2910 MK2 Swerve Module Release

Speaking of 3D printed swerve modules… All parts except gears printed out of black PLA on a CR-10, 100% infill (slightly overkill, we’ve learned more about 3D printing since then. Still, nothing broke)

The printed bevel gears (Taulman 910, 100% infill, 0.2mm layers) actually worked surprisingly well. We reinforced the large gears slightly by connecting the teeth on the inner edge before printing and I still have all four gears. We only ran these printed modules on a small practice chassis but they held up extremely well with almost no wear.


Under what conditions was this chassis run? Did it encounter any pushing, drops, etc?

1 Like

Imma just take that thank you…

1 Like

We didn’t do much dropping with it, we were trying to avoid breaking the mounting plates because those would not have held up to a big drop It was mainly a test bed for the software team to play with. If we had dropped it from a significant height directly onto the modules we probably would’ve broken the plates. That being said, I’ll make a bold statement and say that if you made the main plate out of metal (and probably the bevel gears) you could run a full size FRC robot with the rest 3D printed.

Hi is their code also written for swerve drive or is there a link to either learn about the implementation for swerve drive or even maybe get an idea of approach thanks

1 Like

Check this out: paper: Stryke Force 2017 swerve

And this is 2910’s 2018 code

That’s super cool! Could be a cheap way to build a test mule to learn swerve programming before dropping the big dollars on comp-worthy parts.

Definitely, that was our thinking as well. Still cost a good bit but was definitely less of an investment to get started.

The pla face plate on the cim will likely warp a little if CIMs warm up much at all, probably not as bad as ours did. The plate like you said and maybe where the wheels mate with the fork our the most vuneeable and could break easily enough with drops or hard hits on floor elements, like the cargo hold boundary bar.

We ran nylon plates on our 2018 modules. The main plate that the module bearing mounted to was a 3/8" thick nylon plate that was admittedly conservatively lightened, but we never were remotely worried about breaking it. I guess unsure if it would have survived dropping off of HAB2 this year, but my guess the plate itself would not be the fail point, but either the wheel forks or the module bearing itself.


How long did it take for your CR 10 to print one of those modules? It looks awesome!

This could happen. We used Neos on our full size robot and they never heated up significantly. I think we did crack a motor plate on the test chassis once because when we accelerated the weight of the cim flexed the plate a little far.

I kinda forgot about this. The edges transferring the force from the wheel/pulley to the bearing/plate are pretty small; it’s probably pushing it for 3D printed parts. (Though with some CAD time it could probably be modified fairly easily)

Thanks! Looking back at our printing spreadsheet and adding up the print times it looks like it evened out to about 50 hours of print time per module. However, we printed these at 100% infill so it would not be hard to get that time down considerably.

I guess the big question is do you need the MF for that part - 323’s swerve last year used printed pulleys out of 910 Nylon with no noticeable issues.

I have a hunch a non MF printer could churn through these parts faster and free up print time for parts that require it.

I guess with a markforged, the wall thickness and added continuous fiber will cut this down to 35 hours right. This is an insane feat. 4 years ago I bet no one would have thought about 3d printing swerve.


Our CF20 PETG ring gears held up considerably well. But before that we printed these in PLA on the prototypes, and I sort of felt they would maybe held up, the main concern being the gear teeth wearing more. We had on of the CFs fail on an edge impact before we thickened it a bit. All our PLA pulleys held up (steering, elevator). We only prototyped belt pulleys, but I think they would have worked well enough. The PLA issues we had were thermal-induced warping and impact related damage.

So yeah, I think you could make a good amount of parts on a non-MF. I’d trade some time unbalanced time for MF access.


@PatrickW Is the encoder assembly, the gear and the gear/attach to the machined pulley part 3D printed?

Yes, The 2 gears for the steering encoder are 3D printed. I am currently printing them from Polymarker PolyMax PC.

@PatrickW we just ordered some of the module bearings from your site. Have you tried sourcing a dimensionally equal bearing that is shielded or sealed? I’m glad to reduce the weight of the latest version of our module by using your bearings over the ones we used previously (6816-2RS) but I think it would be good if the bearing was shielded or sealed as it’s relatively close to the ground and exposed. The dust cover helps but I would prefer something more robust. Just curious if this is something you have looked into.


On the 2 fastest drive ratios, the first stage uses what I assume to be an 18 tooth cim pinion. Where do you get these? Or are you using 18 tooth 1/2 hex gears with a cim adapter?

1 Like

I have not tried to source a shielded or sealed version of the bearing yet. It is good to know that there is interest!

On a related note, 2910 drove the mini swerve robot on a grass/dirt field for hours at the FIRST Washington Seafair booth last summer. Definitely not the team’s best decision.

The robot remained functional the whole time and was completely disassembled for cleaning after the event. The big steering bearings weren’t turning as smoothly as they originally did, but luckily we were able to completely revive them by flushing all the dirt out with WD-40 and re-oiling.