At Houston champs, several people asked if the CAD for our swerve module would be available after the season. Indeed it is. Here is the link.
The modules held up well after a season of use. The few failures I remember were all either related to the skis that we added to the bottom, or bolts coming loose even though we used locktite on them. I guess next time I would recommend using high strength retaining compound on some of the critical bolts that don’t have to be removed. I’ll know a little more about how they held up after we take practice bot apart.
Here is the original version I posted a while back.
Close to 4 lbs for a module is pretty impressive. Well done.
I like the 3-D printed parts - especially the covers for the encoder gears. Very slick and professional looking.
What were you using for your steering encoder? Did it come pre-wired, or did you have to solder those wires on. We are looking for ways to make the wiring of our encoders more robust as the soldering of those wires is probably the weakest link in our design.
I’m surprised that you used a 775 for steering. Seems like a lighter, cheaper RS-550 motor would be plenty.
We are using a magnepot absolute encoder from digikey. They have three terminals for soldering that we have found to be robust, and a flat on the shaft so they can’t slip, so they are good for this application.
I agree with your points about the 550, but they are only slightly lighter, and slightly cheaper, and we would have to make a separate order to buy some. We use a ton of 775s, so it was convenient by comparison.
To make getting Into the depot smoother, we cut the ramps seen in the CAD model out of 0.375" aluminum plate. One failed from a hard sideways impact, and several bent their mounting bolts over time.
The wheels held up well over all. We had one need replaced on practice bot. Its flange peeled over because the tread was only attached to the wheel on the ends. (only half the rivet holes were used)
To install tread, we started by making a template out of u channel to drill the holes and cut the tread to length. Then we stretched the tread over the wheel and used 3/16" rivets to mount it.
This was consistently painful because the tread was cut too small, so it was under a lot of tension once installed. It was made more difficult because we put backing washers on the inside of the wheel to spread the load, which were difficult to get into place.
In the future I would probably use slots for square nuts to fit into, and bolt the tread on with 8 32s.
When you right click you should be able to hit export and then use the drop down for STEP. If that’s not working I might need to change the permissions.
What material we’re the wheels printed out of and on what printer? I wonder if threading the wheel itself if it was printed out of say, onyx on a markforged would work better.
Did the template purposefully space the holes in the tread shorter then the hole distance along the wheel so it would be stretched around? If so how much shorter was the C-C made and how did you arrive at that amount?
The wheels were indeed printed out of onyx on a markforged. We have put tapped holes in printed parts before, but I certainly wouldn’t trust it to hold the wheel tread on without testing.
We intentionally made the template somewhat longer than the circumference of the wheel to give ourselves some wiggle room. The problem was that the pitch length of the tread isn’t along the back, or even centered in the backing material (where we assumed it would be). It’s actually somewhere near the transition from the backing material to the blue tread material, making the effective circumference needed larger than expected.
Thanks! 2471 does not use Onshape, we use inventor. I don’t use onshape either really, but after hearing good things on Delphi, I decided to try it for a fun printed swerve idea last year. Then, after gaining access to some markforged printers, and the game turning out swerve friendly, the team decided to use the design this year. I’m hopeful that the team will move to onshape soon, but that’s TBD.
The planetary is a stock versa lite gearbox, except that we made a motor mount plate out of aluminum.
No specific thrust-only bearings were used. However thrust loads were carefully considered, especially in the design of the slew bearing taking the robot weight.
Did you do any stress analysis in Onshape? I talked with Bud and it came up that he was concerned about some of the parts until you had some analysis in CAD. They were thinner and stronger than I imagined for printed parts, but I’ve not had experience yet with using a markforged. Anyways, really nice and beautiful work.
No, we did not. I don’t think Onshape has FEA built in. It might have a plugin, but I haven’t gotten that far into learning the software. It would also be very difficult to get convincing results from FEA when working with such unknown material properties. We were definitely worried, especially about the wheel housing. So during week one, we printed a test wheel housing, and cut a test mounting plate, put some of our ball shifter balls in the bearing groove, put a dummy wheel and shaft in the housing, and bounced around with a person’s body weight on the assembly. This gave us enough confidence to go ahead with the design.
No, I don’t. I can try to remember to take one, but I’ll describe it to you in case I forget. We took a 1.5" wide, 0.25" wall U channel, and cut it to the length we wanted the tread. then we drilled the 8 holes into the bottom of the U on the Bridgeport, and finally cut a 1" wide strip of plywood for clamping the tread into the template.