33 Killer Bees 2019 Winter Swerve

(Falcon bodies hidden in section view, as I have some error issues with the files when trying to do a section view currently)


We wanted to get the latest iteration of our swerve module published just before the 2020 kickoff.

We posted this back in May. We ended up making some of the parts for this and got partially through assembly, but then got busy with other things in the summer. A couple other teams have actually built a full chassis of this design, so we were able to get feedback on its performance and issues with the design.

We ended up making some iterations to the design at the end of the summer/beginning of the fall, and ended up with a design we felt was superior in a number of ways. We went for a narrower orientation then the May version, and while we sacrificed some of the height that we saved on the previous version, we were able to keep it below where the top of the bumper would be. This module with NEOs ended up being 4.5Wx7.5Lx7H and weighing 4lb 0.75oz.

We ended up building this module and have an entire chassis assembled and working as our programming development chassis. Overall we were very happy with this design. We spent some time dialing in a tread template that we made to make tread that easily fits the wheel, but otherwise we have not had any issues mechanically at all. This design now allows for us to easily have a frame gap if needed, and a standard 28x28 frame weighs under 20lbs.

We ended up at the final revision that is shown at the top of this post after seeing some of the concepts that 2471 implemented on their latest module, specifically the smaller 3" ID 3.5" OD bearing. This allowed for even tighter packaging. This latest version uses the same gear stages for the drive that the one we built up uses, while we actually ended up getting a little more reduction on the steering. Below you can see a comparison of the improvements made from the fall version to the winter version.


Bevel stage is a 20:40T 1.5 modulus bevel pair, purchased from KHK. The 20T is broached to 3/8" hex bore, while the 40T has the bore opened to 0.874" and a 1.25" bolt circle added that are tapped to #10-32 to allow the wheel to be bolted to the bevel gear.

Spur gear stage from the motor to the bevel shaft has options depending on what size pinion is desired to result in different final free speeds:

08:30 10.81fps
09:30 11.14fps
10:28 13.27fps
11:28 14.59fps
12:28 15.92fps
13:26 18.57fps
14:26 20.01fps

This gear setup allows us to have great flexibility on what drive speed we can use depending on the season’s game task, rather then locking us into a narrow window of speeds. This was the driving factor in choosing to use the bevel pair we ended up with.

The wheel is a 3" custom printed wheel with blue nitrile tread mounted using #10-32 bolts threading into threaded inserts that are pressed into the printed hub. 3/8" round bearings are pressed into the wheel on one side and the 40T bevel gear on the other to allow for a dead axle setup, which makes removing the wheel a one-bolt operation.

There is an 8mm bearing on the module for the end of the motor shaft to be supported by. If using a Falcon, we have an adapter we can print to go from the spline profile of the motor shaft to the 8mm bearing, or the TTB 8mm shaft is also an option for this.


The first stage is an 8:30T spur gear reduction off the motor, followed by a 20:86T GT2 115T belt reduction to the module turret. If a falcon is not used, we would be comfortable with using a 9T gear on the first stage instead.

The intermediate shaft for the steering reduction is turned down to 3/8" round on the ends of the shaft, and a bore in the main module plates act as bushings for this shaft to spin in, eliminating the need for additional parts and weight.

This latest iteration of our design also worked to integrate the 221 Lamprey encoder into the design. We are big fans of using this for the sensing the position of the module, so that if we have to pull a turret for maintenance, we do not have to worry about re-zeroing the sensor.


4" Width x 6.25" Length x ~8" Height (With Falcons, ~7" Height with NEOs)
4.3lbs projected weight with Falcons
3.9lbs projected weight with NEOs (not including SPARK Max)

This weight projection should be pretty accurate, as the CAD of our previous module was actually over what the actual weight ended up being by .2lbs. Many of the parts are similar on this model, so the actual weight of this module may end up being lighter then what I listed above.


For mounting to the frame, we have 2x1 box tubing between the two main plates. The thunderhex standoffs go through holes cut in the tubes to locate and mount the modules to the frame. We have found that a 0.377" hex profile with 0.063" radii is a perfect slip fit for 3/8" thunderhex to go to. This gives us a robust way to mount the module, while also making it easy to assemble and mount to the frame, as well as take apart when/if needed.

This was all the interesting points I could think of, let me know if you have more questions about anything, and good luck to everyone in the 2020 season!


Can you elaborate on how the modules mount to the 2x1 tubing? How do they remove? Slide sideways away from the tubing and then up/down?

1 Like

Here is another pic that shows this a little better:

So the thunderhex shaft goes through both sides of the 2x1 tubing. This means when you screw into the standoffs to mount the plates to the frame, the standoffs cannot rotate because of the hex profile machined into the tubes for the holes.

There are trapezoidal shaped cutouts in the frame tubes to fit the module. This was done to get the wheel as close to the edges of the frame as possible (the center of the wheel is 2" from each frame edge). Maximizing wheel footprint was an important point for us. We are able to cut most of these cutouts easily on our Omio, and then finish the cut with a bandsaw.

If you had to replace an entire module (should not happen), you would remove the 6 screws from the top plate and the entire module would come off from the top and bottom, with the thunderhex standoffs sliding out of the tubes with the bottom plate. You could also do the reverse, but then the turret would also become off the bottom plate.

Generally we have found we never ever have a need to pull an entire module. This design has made it really easy to service (1 screw pulls the wheel, 4 screws pulls the entire turret pod).


Maybe I’m missing it but I don’t see the belt. It looks geared?

Looks like the falcon one is belted and neo one is geared? The first render has a belt.

The latest version pictured at the top of the post, and showed again at the end of the post uses the belt. The iteration we designed and build in the fall had a gear stage instead.

The packaging utilizing the belt stage ended up packaging better on this latest version. Teams like 27 and 3539 who built our first 2019 iteration that we posted back in May proved to us that the GT2 stages work just fine for steering, so we are confident this latest design would work great.

1 Like

That’s a nice design.

From the title of your thread and knowing you are from Michigan, I was expecting studded tires :wink:

Where are you guys sourcing the 3.5x3 bearing from? WCP and Thriftybot all seem to have the larger one as the only option.

You can order these from Alibaba suppliers for $60 for 4, shipped via DHL. They’re quite cheap. I don’t know where 33 gets theirs but I’ve bought this way before.

ThriftyBot also plans on carrying this bearing, they will have it in stock at some point soon.

Also be sure to check out the newest iteration of this design here.

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.