3863 Pantherbotics 2019 Offseason Swerve Release

Team 3863 Pantherbotics is proud to release the latest iteration of our swerve drive!

Compared to our previous offseason swerve, this design is intended to reduce the number of unique parts, minimize cost, and increase assembly speed. The top and bottom plates were machined on our CNC router using 2D operations only. We were able to do so by removing the bearing pocket on the bottom plate and switching to use bolt heads to retain the X-contact 6816 bearing. The forks were cut out initially on the router, then the holes on top were done on our mill.

Regarding the steering, we chose to avoid using a VersaPlanetary as we found it was significantly more expensive than using spur gears, belts, and pulleys. The first stage is a 100:12 using 32DP COTS gears, and the second stage is using a 5mm pitch HTD belt and 3D printed pulleys. Currently, the total steering ratio is around 30:1, which is definitely a little on the low side. We’ll have to do more testing to see if this ratio is sustainable. The 32DP first stage could be swapped for an equivalent C-C 20DP reduction, which would net us a higher ratio. We are using a TT-6127 absolute encoder with a separate encoder gearing off of the first stage to sense wheel orientation.

For the drive, we are using a 3D printed wheel. For our ‘proof of concept’ modules, we have been printing them out of PETG. They’ve been surprisingly durable! However, we plan to print the final modules out of Nylon 910. The printed wheel has six 0.201" holes in a 1.5" bolt circle to affix the bevel gear to the wheel. The total drive reduction is 7.50:1, netting a free speed of 13.21ft/s with a NEO. It can be geared faster, if necessary. We plan on using the NEO hall effect sensors for sensing wheel velocity, but we can also opt to use Munchskull’s CIMCoder if necessary.

Google Photos link to an assembled module, here.

GrabCAD link (includes both STEP and SolidWorks native files)

As always, don’t hesitate to ask questions!


Why not use a Neo for steering? You can get away with a lot less reduction with it. Other than that, reminds me a lot of 1323’s modules.


I would guess to make encoder integration easier, cause he can just plug encoders into a Talon and run PID loops on it. Plus, a NEO would still require 2 reduction stages.

Echoing what Anand said, the encoder integration with TalonSRXs is really nice – it’s hard to beat CTRE’s established library for ease of use! Furthermore, since we’ve already built up a substantial stock of Talons, it’s more economical for us to continue using 775pros. Once the brushless ecosystem settles, we’d take a closer look at changing our choice of motors.

Very clean design. Do you have an estimated cost and weight per module?

Also, do you use a mirrored version for the opposite corners or no?

Looking at the BOM, we’re looking at about $260 for everything except for the plate and the shafts. Including those would probably add another $20~30ish to the cost of the modules. Chinese-sourced bevel gears would lower the cost by around $30.

The module weight, with NEOs, is just about 5lbs according to SolidWork’s estimates. We don’t have our scale in the shop currently so I can’t weigh the assembled module, but that feels about right.

We do use a mirrored version for the opposite corners. Luckily, we don’t have any configuration-specific features on any part, so we can use the same plates for both modules.