My team has started to use swerve this season, and I’d like to know how well different kinds of swerve drives are surviving the 2020 competition. I hope teams using swerve can answer these questions：
How often did you have to repair your drivetrain?
What broke your drivetrain? (Being rammed by defense, going “full send” over the barriers, etc)
What kind of module and drivetrain setup are you using? (COTS or custom modules, wheel diameter, bearing setup, 3dp or machined parts etc)
Did you have to do maintenence such as greasing the modules or removing field debris?
Probably many new swerve teams like us still haven’t had the chance to touch the carpet due to COVID-19, so we’d love to know in advance what to expect!
We used the WCP V1 modules this year and ran into a few problems. In order of frequency:
Screws coming loose. I recommend either red loctiting or spring pinning the forks that hold the wheel in place, as well as the buttonheads on the underside of the module holding the big bearing in place. WCP supplies both pins and screws, and we used the latter with blue loctite. Still saw a few back out. We never had to repair anything down there, so I would highly recommend making those connections permanent.
Wheel tread wearing down. Lots of tread wear on swerves, so make sure you have spare treads available. We would often split the tread right at the edge, so make sure your tread-hole-drilling-jig is up to scratch. The Armabot jig does not work for swerve wheels- you’ll need to make your own.
Invest in snap ring tools to make sure all your snap rings are on correctly, especially the one on the end of the small bevel gear. Make sure you assemble everything in the proper order or that one will never go on.
A fair amount of field debris got wrapped up on the wheel axle during practice but it never affected performance, just had to be cut off every once in a while.
We heard about wheels getting trashed on the Rendevous Point barriers, so we elected to print some module shields to protect the wheels.
Programming was easy using WPIlib, and you get a lot of freedom in how to do things. Making sure your modules are perfectly aligned helps improve things like motion profiling and odometry (implemented using WPIlib). A simple straight bar clamped to the side of one wheel allows you to align two modules at once. If your mechanical aspects are on point, programming is a lot of fun. Almost too fun. Make sure your features are all dialed in and stable before you move on to the next one.
The swerve didn’t break outright until we had one module lose every single screw on it (this alerted us to the backing out problem) and the wheel fell out. Still drove ok up until that point, it was just throwing off autonomous control. Having a way to swap a module wholly without removing too many screws and wires (so many wires) would have saved us a fair bit of time.
Ran both events this year, didn’t have to do any major repair to the modules themselves. Did a good bit of driving over the center but our main run was trench. At our second event though we did have one of our gears slip a little but it was just a loose shaft collar. As for maintenance we applied extremely liberal amounts of grease any time it looked even slightly low just to make sure. Sometimes carpet string did get caught in our bevel gears so we had to clean that every once in a while as well.
We made out own swerve modules and they were only really working about the week of the competition. Once they were working they seemed pretty reliable. And then came the competition. We broke one of out swerve modules by running over a ball and it getting sucked into the wheel. This cost us the entire competition. I am not part of the build team so I don’t know the exact dimensions.
Really? For us the treads hold up extremely well. In ISR 2 we only had to change treads on two wheels, after a rough landing on the steel barrier(ran over a ball which made us come from a height and put a small bite in the tread)
We have been running swerve for 6 years now (the first two years with the Revolution modules and the past 4 years with our own custom design). We always build a spare module that is fully assembled and dressed out with motors and encoders to that we can swap out the module if we need to at an event to replace a damaged module. Pretty much everything on our modules can be repaired in situ, but sometimes, it is quicker to replace the whole module).
In the past couple of years we have not replaced any modules. This year, although we built the parts for our spare module as always, we actually never finished fully assembling it. The major sub-assemblies are done, but the entire module was net fully assembled.
Our failure modes have been mostly associated with the steering motor and encoder parts of the assembly. As recently as 2018, we would have the steering motor come lose which would cause our encoder gears to skip teeth in the mesh. I don’t recall any instances of this specific issue in 2019 but there was also another issue that can cause these gears to come out of mesh which would be if the encoder itself comes loose in its hole. I think we had one or 2 instances of that during the year.
We also had steering gearboxes come lose (the motor output collar on the VP gearbox set screw coming lose). We have drive motors come lose, but we have never had one of those cause us to stop functioning during a match, We have had steering motors come unplugged. We have had encoders come unplugged. We burned up our 775 drive motors in 2017. We cracked drive motor cases on our NEO motors on the cargo ship in 2019.
Over the years, we have been steadily improving the design to address each of these failure modes and for the most part our 2020 design has eliminated all the ones that can be eliminated with design actions (as opposed to good assembly practices like using loctite).
At this point, our modules are pretty reliable and our issues this year have been related to assembly practices. For example, we have to remove the VP output shaft from the gearbox in order to cut it to length and then re-assemble. Getting the snap ring on the output shaft on the VP Lite Gearboxes is annoyingly hard (and typically requires an arbor press to get the wave spring between the bearings to compress enough to get the snap ring in place). We had one gearbox where this snap ring was not installed properly and the output shaft dropped out during one match. This led to a hasty repair between matches and in our rush, the motor did not get plugged back in to the motor controller, so that module was dead for a second match. This led to a lot of wear on that wheel as we were dragging it around the field and over the rendezvous zone perimeters.
Over the years we have gotten into the practice of periodic inspections to look for things getting loose and generally tighten them before the progress to the point of failure. So even sloppy assembly practices are less of an issue than they used to be. But I expect we will continue to get better in that area as well with better quality control during assembly and better application of best practices.
In 2018 and 2019, 1072 used a colson wheel WCD and switched wheels perhaps once the whole season on the practice robot. The swerve drive this year needed new treads every two weeks at the absolute most during periods of heavy practice.
We use 3in diameter 1in wide wheels, so that inherently causes tread to wear out faster then a 4in wheel thats 1 or even 1.5in wide.
When I told you about the issue, we did have some un-optimized swerve software, fixing that did decrease the wear significantly, but obviously a 3in wheel will never be at the same wear level as a 4in wheel. Currently we swap tread about once every 2 or 3 practice days, and each day is about 3-4ish hours of practice.
Wheels had a full season on them. We determined this is not caused from scrub but from rapid cornering. This swerve could change heading 90 degrees in 100ms however a robot moving at full speed can’t.
Same driver slower game. Picture taken after IRI. This setup held up real good even after 3 events. We thought we had a winning recipe.
I don’t have a picture but 2020, new driver, same tread. She could destroy the rough tops in about 2 weeks of practicing. To be fair, 2020 is more of a full court sprint and with the added power of Falcons…
1899 used the andymark swerve in 2019. We had to replace modules countless times in the middle of competition and fix them. The bevel gears were pretty worn out by the end of the season. For the 2020 season, we used the SDS Mk2s and we never had a problem at our 1 district event. The only issues we had during the build season were issues with the Neos, not the module.
2019: a lot. 2020: rarely.
2019: playing defense. We broke a module at an offseason when another team climbed onto a module and drove back. That swerve couldn’t be fixed because we couldn’t take it apart.
Completely COTS. 2019: Andymark swerve, we switched to Neos as the drive motor near the end of the season. 2020: SDS MK2s with Neos.
We would regrease the modules occasionally. Never had to remove field debris from the modules.
4020 went to the SDS MK2 modules this year. Thus far, all we’ve broken are VEX Versawheels. The modules have been solid. We went through 10 wheels at Palmetto and all 10 were broken and heavily worn, but none ever broke to the point at which it impacted a round. The tread seems to hold the plastic rim together. We machined some AndyMark Hi-Grip wheels to fit the MK2s and were starting some wear and impact testing before Rocket City, but were rudely interrupted. Mostly because of the COTS swerve, our bot was right at the $5K limit so we didn’t really have an option to try the SDS billet wheels with tread. We couldn’t “afford” it.
Remarkably, doing module maintenance after Palmetto, we found that the module mounting screws were dimpling our base frame rails. Our rails are 1/8" thick so that had to be quite a lot of impact force. The modules seem to have handled it. If you aren’t practicing on metal boundaries that are incredibly rigidly held down, you aren’t getting a good simulation of the impact from a competition field.
We found a good bit of wear on the 2nd stage driving gear on the drive geartrain and had to replace those on every module. We weren’t too generous with the grease on the open modules. We’ll go a little heavier in the future. The rest of the gears held up well through fall practice, pre-competition practice, and one competition event.
We’ve had only one issue on the electrical side of things. It appears that we had a Spark Max on steering duty for one module reset during a match. I’m not sure how we might have gotten into current overload on the steering side, and we aren’t sure how else it could have reset. It only ever happened once and only on one module. We use the NEO encoders for steering after initializing with the MA3 encoders, so this became a real issue in the round. Since then, we’ve implemented the ability to re-initialize wheel orientation on the fly but haven’t had to use it.
We used the SDS Mk1 in 2019 and we never had any failures. The most we ever needed to do was replace the Versawheels because they wear very hard. Also when we swapped the drive gear ratio with our practice bot at district champs we found the larger aluminum gear had worn noticeably but we never had any using same gears through the rest of the season.
We didn’t have to do much maintenance aside from swapping Versawheels every so often, the SDS modules are pretty robust.
I can’t speak to 2020 definitively but I think we used the same Mk1 modules (probably rebuilt with fresh gears, etc.) and I don’t believe we had any significant failures.
That’s interesting, we didn’t have any of the versawheels break on us. IIRC, we went through 8 in the entire season. The andymark wheels we used in 2019 last a lot longer than the versawheels. I didn’t take a close look at the modules after our district event, so I can’t say anything about gear wear. We didn’t notice any issues with the mounting bolts either, that’s probably because we did more trench runs and only went over the metal bars for climbing.
Are you running the aluminum or steel version of the mk2s? We are using the steel gear version.
