pic: New COTS Swerve Module



This is the production version of the 2910 MK1 module. Many small improvements and simplifications have been made.

I’m excited to announce that I am selling kits and parts for this module through the website swervedrivespecialties.com. If there is the demand, I plan to stock the inventory to qualify as a legitimate COTS Vendor for these parts.

More details and the CAD model can be found on the product page.

Please feel free to ask any questions you have about the modules here.


Wow, this is quite interesting.

I presume that you’re going to have the end user be responsible for the programming of it? Or do you plan to include sample code.


I’ve been predicting that more COTS swerve drives will start showing up on the market (VersaSwerve™ 2018 Winter release? Big if true). Anyway this is a neat option, and I can’t wait to see 2910’s swerve up close at CC this weekend!


I will not be supporting the programming. It’s not my area of expertise. I do know that 2910 already has a swerve Java library that is up and running well and works great with these modules. This will be made public before build season. Our code that was made public from the 2017 off season should run these modules great as well. It’s just unpolished in comparison.


I’ll be at Chezy Champs as well. I’m 2910’s drive coach. I’ll bring one of these new modules for anyone to check out.


Wow, very nice! I was a fan of your team’s swerve when I inspected you in Houston last year. I’m glad now other teams may get the experience of using it.

That being said, the price is a bit steep. I figure with all of the add-ons the total price comes out to about $650, compared to about $450 total for a Swerve and Steer. I know expense is something that comes with the territory of swerve, but I imagine the type of team that would be buying this is also the type of team that would be interested in saving $200/module. Can you do a bit of self-promotion as to why they should buy yours instead? (I have some thoughts, but I’d like to hear yours)


How did you like using those wheels for your module in 2018? Did you try any other wheel types?

I’m on my phone so I cannot open the cad right now; is there a bearing supporting the shaft underneath the 15T bevel gear?

I’m so so so happy that you will be selling the bevel gear sets separately. Any idea when those would be in stock? Where are you sourcing them from?


The wheels worked quite well. The only thing was the wear. We replaced them every competition so that we didn’t need to compensate for the change in diameter for auto driving distances. If we were compensating for the diameter we could have easily gotten 2 events (probably 3) out of each set of wheels.

There are 2 bearings below the small bevel gear. Same as the modules 2910 ran last year.

The bevel gears are genuine KHK stock gears with secondary machining. I’ll have some in stock for sure in 3 weeks, but probably a little sooner.


For those reading that haven’t visited the website, the base price of the kit is $425. This price doesn’t include some of the COTS parts that your team might already have.

A lot of the higher price difference has to do with higher cost components. The main rotation bearing and bevel gears in this module are inherently high cost components. There are also quite a few custom machined parts in this design which are not inexpensive.

I will say that I am confident this module is quite robust. It is very similar to the design 2910 ran last competition season which hasn’t had a single hardware failure yet. They are still going strong after 85 matches. You can expect a very high level of build quality from these parts.

I have not used the Andy Mark Modules before, so I do not want to comment on how they work or hold up. I also would hate to put someone else’s product down. I think it is an impressive feat that they are able to sell a swerve module for $325. That being said, some objective advantages to this module are the lower weight, shorter height, and larger more robust bevel gears.

Ultimately, I think there is a place on the market for both this module and the AM module at two different price points.


Glad to hear it! I do strategy for 3512, but I checked the match schedule and we don’t have any matches with or against you guys unfortunately!

In any case, have a safe trip and I hope y’all enjoy your time in Cali!



This module is beautiful. Had the opportunity to hold one in my hands in Houston this year and was really impressed with the clever leveraging of some manufacturing tricks on the module, such as how the final steering pulley is made. I like the DeCIMate configuration that fits within this module too.

Good luck with sales!



Could you reduce the machining on the main plate if the steering pulley support bearing hole was all the way through instead of a blind hole? That way all the milling or routing would be from one side and you don’t need to flip the plate to machine from the other side.


Yes, you could certainly do that. I choose to leave the material below that bearing because there is already pretty minimal material between the steering support bearing and the main rotation bearing.

For looks and the perceived quality of the part, I want it be machined on all surfaces and have a slight chamfer on as many edges and holes as possible for deburring. This makes the part need to be 2 sided machined already.


A few people have asked about the tooth profile on the base pulley part. The belt is HTD-5M profile. The tooth profile on the pulley only has a full tooth every 3rd tooth. I designed this profile to make the part faster to machine. I can use a 3/16" end mill to cut the tooth profile. This profile has no negatively effect the operation of the module. It is such a large pulley and has over 180° of wrap so it works perfectly fine. The modules 2910 ran all last season used this same profile as well.

This photo shows a belt engaging with the toothed pulley.


I was really impressed by this clever trick when I saw the modules in-person at Chezy Champs. Well done!



Perhaps I’m missing something here, but to me this looks like a pound of prevention for an ounce of cure. How many of these large steering pulleys did you make? If the number is small enough that they could all fit in a cigar box, how does the savings in machine time justify the engineering trade off calculation and design times?


To machine the true tooth profile you can at the most use a 7/64" diameter bit. The stick out required to machine the tooth profile is about 0.82". This ends up being a stick out to diameter ratio of about 7.5:1. Tool deflection becomes a real issue and it is quite difficult to avoid chatter while taking a decent cut. With a 3/16" tool, the stick out to diameter ratio is about 4.4:1 This is much more manageable. I’ve machined about 30 parts with this tooth profile if you include the ones for 2910’s off season bot, comp bot, spare modules, as well as the inventory I’ve built up for sales, so it has been worth the time savings.


What was your reasoning for having two different main plates, RF, LR and LF, RR? Would it be possible to just have all of them be the same? That way any module can be used on any corner.


The footprint of the module is not symmetric across the diagonal. Layout Drawing The main plate is 1" longer in one dimension then the other. This is mainly because the versaplanetary for steering is off to one of the sides.

The two different versions are essentially mirror images of each other. You could use the same module on all four corners of the robot. It would work totally fine, but it would be kind of strange because it wouldn’t be symmetric across the center line of the robot.


In your drawing, the center of the module wheel is the same X and Y distance from the corner, so although the modules wouldn’t be symmetric, the kinematics would remain the same. Is this mostly a aesthetic choice then?