pic: Light Weight Swerve Module



After posting designs for a shifting swerve drive on these forums a few days ago, I was in a CADing mood. Designs that I’ve done previously prioritized the horizontal compactness very highly, so in this one I tried to make it as low as possible while still having a vertical CIM motor. It’s also the lightest weight module I’ve designed so far. It weighs in at 5.25lbs each according to inventor.(includes most hardware)

Is that with the motor weight?

I tried to match the motors to actual weight by changing the materials in inventor.

here’s a breakdown of weights as inventor sees them.
cim: 2.78lbs
Planetary: 0.4 lbs
AM 912: 0.51lbs
Other: 1.56lbs

Looks great. Are you going to use it on any of your robots?

You can key in a custom weight for an individual part or assembly.

5.25lbs with the motors, shifting gearbox, AND module!
You win swerve. Seriously, this weighs less than a WCD. If you make a 4-cim, 4 minicim version then there would be very little disadvantage to using this apart from the programming and driving. I also think you could make this into a shifting version without too much trouble.
How are you tensioning the belt? If it just slides along the track you have and is clamped by screws, I would defintiely change that to having something similar to 192’s gearbox wehre you have seperate holes for each tension. Sliding tensioers with friction holds are sketchy IME. That was the primary reason why our drivetrain failed this year. It might not be as bad on the cim, but it won’t add weight to change that slot to holes.
How much ground clearance for the cim and module? Why banebots over versaplanetary? A BAG motor and that massive gear reduction will allow you to use a single-stage versaplanetary for the turning motor.
Where is the top plate? If it’s not present, that should be okay, but testing this would be of high importance IMO.

If you could release the CAD, that would be a boon, but I understand if you don’t want to.

EDIT: Wait, is that CIM held on by only two standoffs? That seems very insecure; add at least one more for strength…

What would the clearance be for the CIM wires?

Wow, that’s amazing! Just a few question though. Do you feel that the standoffs will hold-up on the potential forces acted on the assembly, especially if they’re that long? Why is there a bushing on the swerve module without a supporting plate? And how is the rotation gear held onto the bottom plate, just by the part where the CIM powers the wheel, or is it not in the CAD yet? What is the calculated FPS? What is the estimated cost?

And building.

Well yes, but I would presume that getting the drivetrain machinined could happen before the manipulators, thus making the assembly have much less of an impact. Driver training would be slightly shorter, but any driver of swerve should really have many hours of driving swerve under their belt pre-season.

EDIT:
The banebots planetaries are 0.5lbs according to the website, not 0.4lbs. Just a quick heads up.

Looks great. Are you going to use it on any of your robots?

To minimize the time commitment required to build swerves, our team built four swerves last year in the off season that then got reused on the practice bot for the competition season. This year, we are in a notable money crunch, so we’ve decided to do limited off season swerve building. Because of that, the odds of us putting this to use in the build season are minimal.

You can key in a custom weight for an individual part or assembly.

Great! I did’t know that you could do that. I’ll have to look it up.

How are you tensioning the belt? If it just slides along the track you have and is clamped by screws, I would defintiely change that to having something similar to 192’s gearbox wehre you have seperate holes for each tension.

It has a single hole on one side, and a curved track on the other. I like this setup because it is adjustable enough to allow for different gear ratios without changing the belt.

Why banebots over versaplanetary? A BAG motor and that massive gear reduction will allow you to use a single-stage versaplanetary for the turning motor.

The bane bots is cheaper, smaller, and lighter. I couldn’t find a configuration that would allow for a single stage planetary without being really large.

Wait, is that CIM held on by only two standoffs? That seems very insecure; add at least one more for strength…

There are three standoffs. I’ll try to get the CAD up here pretty soon I suppose.

Where is the top plate? If it’s not present, that should be okay, but testing this would be of high importance IMO.

I’m not sure what you mean by “top plate.”

What would the clearance be for the CIM wires?

0.59"

Do you feel that the standoffs will hold-up on the potential forces acted on the assembly, especially if they’re that long?

I’m pretty sure they will be fine. The forces on them are from holing up the CIM motor and encoders, and from the CIM torque.

Why is there a bushing on the swerve module without a supporting plate?

There isn’t. I think you are referring to the 3D printed encoder gear.

how is the rotation gear held onto the bottom plate, just by the part where the CIM powers the wheel, or is it not in the CAD yet?

there is an integral trust bearing cut into the turning gear, and bottom plate. It’s much like what team 16 has done in the past. Basically there is a track for a 0.25" ball bearing cut into the bottom of the plate and the top of the gear.

What is the calculated FPS? What is the estimated cost?

It’s currently geared at 3.22 to 1 on a 2.75in wheel. So about 16fps, although it’s easily changeable.

Thanks for all the great questions and comments, I hope my answers make sense.

That tensioning method could be unreliable. Testing would be paramount, and easier would be to have discrete mounting holes for each tensoning distance.
By “top plate” I mean an upper support. Right now, the only thing keeping the module in place is the weight of the robot. You need to have some kind of way to secure the module from above as well, either by means of a plate on top of the module or by using some kind of loose clamp on the bottom plate.
As for gearboxes, I see what you mean. I’m cadding one like this but shifting right now, and the banebots is considerably lighter than the versaplanetary. However, I hear that they can be unreliable.

Wow! That was a fast response time.

We ran a swerve drive with a setup much like this last year but with two straight slots instead of a hole and a curved slot. We didn’t have any issues with it. Truth is, you don’t want a whole lot of tension in the first stage of reduction because it will cause resistance.
The caster is held from falling out of the robot is the snap ring groove above the timing pulley. (The groove is hard to see in this render)
I’ll look into the reliability of the banebots. Thanks for the heads up.
I think this could be modified to make a great shifting design. I might CAD one up next time I get a notable block of free time.

Ah, so that’s how it’s held. Very interesting.

Great design!!!
The more designs someone does the more they learn and the better they get.
Your designs get more compact and innovative every time, keep them coming.

What is shifting the gears? I don’t see an air cylinder or servo in there.

I prefer the Vex Versa gearboxes also, due to the fact the output shaft has a flange. The Banebots gearbox doesn’t have a flange on the output shaft and that can cause problems if not used correctly. We tried using a round sleeve and a screw/washer to pull the gear up against the Banebots gearbox bearing with bad results last year. We switched to clamp collars and everything ran smooth after that. I agree they’re lighter and that’s a benefit.

I am a little nervous about retaining the steering balls with a single clip at the top. We went through a discussion about using a clip on our 8WD setup and found they can take about 600lbs of force for a 1/2" shaft so the clip isn’t the issue. I am more concerned with the strength of the upper plate that will be seeing that load. I think there are more forces there than you are considering. In normal operations the weight of the robot will bear down on the balls and this will be a non-issue. When the robot is pushed to the side and the balls try to climb out of their grooves you will see vertical load that might cause issues. Again you might be fine.

The other consideration is contamination getting into the steering ball tracks. We had issues with aluminum chips from drilling on the robot, debris from the field and the carpet fibers getting kicked up from the tires getting in the steering grooves. This is the reason that we are switching to sealed ball bearing on our 2nd generation swerves. We disassembled our units and replaced the balls multiple times last year. I would guess Bomb Squad doesn’t have this issue because they leave the steering bearing setup above the wheel and not right next to the carpet. I would be interested to hear from them if they have seen this issue also. Maybe consider a labyrinth seal (a winding path to the balls instead of a straight slot between the plates). I was considering using two opposing L-shaped walls instead of two round grooves.

This is a single speed design. sorry if the description doesn’t make that clear.

As far as the retaining clip, in order to cause the balls to try to crawl out their grooves and push up on the robot, the robot would have to pull several Gs of horizontal acceleration.

Thanks for warning about grit getting in the bearing. There is only 0.04" of clearance between the turning gear and the bottom plate, but I’ll start thinking about ways to seal that gap.

What bearing did you use on your most recent module?

I’ve seen you rule out the slot tensioning method in a few places. Have you guys had a bad experience with it?

We’ve done it numerous times, on 550s all the way up to CIMs, without any issue. It’s plenty reliable if the fasteners are tight.

You shouldn’t use a socket head by itself as it’s small head diameter doesn’t grab a lot of meat, so we use buttons or a washer in addition.

The primary reason reason for out drivetrain failing this year was tension-held bearing blocks on the corners of our WCD. There were two screws going through the bearing blocks on either side of the 2x1 with nuts on the other end. No matter how much we tightened the screws, the bearing blocks would always slip and cause drivetrain failure.After adding some sandpaper-like gripping tape in between the bearing blocks and the 2x1 post-season, we no longer had problems with the drivetrain failing. This is why I always recommend against any kind of friction fits.
Now, the forces are less extreme on the first stage of a transmission, but I still wouldn’t do it in the middle of the season.

Ah.

The loads here aren’t even comparable.

No slot obviously is more secure, but slots can be used for motor c-c adjustment without issue in FRC.

A nice trick is to angle the slot away from the direction of force (for belts/chain) as much as you can fit. This not only increases your tensioning resolution, but decreases the friction required to hold position.

For gears you actually want to aim at the other shaft, as the forces pushing the gears apart are the lowest of the forces generated in that interaction.

I adjusted the weights of some components, and added all of the hardware accept snap rings. The new estimated weight is 5.4lbs each. Slightly heavier that the original estimate, but still pretty light. I’ll start working on a shifting version soon.