Trackball Smoothness Revisited (Roller Bearing Mod)

After upgrading the CPU, battery board, and side panels in my Reform, I decided to investigate trackball smoothness again. The community previously converged on static ball bearings embedded in the cup: Trackball Replacement Cup with Bearings. The Ploopy open source trackball team uses roller bearings (MR63ZZ) instead of static bearings, so I decided to give that a try.

I had to redesign the trackball mount to accommodate the larger bearings. I started with @Chartreuse’s OpenSCAD model, but this is basically a new project. Only some original measurements and comment fragments survived the transition. The project is here.

Compared to the original cup, it’s a little chunky, but it still fits fine in the Reform. I printed it with the cup facing upward (no supports needed). I used the bearing dowel and bearing press models from the Ploopy project. I printed the dowels vertically.

Is it an improvement? Maybe.

These are my impressions, but there are a lot of variables in play here, so you may have a different experience. When I did the original static bearing modification, I used 2.5mm ceramic ball bearings.

The roller bearings are slightly quieter, and seem to have less stiction (resistance to moving from a standstill), but more friction (the static bearings allow the ball to roll longer if you give it a hard spin). The roller bearings generally feel very smooth, but occasionally I feel a small “catch” in the movement. Some of this may depend on the quality of the bearings. I somewhat randomly chose these (Amazon link). I don’t know if they are high quality, average, or low quality.

If you’re happy with the static bearings, I’m not sure there’s a strong reason to try this, but I do think I slightly prefer the roller bearings. I’m going to try living with them for a while. Here’s an interesting comparison of trackball bearings, including more exotic ball transfer units: Adventures in Ploopy-land

I’m happy to answer any questions about this project if you decide to experiment with it.


Very nice! I’m curious how the linear travel of those bearings would be an improvement over the static less resistive bearings. Are you able to quantify that a little bit more? I think from a logic stand point this would be a big improvement. But your experience doesn’t seem to make that that sure conclusion.

Excellent work regardless! Thank you very much for sharing!

Really nice mod. One thing I want to try is moving the sensor from the bottom of the cup higher up one wall. Sometimes my trackball stops respondin until I turn my reform over and spin the ball a bit. I think it’s dust or debris interfering with the sensor.

Thanks! I’m no mechanical engineer, so this is just my amateur conjecture.

A single roller bearing is super smooth, so I think the intuition is that this should be a clear improvement over the static bearings. The problem is that we need 3 bearings to support the trackball, and the ball’s movement can only be optimally aligned with at most 1 roller bearing at a time (and maybe none of them).

In the non-optimal case, the roller bearing is more like a static ball bearing, but with a less optimal contact surface (edge of cylinder versus point on sphere).

1 Like

Interesting. I use a Kengsington Trackball on my desktop and I just now removed the ball and looked at things. It just has static “bearings”. They don’t feel like they roll at all though. I think they are just a little slippery and because of the weight and size of the ball, this allows the ball to kind of glide in it. When cleaning it out occasionally I notice a huge improvement to the effort of roiling the ball and the smoothness of the movement.

I might just have to give your mod a shot! Thanks again!

Regardless of bearing, the trackball is noticeably louder when it’s installed in the Reform than it is when it’s freestanding. The rigid aluminum case seems to work as an amplifier. I tried cushioning the trackball PCB with some scrap rubber padding around the screws holes, but I’m not sure it made any difference.

When I was doing that experiment, I noticed that the screws holding the 3D-printed cup to the PCB had gotten a little loose already. If I print another cup, I will probably use threaded inserts for the mounting holes to get a more reliable connection. I’d like to do that for the sensor mount too, but I don’t think there’s room.

I also just broke a leg on one of the curved key caps—oops! Probably inevitable considering how many times I’ve removed the trackball from the laptop. Fortunately, it seems to be okay with one good leg. I’ll have to get some spare caps in the future.

Adding fixed metal or ceramic bearings project is linked in the first post. It aligns with what the desktop trackballs are doing, is mechanically simpler, and reportedly is at least as good if not better.

It’s great doing research and trying different things. At the same time coming up with things that improve on decades of prior art is not that simple