pic: Yet Another Swerve Drive

Summer CD? Summer CD.

Driven by a CIM at 5.5:1 and turned by a BAG at 66.7:1. Free speed of 16 ft/sec.

I’m not sure if I need the plate on the bottom, but at the very least it looks nice.

View the CAD here: https://grabcad.com/library/team-100-swerve-v2-1

The bottom plate is probably not necessary unless you are very worried about safety. So, getting rid of it will remove more weight from your module. Looks pretty good so far!

It looks good.

Just looking at the picture, my main critique would be that it seems like it could be made more compact.

I would try to make the whole module shorter by reducing the gap between the top of the wheel and the end of the CIM shaft. This will reduce weight, space, and the moment load on the structure.
I would also choose a smaller belt for the steering and move the versaplanetary closer to the CIM.

I had tried using a smaller belt, but 80t (as in the photo) is the lowest you can reasonably go. There’s 0.39" of space currently, so I suppose you could take it to maybe 78t, but that starts to become a pain to work with.

That’s some good advice regarding shortening, though. I’ll probably recess the wheel into the a pocket on the pulley in the next rev.

I don’t know a lot about designing swerve(yet) but from what I’ve seen most people are using absolute encoders on the module rotation to make field oriented drive easier for programing and match set up so why did you decided to use a quadrature encoder and have to deal with wheel alignment and such?

Mostly, I just wanted to try something different. I’m using the WCP Hall Effect Sensor for zeroing (you can see it on the CIM mounting plate)

Trying something different can be beneficial, but general inadvisable when there’s a wealth of knowledge surrounding the issue. Having to zero a module introduces the possibility of human factors effecting your drivetrain, or a loss of time at the beginning of each match. If those are acceptable risks, keep blazing your path.

We went with absolute encoders for our swerve, and I’d recommend that heartily over any other method.

How would using magnets for zeroing introduce human factors? Plenty of swerve drives have used limit switches or magnets for zeroing without issues.
Absolute encoders can move over time (panel nuts get loose, etc.) so in some ways quadrature+zeroing is a bit more secure.

You will come to dislike the time it takes to zero the modules. No matter how efficient/fast you get it to zero, it will seem like an eternity especially during auton. The next step you’ll take is to pre-position the wheels for auton which is what we did but now there is the human error factor. Lets say that you accept this, you will zero the wheels after auton which means other robots will have moved somewhere between 2-3 ft before you start moving. The best bet is to go 1:1 on the azimuth and use the absolute built into the VEX encoder or hang an absolute on the module as others have.

As far as things moving over time, the same can be said about the magnet and/or the sensor used for zeroing.

If you can figure a way to do it without a zeroing switch, you will be happier, guaranteed.

I should have been a little clearer in my phrasing.

If you have a zeroing routine using the switch/hall, you will lose time at the beginning of each match.

I have also seen teams move their modules to the “zero” switch prior to entering the field, this introduces a human factor.

Either losing time or introducing a human factor seems silly to me, when there is a simple one part solution for it.

Moved the turning gearbox a little closer and added a tensioner.
Removed the bottom plate.
Recessed the wheel into the pulley
Switched from 15mm to 9mm pulley