2834 Dual Wheel Swerve

At the beginning of the 2019 season my team wanted to use a drivetrain with high mobility opting to do swerve. However, with no previous experience with this drivetrain we decided to do Vex’s VersaDrop instead, which ended up causing many issues. Due to these issues and lack of mobility of the octocanum drain train relative to swerve, I began designing my own swerve module, hopefully for future use.

Design Overview:
The overall goals of this swerve design were to be light, small, modular, simple/easy to build, and well-priced. I feel that this design accomplishes these goals but there is still room for improvement.

Unlike many other swerve designs that use large thin-section bearings my design uses a bushing method very similar to 2067 design. The rest of the module is designed around this bushing. A 38T pulley connected with a 1:1 ratio to a versaplanetary articulates the module and a NEO with a 4.71: 1 reduction drives the two 3” wheels at 15.76 ft/s (free speed). The entire design is quite modular so if a larger or smaller reduction is needed, the drive motor assembly can be easily modified.

Most of the custom parts are to be 3D printed out of either PETG or Alloy 910. The other custom aluminum parts, specifically the c-channel mounting plate and the swerve saddle legs, will be CNCed on our router. For the modules we made this summer the bushing was made using a manual lathe but once our Haas VF2 comes in, it is likely we will mill them instead.

Furthermore, the modules we build this off season were based on version 75 of the design which utilized ¼” thick 4” wide rectangular tube stock for the saddle. The CAD linked is version 81 which replaced the tube stock with a 3D printed azimuth and two connecting aluminum “legs”. This was done to reduce the weight of the module and to reduce its swept area for better integration into the chassis. Additionally, if this design were to be used in a competition season, we would likely use a VP lite with a banebots 550 instead of a 775 to save more weight.

Cost Per Module: $247
Speed: 15.76 ft/s free speed 12.77 ft/s adjusted speed
Size: 3.5x6.5”, 5.48” tall, and 3.43” of ground clearance
Weight of V75: 4.9 lbs
Weight of V81: 4.4 lbs (estimated)

Future Changes:
There are only three main things (as of right now) that I would like to change about this design: the height of the module, the amount of ground clearance, and switching the steering motor to a NEO. With the height of module being 4.6” above the 2x1 it mounts to, there will need to be something preventing the motor/VP from begin hit. The other two changes are not too much of a concern, but I feel that if would possible help improve the design.

Side Note:
During my redesign of the module (75->81) I was concerned that the 3D printed azimuth would not hold up, so I ran a linear static simulation of the module. Under a robot that weighs 200 lbs, the max deformation is less than .009 mm with a high factor of safety. If you would like to see the exaggerated results you can find it here.

Link to CAD: https://grabcad.com/library/2834-swerve-module-1

Special thanks to 910, 2067, and 2767 for sharing their designs and answering many of my questions!

If you have any questions, comments, or thoughts I would love to hear them!


Well done! I’d love to see how this looks in a full chassis. Not quite seeing how you’d link this module to the frame/other modules.

Great work!!

On the side of the C-Channel there are 4 bolt holes that will mount it to the chassis. Heres an image that should explain it a bit better:


Wow! The module looks so small in the photo, that’s amazing!

I have a few questions:

  1. While the bevel gears look very similar to the ones from SDS, they look 3D printed. Are they? Or are they the SDS ones but anodized?
  2. Are the wheels custom made? If so, from what material? Are they 3D printed?
  3. Could you please post a section view of the module’s bushing and bearing configuration and of the NEO reduction stage?

Thank you very much! Again, while I don’t have a whole lot of experience in this area (and am still learning myself), the module looks very cool!

  1. The image of the built module does have a 3D printed bevel gear that is only because the ones we ordered from KHK hadn’t come yet at that time.

  2. Yes the wheels are custom. They are printed out of PETG with 1.2mm wall thickness and 15% infill.


Just curious because we’re dabbling with the idea of printing some test modules for seeing our models in real life and to test some software - how did the printed gears old up in practice?

And again, thank you!

We tried several different materials: PLA, PETG, and Alloy 910. The PETG bevel gears were the first ones to fail. The PLA ones are still running. I just recently printed a set out of annealed alloy 910 with reinforced teeth that we have yet to try but they seem promising. I don’t think printed bevel gears will hold up well over a competition season, but I could be wrong since we have done very little testing.

Oh no, I was just asking for testing the modules in our shop (if and when we decide to use them during comp. season we’ll probably buy ones from KHK or something similar). Thanks!

For just testing they work fine as long as you drive conservatively.

What were your reasons for opting for a two wheel module as opposed to a single wheel module?

1 Like

Great question!

With this bushing design, the driveshaft has to be directly in the center, so to have only one wheel the module would have to have the bevel gears above the wheel like 2067’s design. Alternatively, you could essentially take out one of the two wheels I have in my design, but the wheel would be rotating off-center which is not ideal.

I miss-read your question.

A single wheel design (as far as I can think of ) that isn’t very tall would require a large thin contact bearing. We decided to not use a thin-contact bearing because there are not many suppliers that offer them at a decent price and have short lead-times.

Yeah, if you wanted to get them from SDS it runs you $65 per bearing. From what I have heard, their lead times are pretty resonable.

1 Like

There are a few vendors, such as SDS, where they offer the bearings at a good price with reasonable lead times. However if the vender were using sells out of the bearings we would have either find a new supplier that sells the same exact bearing or redesign the module around a new bearing, which is the last thing we want to happen at the beginning of the season.

Yeah, for me personally if there’s a chance I’m going to do swerve in season I would already have the bearings I need before the season starts on hand, but I do understand your worry.

1 Like

Another question - what material are you using for the bushing?

It’s made out of aluminum round stock we get from mcmaster-carr. Here’s the product number: 8974k75

Many swerve drive designs use the metric 68XX (typically 6816 or 6818) bearings, which are a standard part and can be found cheaply through Chinese suppliers. It’s also possible to purchase thin section bearings identical to the SDS part from Alibaba for around $30 the last time I checked.

The vertical shaft can be 8mm (steel) with no steps. The thrust from the bevel goes into the lower bearing, and you don’t need any thrust retention in the other direction. This might give reservations to some but we don’t run any thrust accommodations in the downward direction.

1 Like

For your modules is your 8mm drive shaft keyed? If not, how do you transfer the torque from the shaft to the bevel gear and your pulley to the shaft?