Trackball via Ball Transfer Units *Test Results*


25 Jun 2014, 00:32

This project was originally started here:

Here are the goals:

-Use three low friction ball transfer units/bearings to support the main trackball. Alwayse has shipped me six 11-MI-05-17 units to test*. These units are made out of stainless steal and aluminum, guaranteeing long lasting operation. The design will accommodate easy replacement of the units should they become damaged.

-Use a multi-dpi laser sensor** coupled with a Teensy USB controller for flexibility, try to program "twist to scroll" functionality in to custom drivers that will work on Windows, OSX, and Linux. The high resolution laser will facilitate use of the trackball with high resolution screens, keeping trackballs a viable input option for users.

-Trackball to be size of standard billiard ball for easy replacement.

-3D-Printed housing resembling that of a Logitech Trackman FX***, with the addition of more buttons in easy to reach locations (including DPI switches).

-If testing is successful, create group-buy and or kickstarter for DIYers to asseble their own Ball Transfer Trackball.

If you can code, have conceptualization/CAD skills, laser scanning or 3D printing resources, and would like to contribute, feel free to PM me or reply to this thread : ).

* ... units.html and ... s/7431408/

** ... er-sensor/

*** ... 8875ff.jpg
Last edited by Synnöve on 26 Jun 2014, 23:20, edited 3 times in total.


26 Jun 2014, 22:23

I just received the bearings today, and after some very rudimentary tests with sub-optimal placement/angling of the units, I have to say they are incredibly smooth! By far the smoothest trackball I've used with little resistance in any direction, so much so that not very much finger pressure is required for the trackball to spin several revolutions before stopping. Finally getting my hands on these things has lifted my hopes for the feasibility of this project.

There is one, very important caveat though: they are noisy, think skateboard noisy. Now it's important to keep in mind that they are not in a housing and the ball I'm using isn't very dense, but I think it will be similar to the Model M: loud but probably the best switch in terms of feel and durability.

I'm going to purchase some shaping clay to experiment with housing shapes and optimal ball placement. Will post updates when they happen.

User avatar

27 Jun 2014, 09:35

Have not heard about this project until now. It is moderately interesting to me though. I have many trackballs and use a Kensington Expert primarily. Never thought about building my own though.

Question though. Is the sound coming from the bearings themselves?, or is it the trackball moving against the bearings that is causing the noise?

In either case, the enclosed should dramatically effect the sound, and if it is the assemblies themselves, like you said you could probably do some major dampening with the clay. Assuming I am visualizing what you are talking about correctly.

I like the idea of the twist to scroll idea. I do happen to hate it on my Kensington Slimblade though. Mostly because it doesn't seem to balance the control and the speed of scrolling very well. I think ball bearings would aid that dramatically.

Will follow this thread closely!


27 Jun 2014, 11:30

If you look at a cut-away of a ball transfer, you'll see that it's basically a large ball supported by many smaller balls. This is what makes the transfer a low friction omnidirectional bearing, which means, when you include the trackball, the only sizable friction generated is between the larger balls in the transfer unit, and the smaller ones which support it inside the housing. My theory is that the sound is coming from the support balls inside the.

I agree that when they are mounted in a case (they have threaded screws on the bottom of the housing), this should lessen the the loudness dramatically, especially if there is sound dampening in the case as well.

I haven't used the slimblade, but when I do a twist motion with the trackball sitting on three transfer units, the ball spins around it's vertical axis for many revolutions before stopping... basically, there is little resistance to the ball, so I imagine it'll be able to scroll as fast or slow as you want it to.


27 Jun 2014, 18:30

Twist to scroll will be difficult to implement. But my Kensington Expert is a little too high friction for my taste, it doesn't have quite the spin I'd like. Only maybe 2-3 revolutions with a good twist. A solid case may be a good path to go - high pitches are blocked by solid objects.

User avatar

29 Jun 2014, 15:12

Synnöve wrote: -3D-Printed housing resembling that of a Logitech Trackman FX***, with the addition of more buttons in easy to reach locations (including DPI switches).
How about a minimum case:


Buttons to be added. I'm not a trackball user though.

Also, it should be possible to get balls from huge ball bearings in that size, like 3".
Last edited by lowpoly on 01 Jun 2018, 16:36, edited 1 time in total.


16 Jul 2014, 22:26


16 Jul 2014, 22:31

lowpoly wrote: How about a minimum case:
You want the supporting ball units to be much higher up than that, because they need to keep the main trackball from flying out of the housing when you put sideways force on it. I think it works best if you put them at 3 corners of a tetrahedron, though it might work if they’re a bit further down the side than that.


21 Jul 2014, 01:15

I like the sketch Lowpoly did, as for buttons, how about if you press the ball down to do left-clicks and a switch that is natural for the thumb to press for right-clicks, and twist to scroll (need to be some kind of threshold to distinguish accidental and intended twisting)

maybe use some kind of analogue sensor to measure the weight of the unit and when it is not weighing more than normal it senses twisting, when the weight increases a little it is a mouse and when the weight is high it is a left-click.
alternatively, right-click could be activated by lifting the ball a little, decreasing the weight to less than twist-mode instead of a button..

I need to order some parts now I think :P

User avatar
Daniel Beardsmore

21 Jul 2014, 01:51

Aleksander wrote: I like the sketch Lowpoly did, as for buttons, how about if you press the ball down to do left-clicks …
I agree with you in principle, but are you confident that you could press on such a freely-rotating body without inadvertent rotation?

That's the problem I had with the Microsoft Comfort Mouse 4500: the scroll wheel was free rotating (no detents) and it was almost impossible to middle-click as the wheel would just rotate under pressure. That was the biggest reason why I gave up with it, as it was indeed comfortable.


21 Jul 2014, 02:06

I think it would be ok, I have a 3Dconnexion SpacePilot pro for moving around in 3D space, and it has a similar interface.
With a design similar to what Lowpoly suggests, you would have 3-4 fingers in contact with the ball at all times, and you would hold it stationary when pushing down, when scrolling (twisting) you would probably have the pointing-finger resting on top off the ball, holding it pretty locked in on the Z-axis, and rotating the ball with the thumb and ring-finger..

At least in theory :P


21 Jul 2014, 13:00

Pressing down on it for clicks is going to either require more complicated engineering (you basically need to push everything down together including the ball, the ball units, the sensors, the housing), or not work too well, or both. I think it’s more trouble than it’s worth. A trackball and a SpaceThingy (whichever of the 10 versions they’ve made) are substantially different types of devices.

Unrelated comment: I don’t understand why they don’t add some buttons directly on the surface of the SpaceThingy, and give it a more ergonomic shape than a sphere/puck. In general I think the SpaceNavigator is too small/not heavy enough, and the SpacePilot and similar ones have really terrible button layouts and mostly very gimmicky extra features. I guess that’s what’s profitable (segment the market and try to upsell people on the expensive gimmicky ones), but it doesn’t seem like very good design to me.

User avatar
Daniel Beardsmore

21 Jul 2014, 22:33

Aleksander wrote: With a design similar to what Lowpoly suggests, you would have 3-4 fingers in contact with the ball at all times, and you would hold it stationary when pushing down …
One of the things that irritates me is that I frequently click before stopping the pointer. It took me a long time to figure out why, often, a mouse click would focus a control but not click it. The mouse is still moving at the time I press the left button, with the result being that I'm trying to drag the control by mistake.

I need more convincing that it would be possible to press the trackball without it rotating at all.

User avatar

21 Jul 2014, 22:43

And add to that the trickiness of performing a perfectly simple drag: you'd have to keep the ball above the pressure threshold while rotating it in two axes. Seems like a lot to expect of users, as well as from the design.

Spin scrolling sounds nice though, above a trigger value so you don't set it off accidentally. Indeed, I suspect heuristics will be make or break for this whole concept.


22 Jul 2014, 07:09

Those ball transfer units look nice. In my experience, supporting the ball just below the equator seems to yield the best results. My current trackball has simple metal bearings in a 3D printed housing with a Delrin ball. While I've been using this for years now and love it, I've often thought it would be nice to have a lower drag option and this looks like a good one. I'll have to order a few of those "mini ball transfer units" and play. If you end up needing any help with the sensor, let me know.



22 Jul 2014, 13:38

MrJohnK: are you this John K? ... cal-sensor

If so, I absolutely want some advice about the sensor. I’m going to start in on programming a Teensy 3.0 to talk to it sometime within the next 2 weeks hopefully.


24 Jul 2014, 05:59

@Jacobolus Yes, that is me. Feel free to email any questions you have to I have a Teensy 3.0 here I've never hooked up before, but I'll do my best to help out any questions you have. I've used the Teensy 2.0 on my Trackball.

Post Reply

Return to “Workshop”