Over the past year or so, 1710 has been working on developing a swerve drive. We have attempted swerve once before, and even have some prototypes to show for it, but this will be the first year that we will implement the agile drivetrain into our competition robot.
Our earliest attempt at swerve was a 2 axle design, with one axle for the wheel and another axel above the wheel used to transfer the rotation from vertical to horizontal. This makes the design large and requires that we transfer the driving motion from the first axel to the second axel, adding backlash. Although the design is outdated by today’s standards, this first attempt at swerve is what ultimately inspired our team to make a swerve robot this year.
(3D Printed Prototype)
(Metal Prototype)
Last year we decided that we wanted to have a swerve drive in order to compete on a higher level and to challenge ourselves. Our Chief Operations Officers did lots of research into other teams and cad models for an in-house design. Specifically, we were impressed with the results that teams like FRC team 1690 Orbit had. We even took design ques from their upside down electrical panel design. However, after consulting our coaches and mentors, we found out that the biggest reason our first attempt at swerve was never implemented was due to the logistical nightmare of machining and testing a swerve module in house. So ultimately we choose to use a modular kit system.
(Team 1690 Orbit Swerve Video)
(Early Proof of Concept)
We decided to go with a pre built MK2 module from Swerve Drive Specialties. We were impressed with the videos we had seen online using their modules, offering the latest motors in a compact package. The SDS modules use a large bevel gear to avoid having to use 2 parallel shafts, which makes it more compact. Having jumped the first hurdle, we started planning the assembly of our first swerve bot. We scheduled a covid friendly way to distribute modules to senior team members so they could assemble them without having to contact one another (making sure to wipe down the modules with alcohol pads before and after hand offs). Then, once we had the modules assembled, it was time to put together the chassis itself. The chassis was built by our Chief Design Officer and one of our coaches. Finally, the finished drivetrain was delivered to our Chief Technical Officer to work out the software kinks.
(1710 Outreach/Test Robot)
(MK2 Module)
Once we had the swerve drive train assembled it was time for some rigorous testing. Our CTO drove the robot long and hard in an attempt to find any flaws that could come back to haunt us during a competition. The most common failures were the 3d printed encoder gears, bolts coming loose, and the VEX wheels. The 3d printed encoder wheels split along the layers of the extruded plastic relatively often. This was an issue inherent to the MK2 design and was only resolved when we made the switch to the brand new MK3 modules(Although teams may have more success with metal gears for this application). We also encountered bolts coming loose from the vibrations associated with moving around so quickly. This was mostly a problem inside the gearbox but we were able to resolve it with a healthy amount of Loctite. The final common problem we ran into was with the VEX wheels. The rubber tread wears down very quickly and blisters. To resolve this we switched to the billet wheels for our MK3 design. We also had some issues with the potentiometers coming loose but it was relatively easy to tighten them back down. Overall the MK2 is very flushed out and easy to recommend. However, MK3 solves many of the problems associated with MK2.
(Versa Wheel Blister)
We have recently completed assembly and programming of our MK3 swerve module robot (Linked here is our CTO’s post with all of the programming specifics: Working MK3 swerve). The Falcon 500s upgrade is definitely worth the price. The extra power and integrated motor controller makes them more of a successor than an alternative to the Rev Neo motors we had been using on our MK2 robot. Although beware that the can bus system can easily become a rats nest with the 8 motors and 4 encoders all connected together. The magnetic encoders mean that the previous 3d printed gear interface is upgraded to a magnet directly embedded in the drive shaft. This will save us precious time replacing the gears. There is also less backlash in the MK3 modules compared to the MK2 predecessor. This seems to be due to the new drive gear which is all one machined piece compared to the old drive gear shaft that consisted of multiple gears on a shaft. We still noticed lots of wear on the billet wheel treads, however the Nitrile tread is easy to replace and cheaper than buying a new wheel from Vex. Additionally the modules use the same pattern holes to attach to the chassis on MK2 and MK3 which made it very easy for us to test MK3 on our old MK2 chassis while also developing our robot for the 2021 season.
(MK3 Module)
(MK3 Module)
Overall we are very impressed with the SDS modules. The pre-designed, pre-machined nature of the modules make them a relatively easy option to implement. Especially for teams who don’t have the machines to produce the large bevel gears.