Third Coast CVT Swerve Module

So over the past two months I’ve been designing a swerve drive for our team that combines Team 2767’s Third Coast Swerve Module and Team 1640’s CVT Swerve Module. Another aim of the design was to minimize the number of customized parts required.

Some Helpful Resources
Third Coast Swerve Design Description
Team 1640’s Swerve Central

Most of the main module is similar to the Third Coast Module, with the exception of the azimuth pulley, which is now a Vex VersaPulley in the 9mm configuration.

Where the drive shaft is driven in the Third Coast drive, there is a slightly modified version of 1640’s CVT mechanism. One of the main differences is that the flanges that push the V-Belt together are not 3D printed, but are Flat Belt Idler pulleys from McMaster. The Fractional Horsepower Pulley is also a slightly different model from Torque Transmission.

Both the Azimuth and the Drive are powered by NEO motors, although an adapter is required for the drive.

A total of 10 parts need to be manufactured/machined:

  • Saddle
  • Pivot Hub
  • Mounting Hub
  • 2x Tensioning Arms
  • Tensioning Rod
  • Servo Mount
  • 2x tensioning Pulley Shafts
  • NEO Shaft Adapter

CAD files on GrabCAD

Any and all feedback is welcome! This was my first major design project, so feel free to suggest any changes, ask questions, or anything else really. If you have any feedback on how I can improve my CADing skills that’d be great too, since I taught myself and I’ve definitely missed something.

Here’s some pictures of the models (Please excuse the NEO wires):

Side View

CVT Mechanism


1 Like

Correct me if I’m wrong, but are you direct driving the module’s rotation off the NEO? I feel like there might be some issues with that.


^I think so.

This isn’t a good idea. Ultimately, you want your azimuth to be 200-400 rpm. If not, you will draw too much amps, and it will be hard to control.

1 Like

So what motor could I use? Or maybe a gearbox?

You could always try doing a reduction with the pulleys themselves (have one larger than the other) rather than having them in a 1:1 like they are currently.

Alternatively, if you have space for it, you might also be able to swap out the pulleys for a simple gear reduction instead.

unfortunately, I don’t think this will work. the gear ratio needed is around a 15 - 20:1 so you can’t get it out of a singular stage. your options are to either use a vp planetary, or two stages of gears and/or belts. if you use a vp planetary, I suggest you use a different motor, like a 775.

1 Like

RS550, 100:1 VP, with a VP encoder on the last stage.
BAG motor, 70:1 with a VP encoder on the last stage.


To add to this, you want an encoder on the last stage (and no reduction after) because that getting a 1:1 module rotation sensor is almost essential for swerve. You can do it with a relative (quadrature) encoder and a zeroing hall effect sensor or limit switch, but many agree that it’s simply not worth it.

Quick question: where might I get CAD files for the VP planetary gearbox? The website only has a singe stage for the 775pro and I can’t seem to find the start and end stages there.

You could get pretty close with 32dp gears. A 12t and a 100t gear would give you around 600rpm, which while still a bit overkill, would be a lot closer than the direct drive. You could potentially find 32dp gears larger or smaller than what WCP/Vex sells that might make this better too.

You should be able to find them in the CAD tab at the bottom of this page in the “Example v2 Gearboxes” section:

If it doesn’t have the exact motor you’re looking for, you may have to swap it manually.

Ah. Thanks!

So on looking around for solutions, I think I might go for a 775pro with a 63:1 VersaPlanetary gearbox, which would be somewhere in the 360RPM range theoretically. Any other suggestions?

The only other concern I have looking at the model is the rigidity of the mounting bracket. Obviously this could just be a placeholder, so I don’t want to make any assumptions, but if your intention was to use what appears to be a relatively thin piece of aluminum to mount everything to, you might want to consider either trying to shrink the packaging of the module (thus reducing any torque between the components), or maybe looking at a more stable mounting solution.

Granted, this is just based on the visual appearance of the model, so it may actually be a lot more ridged then it looks, but that’s my impression based on the images at least.

On a side note, what are you expecting the reduction range/spread of the CVT to be? I’ve been interested in playing around with some CVT concepts but I wasn’t sure if it really made sense if the range of available reductions is less than a COTS gearbox.

1 Like

This question might be more directed at 1640 since this is their CVT design, but I am curious why a rotary servo is used for changing the belt ratio? It seems to me that a small linear servo could be more robustly mounted and be more compact than the rotary servo mounted on top of the unit.

While I was not around for the design of this iteration of the CVT with the flying arms connected to a servo I can shed some light on it. It was designed to fit onto 1640’s existing swerve module with no modifications. That meant some restrictions and compromises were made so that it was a drop-in design on the non-CVT swerve module of the previous year. That led to the design of the 3d printed servo arms but you’re probably right a linear servo could accomplish a similar motion.

This summer we’re prototyping a new CVT design where the whole module is built around the CVT. Lots more details are to come soon but it involves moving the CVT drive pulley with a servo to change its C-C distance to the next pulley so create the tension for the CVT. A larger C-C distance creates more tension to force the belt further into the CVT pulley, smaller C-C creates less tension allowing the belt to move onto the outer edges of the CVT pulley.

I forget the difference in pitch diameter on our CVT pulley but on our new prototype this summer the range of speeds measured with a tachometer were ~7fps-14fps so you can get a sense of the ratio from that.

I see so many designs recently wanting to integrate our CVT which is great, they just need to wait till we have our new and improved CVT out this year!
@Gdeaver can talk more about the design of the old CVT with the servo arms.


You’re correct; I do need to make the mounting bracket more rigid. I just put it in there to see the clearance with the frame. I’ll probably have to thicken it, and I’m going to move the components closer together.

As for the CVT range, going off of the specifications of the variable speed pulley, it can go from 1.5:1 to 3:1.

Could you elaborate on how he linear servo would work? I’ve never worked with those before.

So stupid me didn’t think this through. I did some mental math and it turns out that my speeding up of the NEO drive brings the theoretical speed to 90 fps, or about 61mph. Now I have to go back and change that!

1 Like

I have a couple of questions about your swerve modules:

  1. How are you driving the VPS-15 pulley from the CIM? Is it a custom adapter?

  2. How much does each module weigh? Couldn’t find it on swerve central.

To be fair, I’ve never used one, so I’m not sure how well it would work, but my thinking was to use something like this. You could mount it perpendicular to the belt run either on top or bottom, with one pulley fixed and the other mounted to the end of the servo. There might be a reason the rotary servo works better though. You would have to check the forces involved.