2020 Summer? Swerve

This is the latest iteration of the swerve drive module that I have been working on over the past few years. For previous iterations, check out this thread.

CAD for this module can be found here.

Overall, this iteration continued to get smaller and lighter. Also, this iteration utilizes a Neo550 for the steering with a single stage UltraPlanetary gearbox on it. This helped further reduce the size and weight of the module.

Some overall specs for the module:

Full footprint from edge of frame: 4" Width x 5" Length x ~7.4" Height
Distance of wheel from edge of frame for wheel base: 2" from both sides of frame

Full weight: ~3.45lbs (SW Mass Properties, all parts either have assigned weight based on COTS reported weight, or correct material applied; printed parts have estimated weight from Eiger assigned)

Drive ratio is a 20:40 bevel gear ratio, first spur stage off the motor is 10/11/12:28. Optional 9:30, or 13/14:26 also available.

Drive Speeds (free speed):
12.16fps (09:30 first stage)
13.27fps (10:28 first stage)
14.59fps (11:28 first stage)
15.92fps (12:28 first stage)
18.57fps (13:26 first stage)
20.03fps (14:26 first stage)

Steering ratio is a 5:1 (5.23:1 actual) UltraPlanetary Gearbox, followed by a 16:86 GT2 belt stage. Overall ratio of 28.1:1, ~391rpm free speed.

Steering position is determined using a Lamprey Encoder mounted to the top of the pod. Local control to get to position can be done using the encoder on the n550.

Module is symmetrical, allowing for same parts to be made for left and right hand modules.

Module is also set up so that it can easily be mounted for either a standard frame or for a frame with a bumper/frame cutout.

I will have another version posted tomorrow that has the drive motor flipped and the steering motor lowered. This version will be ~5.5x5.5 while being only 6" tall so there is an option to fit under the bumper height. This version will be right at 4lbs.


Do you know what the total mass of all of the printed parts is? I’m curious if there’s really much more room for improvement in weight. Not that a fraction of a pound matters all that much, but it’s fun to speculate about the theoretical limit of swerve. This is a beautiful design! Thanks for posting.

1 Like

The following also shows how the module mounts to the frame.

The first picture shows how you would mount to a frame with a bumper/frame gap. This setup would require an additional 2" of frame support to be added to support the full 6" of the bumper corner.

The second picture shows how you would mount to a normal frame with a tube on either side of the corner of the module. The one standoff in the corner would be replaced with a 3d printed block that allows both tubes to bolt into it from the side while the top and bottom plates of the module can be bolted into it as well.

The hexagon holes that go through the tubes to allow the standoffs through the tube works really well. The standoffs are made with 3/8" thunderhex. The hex holes in the tube are machined on a router; they are 0.377" hexagons with 0.063" radii so that they can be machined with an 1/8" bit. This hexagon hole ends up being a perfect slip fit for a thunderhex standoff to slide through. This allows both the top and bottom plate to mount to the frame, which makes for a really rigid mount, and the standoffs going through the tube means you won’t be crushing the tube walls at all when you mount the modules, and also allows for easy removal.

1 Like

2020 should be called the ‘Summer of Swerve’


Do you need to mill a cutout in the tubes for the turning pulley?

This looks very nice. I like how you place importance on the footprint of the module. I look forward to seeing the next iteration that packages the functional parts within the bumper zone (5" tall x <6" wide x whatever long). That seems to be the bogey for swerve module design. If you can stay within those dimensions, you can be reasonably confident that it will usable in any game and you won’t have to design the “top end” around the space claim for the module.

It would be nice if the Falcon 500 offered a version without the integrated controller (or at least packaged it differently) so it was as short as a NEO.

Have you had success with the Lampreys yet?

Overall, the module looks really great! I like some of the new REV components that you have brought into the design. How much lighter is the Rev Ultraplanetary compared to a VP Lite gearbox? One of our FTC teams used the UltraPlanetary last year and it seemed fairly heavy, especially the output cartridge, but I have not actually looked at the weight compared to the VP Lite.

I’m assuming that the wheel size is 3"? How easily can this module be converted to handle a larger wheel if the game called for it?

I like the idea of the frame cutout in the corner to get the wheels as close to the corner as possible, but It raises some concerns about frame strength. These corner cutouts are going to weaken the ability of the frame to resist racking forces that can occur due to impacts. Unless you have other frame members or are using the belly pan to resist the racking of the frame, these corners concern me. It would be relatively easy to add “tabs” the the main plate of the module that run along the frame members to allow 2 or 3 bolts along the frame to allow an attachment to the frame that can resist the bending loads associated with racking.

Have you been using the lamprey with the SparkMax yet? We have been running that this season and still have a few bugs to work out. If you have been using it this year, I would love to understand how you have solved some of the issues that we are having. Overall, I think the Lamprey is going to be a great product for this application, but we need to get through the growing pains first.

Looks Great! And nice animation! I really like how you were able to integrate REV’s UltraPlanetary so you don’t have to use a large spur gear.

How are you fastening the tread to the wheel?

How are you estimating weights of the large 3D printed parts? Do you export and slice them, or just use a rule of thumb density?

Looks awesome! A few people on my team are curious what material the wheel is made out of? I’m guessing it is custom but could a COTS wheels be utilized in its place?

Another beauty!

I feel like a unibody sheetmetal chassis could be a really interesting alternative. The thin section of the angle concerns me. Alternatively… a frame shaped like a # hashtag? (Welded joint? Cross joint?)

Yes, we cut the trapezoid shape on our router, and then cut away the remaining material on the bandsaw. Doing this is easy as the router has already cut the angled portions of the trapezoid, so following that line to the rest of the edge of the tube on the bandsaw is easy.

! single stage UP weighs ~0.3lbs plus a few screws according to the REV website. The reason I did not use the VP Lite is that it is not as short in its packaging. A single stage VP Lite would weigh 0.4lbs.

It is a 3" wheel. I am looking at options for changing the fork to use a 4" wheel and adjusting the bevel gears as needed. However I do have a bolt-on skid design that can be mounted to the existing forks for going over things like the 2020 bump.

Take another look at my above post regarding the frame. In either configuration, each frame tube is mounted in multiple locations. The swerve chassis we built last offseason used a very similar frame mounting method and is very robust. The combination of the frame tubes extending to the end of the module while also being completely gusseted through the module plates make for a very rigid mount. The thunderhex standoffs going through the tube helps as well.

Also take a look at this thread, where you can see the same style frame was used on our offseason chassis. This has worked out great, we have no concerns about strength here.

We did some initial testing with the Lamprey before the 2020 season with the limited time we had them in hand and overall liked them. We have not fully integrated them into a swerve drive physically yet as we did not have the time before kickoff, and obviously have not been in the shop recently to do any development work.

There are #10-32 threaded inserts pressed into the wheel, and button head screws retain the tread to the wheel.

See the description in the initial post:

Due to the way the bevel gear packages very tightly in the pod with the wheel, it would be very difficult for any current wheel to satisfy the packaging requirements to be able to be used. Also there are very few 3" COTS wheels currently available.

1 Like

The picture that you show in this post highlights my area of concern. Three of the 4 sides of the rectangular tube section have been cut away leaving only the outer side. Thus the bending strength of the tube through that section that has been cut away is significantly weakened.

If the frame were subjected to a significant impact load that tried to rack the frame and there was no other resistance to this racking, I feel like the frame would bend in these thin section. The frame member would pivot around the thunderhex standoff at the corner of the module and bend in the thin section. If there were 2 connection points at the corners of the module that are further away from the corner of the robot, then I think everything would be fine.

You actually did this on one side of the module in that previous design. So that is good. However, looking at the chassis that you built for that offseason project, I think that the majority of the racking loads will be resisted by the T gussets between the side rails and the cross rail that is set back from the open end of the robot. That might be enough.

Of course if you use a belly pan to resist the frame racking mode, then you really don’t need much bending strength at the corners of the frame. So, the decision to cut out the frame members like this needs to be looked at in light of the overall load path strategy of the frame. This could be perfectly acceptable. In your offseason design in that previous thread, it looks like there is a polycarb belly pan as well? So that probably carried a lot of shear load to keep the frame from racking.

This particular image is out of date, please see the post above where I covered how the frame is done for this module.

In either case, the trapezoidal shaped cutout is done to the tubes regardless. Part of the top and bottom walls remain so it is still (albeit shallow) a c-channel shape. This is how we did our frame last off-season and was very happy with the strength. That frame does not have any kind of belly pan. There is a thin piece of polycarb on the top of the frame to mount electronics, this was mounted using some wood screws and double sided tape; it did not provide any significant strength to the frame.

As I showed in the post above, the normal corner frame configuration has the two tubes meeting in the corner and are bolted together.

OK, so there is still a shallow C-section there? It is hard to see that in the plan view. That will certainly help keep some strength through there.

It is clever, and I do like the idea of the cutouts. I just think it needs to be done carefully to make sure you don’t weaken things too much.

Thanks for the explanation and for sharing your latest design. Lots of great ideas in there. It is neat to see how the brushless revolution is driving some interesting iterations in swerve designs.

What was the reason to use the 3d printed shaft/pulley part over a 5mm hex shaft as the output from the ultraplanetary?

Why would using a shaft be better? The output of a UP provides a nice bolt circle to mount to, which means you aren’t relying on a female shaft interface to transfer your torque. This solution is a more robust way of transferring that torque, while also being lighter. In addition, there are not FRC pulleys available with a 5mm hex bore.

1 Like

Also, there would be less backlash in a single, bolted on piece than a hex shaft and pulley which would have two additional joints with some level of torsional play in them.

1 Like

I now have CAD uploaded for a flipped drive motor version that decreases the overall module height at the sacrifice of making the length and width a bit larger. CAD for this can be found here.

This module is 5.5" L x 5.5" W x 6" H. The wheel remains 2" away from both sides of the frame like the other version.

This module is a bit heavier at ~4lbs total due to the extra belt stage added to the drive motor, and the increase in size of the plates. However, being at 4lbs flat is still good.

Below are some pictures showing the dimensions, as well as an example of a 28x28 frame and how the module looks in size compared to a standard bumper with a maximum size cutout, with the bumper 1" off the ground.