Super Grippy 3D Printed TPU90A Tires for FIRST FRC

Is the CoF the same when force is applied sideways? (Locking the wheels at 90 instead of pushing towards the other robot)

Similar to the V38’s we ran with week 1, V33 tread has fewer and larger spikes than the V40. It also has a stiffer suspension pattern than future iterations. The V40 tire we found was more durable than the tires with larger spikes while retaining the high CoF.

Also, for anyone printing reference models, in our endless efforts to rid disgusting 10-32 bolts and instead use the far superior 10-24, we have been using our DiffSwerve bevel gears on the wheels, so the normal SDS bevel will not currently fit. Easy change for us to make in the future, but the current posted model will not work with the stock SDS bevel gear that attaches to the hub.

1 Like

Week 1 with V38, we changed them before quals, part way through, before playoffs, and they were gone after finals. That being said, the larger spikes on V38 would wear down rapidly when the wheels were doing burnouts, while it seems to take a lot more for the V40s to start to show signs of wear. We did lots of burnouts at SE MASS between autos and needing some driver practice.
Week 3 and on we were getting about 12 matches per set, usually just changing them between quals and playoffs. Most of the time we could have kept using the tires from quals, but we haven’t done friction testing on worn tires yet, so we just put a fresh set on to be safe (and they are quick and easy to change). When they wear, they do not disintegrate, but the studs get smaller. We’ve hit the charging station from the side during auto in such a way that it took a small chunk out of the tire, and we did not notice until we went to change them later in the day. Even with a small piece gone they drove and held up great.


Yup. They are omnidirectional, including the suspension. Unlike black nitrile, the spikes themselves are symmetrical and will behave the same from any direction. Suspension wise, they have staggered leaf springs going opposite directions to make sure their performance isn’t direction dependent. Hitting a bump from the front of the tire will feel the same as hitting one from the back of the tire. Here’s a better look at the leaf springs.


I’m super curious about the Onshape workflows to create geometry this complex. Really hope some of the native Onshape iterations can get released at some point, in addition to the grabcad snapshots I saw upthread :slight_smile:

Edit: the more I think about it, the more it feels to me like PTC could be pitched on sponsoring/showcasing this project directly. “yes, our tool can create the impossible geometry that 3dp enables” e.g. you don’t need {majorbrand} to do epic stuff.


What made you want to introduce 10-24 into a FRC environment that’s loaded with OTS components using 10-32 threads?


I’m hoping there’s a huge level of scarcasm but it’s hard to say


I don’t think there was any sarcasm in that question.

1 Like

Not the question just saying the “far superior” 10-24


If we were to sell these tires they would likely use the standard SDS bevel because of how much easier it would make it for other teams. We used our own bevels because we had a large stockpile already made and our standard bolt is the 10-24, making it easier for us to use. Just a running joke on our team to slam on the 10-32.

I wasn’t on the team when TJ made the 10-24 standard, but from what I’ve been told, part of it is that in most cases where you need a #10 bolt, you have the space for the coarse thread, and the coarse thread is easier/quicker to bolt in without cross threading.

Edit: it may be ambiguous as to how similar our bevel is to the standard one; the only difference between the two package wise is highlighted in red here. This was taken when we originally made the conversion at V20.


When we release the design files we’ll show the native Onshape as well. This complex geometry really approaches the limit of what Onshape (and PreForm for that matter) can actually do. With a good internet connection, it takes over an hour to upload the print job to the Fuse (very unusual for that machine) and Onshape takes FOREVER to load changes. Interacting with the model AFTER it’s been generated has been fine, but we’ve crashed the document numerous times trying to generate the literal thousands of studs on the tire.


What does the wheel look like when it is on the robot? How much is it compressed? (On a tile floor would be great for showing this)
I’m looking for a starting point in designing my FDM printable copy.

1 Like

I love the idea of optimizing for FDM!
You might look at spiraling the suspension ribs and making them solid strips (not a nest of beams). I don’t think a simple angle is going to have a strong directionality to it, despite looking like it might.


Any idea on when the design files would be released to the public?

1 Like

I wonder if you could get all of the suspension that you want just by adjusting number of walls, infill type, and infill density.

Saves a lot of design complexity that way!


IIRC 125 did custom wheels a couple years ago, with larger beams than this design. @Brandon_Holley did y’all do similar benchmarking that you could share?
(Perfect world - it would be so cool to send a couple old wheels over for 88 to benchmark on the same rig they tested their design on, but that’s a lot of work/coordination)

1 Like

I think 88s and our approach were pretty similar in what we were trying to accomplish, but with some fairly notable differences. Because we were making our wheels out of PA12 (Nylon), we ended up focusing on making the “rim” of the wheel as rigid as possible, and allowing that rigid rim to ‘float’ around the center hub via the connected ribbing.

We also were building our wheels for an 8WD, with each of those wheels being 3" wide. In terms of stiffness needs, this is a substantially different system than a 4WD swerve with 3" diameter and 1" wide wheels. We had so many “springs” between us and the ground that all needed to be very very compliant for us to get the effect we really wanted.

Because we were working with a relatively stiffer material (versus something like a 40A TPU) we started from the beginning with the ‘thin strip’ approach. You can see here in our first trial wheel the nature of that ribbing pattern. This particular wheel ended up being WAY too stiff.

We also disliked that the entire center of the wheel was disconnected to the center hub, so over time and various trials we ended up trying several iterations of rib span, thickness, quantity and width. We got to a happy place on overall stiffness and performance with the ones in this photo (Pasta for scale)

The last major challenge for these particular ribs however was related to how they performed under hard impacts. It wasn’t the initial impact that damaged the ribs, it was the fact that there was virtually no damping in the springs. After the spring was loaded and rapidly released, like when you drive over a bump and are floating through the air for a brief moment, the rapid snapback to its neutral position would cause the ribs to snap.

Our solution to this problem was to add damping material! We made a clamp on mold, and filled the wheels with molded foam to ensure the snap back action wouldn’t destroy the ribs. This ended up working very well and was our final solution we put on the robot.

88s solution is a lot cleaner overall, with TPU being a better spring material for this application, and also enabling the tread to be printed into the same part. The thin ribs we used did allow for a ton of tuning and flexibility, and I’d expect them to survive a lot better if they weren’t as long and had a material with much more inherit damping like TPU.


Can we make this the new standard for scale? No more banana, only cute doggos?


Unfortunately, I think dogs vary in scale a bit more than bananas. Maybe if you specify a specific dimensional range for the dogs?


You can only use dogs that are 18" ± 3" long , 12" ± 2" tall, and 10" ± 2.5" wide.