properly aligned).
The Expert Mouse may be one of those products that sells well enough to continue making it, but not well enough to justify making it better.
Because the revised version has a different detector on a PCB hacked into the original location, I think the problem really was bad hardware. If it had been a physically reversed detector, surely they would have figured out how to reposition it on that new PCB, rather than spin up a whole new part number.
I just set up my last spare on the laser cutter and all three continue working. I’m afraid to look inside any of them for fear of disturbing the magick!
It would be interesting to see what the signals coming out of the detector are and if you could simply fake them with a pair of microswitches. If the detector is just a pair of photodetectors I don’t think that would be possible since you need pairs of signals over time, but if it does do the “integration” internally then it might be as simple as pulling the right pin high.
I don’t find the motion of the scroll wheel so smooth even when it works that a simple scroll up or scroll down button wouldn’t do as well (I suspect).
Also there are a bunch of test points on my PCB, maybe one (or more specifically, two) are usable for sending scroll events.
So many projects, so little time … [sigh]