Hi, Everyone! I just was curious if anyone has had the issue of play in the motor axel of the turning motor and/or wheel of your AndyMark Swerve Drive? Has anyone attempted to use belts around the gear to reduce the play? If you could let me know if you had the issue and how you fixed it, that would be amazing. Thank you! 
Sincerely, Seth T
What kind of play are you talking about? In what parts exactly? What direction? Sharing a video may help you be more descriptive so we can help better
Are you talking about the bevel gear for the drive wheel? Can you elaborate?
I’m sorry, I meant like unwanted movement/loosness in the axel of the turning motor and the gear on the axel of the wheel that connects the wheel to the motors on the swerve drive(the bevel gear). It causes the wheel to be able to turn left and right slightly without aid of the motor (which is not good) and the wheel is able to move forwards and backwards slightly as well. I do not have our swerve drive available at the moment because it is in the middle school and we are not allowed to go in there. I am really sorry.
Yes, that is what it was called. Sorry for not explaining it well, I’m not that good at explaining. It seems to be the bevel gear on the axel of the wheel that connects it to the motors and the bevel gear for the turning motor.
I think you mean backlash? It is a common problem in swerves and teams put a lot of work into minimizing it. Should be fine for regular teleop but it can cause problems in complex auto routines.
Someone correct me if I’m wrong but I’ve heard that properly tensioned belts have less backlash than gears and chain.
It sounds like you are talking about backlash in the steering.
Every gearbox/geartrain system has some amount of backlash. Even belt systems can have some backlash but generally the friction between the belt and the pulley between the cogs will be enough that you won’t feel the backlash.
Reducing backlash is good. But there will always be some backlash. In the AM swerve module, you have a PG steering gearmotor that has, I believe 3 stages of planetary gears inside, each with some amoung of backlash. Then you have the backlash between the output shaft and the driving gear and then the backlash between the driving gear and the driven gear. With the AM setup, I’m not sure there is much you will be able to do about the backlash.
For your steering feedback control, are you using an encoder on the module itself (MA3 or Lamprey) or are you only using the PG encoder? If you are only using the PG encoder, then the backlash will not affect your control loop stability, but it will affect your steering accuracy. If you are using an encoder on the module itself, the steering accuracy will not be affected by the backlash, but there will be a deadband of motor rotation that will not cause any module rotation and your control loop will need to be able to handle this deadband.
All swerve steering systems have some amount of backlash and it generally does not cause any significant problems in how the robot drives or the accuracy of odometry. I don’t know what a normal amount of backlash is for the AM module, and if you think your modules have an excessive amount, you should review the assembly to make sure everything is assembled properly ,that there is no excessive wear on the gears, no loose joints, etc.
We used the Andymark swerves in 2017, 2018, and 2019. Since the bevel gear is unsupported, wear can accumulate and cause issues. The only way we were able to get around this (if I remember correctly) was to take the entire module apart and replace the bevel gear. Odometry on these specific modules is not easy because you’re running the encoder at the beginning, not at the end. We typically found that our encoder could get us close enough, then any vision system you have can make adjustments if you aren’t completely perfect.
I agree with the others here, this sounds like gear mesh and hex shaft backlash, pretty much unavoidable in systems with hex bore gears due to manufacturing tolerances. I believe the usual backlash at the wheel is about 1°. The lamprey encoder on the 2020 version is the best one to detect this as it is reading directly off the wheel module instead of being removed by one or more gear meshes.
We actually addressed this in the 2020 revision. Instead of the gear and wheel being located on the shaft by e clips they are located by spacers and the shaft is retained with screws in the ends. This, along with the orientation of the bearings means that the yoke is prevented from spreading apart over time. This spreading, even by a few thou, would separate the bevel gear mesh and cause accelerated wear leaving to failure.