Team 3737 Swerve Release (DinoSwerve)

Team 3737, the Roto Raptors, are pleased to announce the release of our first swerve drive system called DinoSwerve.

See our first test:

Details of the CAD, BOM and code here:


  • Weight:
    • Net weight: 3.9 lbs without controllers and motor guard
    • Gross weight: TBD lbs with motor controllers and guard
    • Low center of gravity
  • Height: 8.75” with controllers and motor guard
  • Motors: 2 brushless NEO motors with SparkMax controllers
    • Controllers are mounted within envelope of mechanical swerve assembly, effectively de-populating main electrical board of eight controllers.
    • Steering motors are powered by 20 amp PDB terminals leaving four 40 amp terminals free for other devices.
    • CAN bus controlled for simple wiring
  • Cost:
    • Utilizes integrated NEO encoder in steering motor rather than a separate absolute encoder and complex additional gearing. A simple low cost magnet and hall-effect sensor in conjunction with the NEO encoder provides accurate and repeatable wheel direction.
    • Low cost Delrin balls used for turntable vs high cost bearing.
  • Robustness:
    • Integrated but optional Lexan motor guard
    • Three plane fixing for rigid mounting in chassis
    • Extensive use of grade 12.9 socket head screws
    • Extensive use of Nylock nuts
  • Flexibility:
    • Designed be mounted at various heights within a chassis
    • All swerve assemblies are identical and may be placed at any corner
    • Three speeds are possible by swapping just two gears (10.06 ft/s, 11.7 ft/s, 13.69ft/s free speed)
    • Each assembly is identified and configured entirely in code (controller I.D number and rotational offset)
    • Independent adjustment of bevel gear and turntable height
  • Accessibility:
    • All moving parts are visible for inspection and lubrication without any disassembly
    • All gears are accessible and replaceable by removing four fasteners (except 94T gear) making speed change or maintenance quick and easy.
    • All parts are accessible and removable without removing complete swerve assembly from chassis
    • Complete swerve assembly can be quickly removed from chassis
    • Rapid and simple wheel replacement
    • All controller ports are visible and easily accessible
  • Fasteners:
    • Almost all fasteners use #10-32 threads for common nuts and tools.
    • Extensive use of socket button head screws for clean appearance
  • Innovation:
    • Certain custom parts manufacturable without use of a mill through inventive use of waterjet and fasteners.
    • Turntable bearing made from alternate ¼” and 3/16” diameter balls to simulate a ‘race’ (preventing adjacent balls from opposing rotation while touching each other)
    • Common gear sizes for simplified spares inventory
    • Rotation sensor implemented with no extra moving parts (hall effect and magnet)

Questions are welcome.


Looks great guys! I love seeing all the new swerve projects being revealed.


Thanks. Yes it was great to see this result after 8 months of design, build and coding.

And yes, there does seem to be a lot of swerve designs revealed at this time.


Woot! Woot!

Looks awesome guys! Can’t wait to see it in action!

The NC swerve infection is spreading!

OMG!! We barley got our feet on the ground with a basic drivetrain and the NC swerve teams are multiplying!!! Hopefully 3737 and 1533 will help spread the swerve love to 7443.

Thanks. But also credit to 1533 for the idea of the turntable with Delrin balls. It’s amazing how well they work! We did intersperse the 1/4" balls with slightly smaller ones to create a pseudo ‘race’ and prevent contra rotating balls from touching. Seems to be working well.


It may come down to code… we used LabVIEW and as you can see, it almost defies physics when it moves around. Our programmer did a superb job (and has some more features planned that will provide a significant advantage in competition)! But if you use another language that you are familiar with consider the alternatives out there. 1533 have a well established design that has been refined over several years. And there are many others.


Ater kickoff today, we tried the DinoSwerve over the rectangular hollow section that came in the kit of parts. It went over without hesitation - that was good to see. (and the actual steel 1 x 3 tubing has larger radiuses so we are reasonably confident that our Swerve should work this season).

1 Like

Dope design, 7265 will be joining the gang this year as well

Thanks! Have you decided on a specific swerve system yet?


We’re running something kind of similar to the 2018 WCP swerve with NEOs in our current design, we didn’t have the design capabilities to do our own machining etc.

Nice! What language do you use?

Java, used some of Ether’s papers for examples and implemented a solution to the “180 problem”, need to add some more nice features but for now it drives at least!

My son is fairly familiar with the logic behind swerve systems and all the nice little features available to optimize them. His email is if you want to pick his brain about anything. (his prefered language is Labview but the math will still apply)


One other feature of the code is that you can select any center of gravity within the perimeter and the robot will rotate about that point. So find your CofG when fully built (place robot on a board and use a short post about 1" - 2" dia to balance). Then just feed the center position of the post into program (x and y) and the robot will spin cleanly about its weight center.


1 Like

We use Java as well, so if you need help with anything, let me know and I can put you in touch with our programmers.

Did you cnc the gears that are used for turning the module? If so, do those work well?

The large gear was laser cut in 1/4 inch high strength aluminum plate. All small gears were bought from VEX. After extensive testing, the gears are still meshing well.

We use a similar technique to 3737 to make the large gear that turns the module. We have had this gear cut both with a waterjet and with a laser over the years and both work fine. There is not a lot of load in this mesh, so the relatively rough surface finish is not really a problem. We have, on occasion, lapped the gears by spinning the module continuously for a particular length of time, but we have never actually noticed a difference between the modules where we have done this lapping and the ones where we did not.

We also had a set waterjet cut and to be honest, they were almost as good as the laser cut gears and I would judge that they will work as well. Since we had both (water and laser cut) we put the laser ones on our competition bot and kept the waterjet ones as spares (but I really doubt we’d have needed them even if we had had a full season). I think TS’s comments above are totally valid.