4926 Offseason Swerve Drive

This off season, my team is pursuing a swerve drive and this is the design currently proposed. Since this is our first time ever attempting swerve, my goal with this design was to make the module robust, simple, and easy to manufacture/maintain. I took inspiration from many other teams’ swerve modules including 2910, 1640, and 118 Revolution Pro. Even though my design looks very similar to 118’s, I actually was almost finished with the CAD when I saw their design. Great minds think alike I guess!

Following my design goals, I decided to go with the simple, “classic” swerve over bevel beside wheel and other more efficient variants. It is geared for 14.6 fps but can be configured to 17 fps or 12 fps as well. The weight, according to Inventor, will be about 7 pounds per module.

Some unique things about the design:

  • First stage drive reduction belt is tensioned by sliding the NEO along slots and tightening the motor screws. There is a picture attached of the slots.
  • Designed to include shims in shafts where tolerances might be an issue. Some of the parts will need to made on our new CNC router and we are still figuring out how to optimize the machine so I wanted some way to make sure everything will fit together well regardless of what kind of tolerances we ended up with.
  • Going with the above point, all the parts are either COTS or simple routing operations on a CNC router (with the exception of one 3D printed spacer)

The only issue I currently have with the module is how tall it is. However, that is a trade off I am willing to make with how simple the module is.

What problems or concerns do you guys see in the design? Please be critical. Any problems I can catch at this stage will be one less headache in the future.

Link to GrabCad:


Perhaps you could flip the motors upside down and have them underneath the module instead of on top. You could also swap the steering motor to a NEO to cut out some motor height and gearbox stages. This change might not be suitable for games like 2018 and 2019 with mild terrain, but would be fine for flat fields like 2014 and 2017.


First of all, thanks for the feedback!

I was thinking of flipping the drive motor as well to lower the CG and make it easier to swap the motor if necessary. However, the swivel pod runs into the motor if I simply flip it with the current design. I do like that idea though and I think I can make the center to center distance from the motor to the swivel pod greater with a bigger belt so I can flip the motor and not have that issue.

I chose to use the BAG motor mainly to save money since we already had some in stock. However, I’ll look into switching to NEOs since that would cut an extra stage off the gearbox, which would also save money and weight.

1 Like

This looks similar to a lot of swerves I have seen in the past. I have a couple questions for you.

  1. What drove the reliance on the Vex bevel miters? Using the SDS bevel set, you could cut down on height
  2. Why so many shaft collars? A 1/4-20 bolt and washer work well and are less prone to failure.
  3. It looks like the rotating pod may rotate into a frame. Is there enough clearance?
  4. Why gears for the final azimuth stage?
  5. Why use the shoulder bolt on the bottom in lieu of tapped thunderhex? It’ll work, but the thunderhex will be cheaper and more readily available in an FRC setting.
1 Like

First of all, thank you for the feedback!

  1. The main reason for using the vex gears was that we wouldn’t have to hex broach them. Our team doesn’t have those capabilities in-house and I didn’t want to rely on outside parties for something so important to the swerve.
  2. That’s a good point! I’ll change that in the next iteration.
  3. The pod is designed to not run into the plane made by the sides of the square tube chassis. Since the module mounts onto the bottom of the square tube, it won’t run into it.
  4. Last year we ran belts on our WCD and we had one of them tear up during elims. Luckily it wasn’t too hard to replace but on a swerve module, it would be impossible to replace a belt on the inside without taking the module apart and I didnt want the pulleys on the outside of the module because I wanted a dead axle for the wheel. In then end, I found it easier just to use gears since the center to center distances worked out fine for the gear ratio I wanted.
  5. Wow I didn’t think of that! Thanks!

This company sells beveled gears and x-contact bearings. If you want to go through the headache, you can get the bearings cheaper on Alibaba by RFQing. I’d recommend these beveled gears and that bearing for turning instead of the current setup. What’s driving you to pioneer your own swerve vs. just using a COTS module?

A belt in the azimuth of a swerve is a far different application than the side of a west coast. Don’t let that one failure scare you away from belts, they are lighter and cheaper and an excellent way to transfer power.

Read through this one. https://www.chiefdelphi.com/uploads/default/original/3X/4/e/4ed7bb78a109bbcb9e93d01df5f3258228d26e93.pdf

What 3D printing do you have access to?


When you buy a solution, you don’t get the learning experience of designing. Our team is redesigning our CVT swerve and the students are gaining fantastic real world experience of design. And then there is the " How do you actually make it" experience.

1 Like

Another useful way to capture shafts is to use a flanged bearing and turn down the end of the hex shaft. I think this would be appropriate on the bevel gear shaft, probably not the wheel shaft.

I am interested in what the specific gear selections are, if you don’t mind.

Any possibility seeing your work in the fall at RAGE or CAGE???

This is is exactly what we do on our design for the shaft that has the bevel gear. It works great. We also add a length of teflon tubing that is a tight fit to the shaft to act as a spacer between the bevel gear and the spur gear. This keeps the spur gear in place without significantly reducing the clearance between the wheel and the shaft the way that a collar would. Helps to keep the module compact.

For the wheel shaft we just use a button head bolt on each end to capture the shaft between the two bearings (the grabcad model shows a long through-bolt in this location, but the actual modules just tap the ends of the shaft and install individual bolts in each end).

1 Like

Peter, I don’t know yet. We may build this single module as a control test bed and if there is success and continued interest maybe! We’ll see. Thanks for commenting!

The first stage belt reduction is 12:30 and the second stage gear reduction inside the module is 28:66. The azimuth gears are 1:1 since that keeps the vex integrated encoder 1:1 as well which I have heard is very useful. All the reduction for the azimuth motor will be in the versaplanatery stages before the encoder.

Team 4926 has a Onyx One so we are pretty set there. We agree with @Gdeaver that the design experience and transition through to “making it work” is invaluable. If we were committed to using it for a competition season maybe we would go COTS, but since we are not, we are exploring this route.

Thank you for your comments to our student designer on this!

I think I misunderstood you when you asked why I used belts for the azimuth stage. I thought u meant the drive but I looked up what azimuth meant and realized you meant the turning of the module. My bad!

The main reason for gears was just that it was easy to find one (I guess that’s not a good reason). I need whatever I use to transmit power to have a large surface area so side loads on the swivel pod won’t bend the coaxial shaft. The larger surface area in contact with the mounting plates helps keep that from happening. That’s also why I have the large thrust bearing. However, I’ll look into how I can integrate belts into that stage instead. I have also heard belts have less backlash which is important in swerve.

I’ll look into those bearings as well. Thanks!

Excited to see you guys with are going for a swerve, can’t wait to see it.

1 Like

Wonderful Design, I love how simple it is. One question I have is, I don’t see an absolute encoder anywhere, would your plan be to just zero the modules every match? If so, I highly recommend using the absolute encoders, that way it takes away from forgetting about zeroing them every match.

Thanks! The vex integrated encoder is attached to the versaplanatery gearbox’s final output. It has both absolute and relative encoder settings and it is geared 1:1 with the module.

Also, to be clear, I was not asking why you were designing your own to discourage you. Designing a swerve is an excellent design challenge, and one I went through as a student not long ago. Keep working at it. I do encourage you to use the Onyx One more. It’s powerful.

Could you post a sectional analysis view of the drive shaft/what is supporting it? I’m concerned that the drive shaft could bend if the module hits something.

The drive shaft is just 1/2 inch thunder hex but there is a very large thrust bearing in contact with the mounting plate. The idea is that a side load on the module won’t bend the shaft because the thrust bearing can’t go through the mounting plate.

Here’s a little diagram in case my explanation didn’t make sense. I plan on making a career out of my graphic design skills :laughing:

I’m looking into using belts to control the module’s azimuth and my main question is: Would this kind of application of belts require a tensioning system? Normally in WCDs, belt tensioning is important but I am not sure if this would be necessary for rotating a module.