As a small offseason project, I decided to design a swerve module with the main considerations being as compact as possible. The entire module is only 6.4" tall and has a footprint of 7.125"x7.125".
This module uses 3D printing extensively for the wheel mounts, the azimuth and the azimuth driving gear, the steer pulley, as well as the MA3 encoder mount. The azimuth itself has two teeth profiles, each of them offset from the other by half a tooth to help reduce backlash. The free speed is 16.1 ft/s.
I might be wrong but looking at that, the 4 corners of your drive base would be held together by the swerve module. Wouldn’t that put an unnecessary load on them and reduce structural integrity as opposed to a rigid frame?
Yes, they would be held together by the swerve module, but also by the belly pan. The reason for this is to push the wheels further out leading to a more stable drive base. The idea comes from 33’s 2019 Offseason Module
Here’s two pictures demonstrating the idea. The gear with the red arrow cannot rotate counterclockwise anymore since there are two points (next to the red dots) on the gear tangent to the other gear. Hence, it is in the position with the most backlash.
However if you look at the profile on the other side of the gear, motion in the clockwise direction (since its the opposite side) is reduced due to the fact that next to the green dots the teeth profiles are intersecting. Hence, the position shown in the first image is only possible with one tooth profile, however if the second offset profile exists, the position in the first image is impossible. Hopefully this gives an idea as to why there’s less backlash.
While it’s a neat idea, I think your backlash reduction is in the wrong spot. Right now, it doesn’t matter that you eliminated the backlash in the rotation drive gear, since you didn’t eliminate the backlash in the encoder gear pair as well. Since your encoder still has the same degree of uncertainty as to the location of the azimuth, the extra precision on the motor doesn’t matter.
Generally, precision on the sensor matters much more than precision on the motor, since error on the sensor side results in a faulty position, while error on the motor can be compensated for in many ways.
#SummerCD ridiculous marginal gains that don’t matter:
You could use the Falcon casings to replace some of the standoffs, if you pulled the set screw out of the side vent and ran an L-bracket from there to the lower plate? Bonus points for using a vented screw there. Extra bonus points if that hole lines up with the lower plate height and you just need to add an operation to put holes in to take advantage of it.
Pretty sure this doesn’t actually save weight, just adds complexity, but feel free to draw it out and prove me wrong. (“Throw a hole in it” probably only works if the plate is already ~1/4".)