TPU Wheels
Over the last few months, a team of four students (including me) have been working on making and testing 3D printable wheels for MAXSwerve. FRC team 88 worked on wheels much like these, releasing a design at the end of the 2023 competition season. We saw their design and wanted to try to adapt it to MAXSwerve.
Why?
Go fast or get passed.
These wheels have a special spike pattern that grips into carpet, allowing wheels to achieve a much higher coefficient of friction (COF) than standard wheels.
What is "COF"?
The Coefficient of Friction, or COF, is the ratio between frictional force and normal force (force pushing down) of an object. Grippier materials have higher COFs. In this instance, what we’re measuring isn’t really the COF, but rather the strength of the mechanical lock between the spikes and the carpet. We’re treating that as the COF.
In numbers, these wheels have a COF of 1.43 while normal treaded wheels have a COF of 0.97. A wheel’s COF directly corresponds to how fast the robot can accelerate. If it’s too low, the wheels will slip instead of putting all of the power into accelerating. This is especially pronounced with light robots and more powerful motors (Vortexes and Krakens). This doesn’t lead to too much added sprint speed, but can help you in handling and being defended. If you have more grip than your opponent, you might be able to push them out of your way, or at least won’t be pushed by them.
Play around with this drivetrain simulator if you want to see for yourself: Drivetrain Simulator - AMB Robotics Calculator
Keep in mind that the sim doesn’t completely reflect real life values.
How to try them out
The design that we’ve been testing is for REV Maxswerve. Many other people have been testing these out for SDS and WCP swerve modules, so if you want to see those, click here.
Our leading design right now is called “TPU Wheel Main” in the Onshape document.
CAD link
The hub is meant to be printed out of Markforged’s Onyx, and I wouldn’t recommend using anything less strong than any Nylon materials.
The tread is meant to be printed out of TPU 95A with 100% infill. Using a relatively low layer height helps keep definition in the spikes. I’ve been using a 0.8mm layer height. You also need to use supports for the part that goes into the hub, but not the spikes, which can be annoying.
As it stands, it uses 67g of TPU per wheel, which costs ~$2, or $8 for a set.
Recently, team 2847 ran our design at an off-season competition. The wheels lasted the length of the competition (11 matches!), and this is what the wheels looked like afterwards:
Very big thanks to them for running these!
If you want to see most of the possible wheels you can try, check out this doc which has almost every published design for MAXSwerve, and a few for SDS.
A word of caution:
These wheels still haven’t been tested enough to be 100% sure that they won’t break very quickly, midmatch. Please validate for yourself that these have a long enough lifespan for you.
Static COF Testing
| TPU Wheel |
Treaded Wheel |
REV Molded Wheel |
| ≥1.43 |
0.97 |
0.87 |
Our testing rig is very simple, all we did was staple an extra piece of carpet onto a plywood board and lifted it up until the robot slipped. We measured this angle and took the tangent of that angle, and that gave us our static COF.
≥1.43 means that the robot fell over before the robot slipped, so we can’t know how much it was exactly above the 1.43. 88’s original wheels had a COF of ~1.6, so it’s probably under that.
Proof
Video of some of the tests
Challenges and Design
During the development of these wheels, we had very many challenges. 88 gave us a good starting point, but they had designed their wheels for SDS modules, which have a 4" diameter by 1.5" width wheel. MAXSwerve, on the other hand, has a 3" diameter by 1" wide wheel. Additionally, 88 designed theirs for SLS printers, which are suuuuuper expensive.
We widened the width to 1.22" from 1" because there was extra space, then we also added a notch to allow the wheel to be hotswapped. The notch goes to the top, and then the wheel clears the pinion driving the bevel gear.
We had a lot of unexpected durability issues with this wheel, leading to many, many redesigns. I don’t know how useful going into detail is here, but here’s an overview of what we’ve changed.
- The tab that goes into the hub got stronger (more like 88’s)
- Replaced the metal part of the hub with a bigger 3dp part to support the wheel on the side better
- Moved away from using gyroid infill for suspention due to my speculation that it breaks when it bends too much
There’s a lot of changes since the original design, but those are the most important ones.
I wish we could have tested these for accuracy in auto, but we ran out of time 
Presentation
We gave a presentation about these wheels for our parents, mentors, and peers at our last meeting of the year (2023). Here’s the poster that we presented:
Overall it went really well! The the text was a bit hard to read but it seemed like everyone understood what we worked on.
There were four other thesis groups and the demo bot group that also presented, and I’m sure you’ll hear about the others soon!
