Over this past summer and fall, my team (2090) decided to work on designing our own swerve drive modules. 2019 was the first season we decided to use swerve, however the ones we used were partially designed in season, and had a few large issues. so we went back to the drawing board, and designed new ones from the ground up.
As you can tell, these modules were influenced by the designs of 2910, 1323, 33, etc. we did make some larger changes however to fit our build style, as well as repair-ability. So far, the modules haven’t had any major issues, and seem to be running very well, however we never got the chance to compete with them. So, if you have any suggestions for improvement, please let me know below, I’m looking for some possible improvements for the future.
Stats:
Weight:
~5 lbs.
Drive:
Falcon 500 drive
overall 5:1 ratio for a 3.5" wheel (ratio is interchangeable)
Adjusted ground speed is ~16 ft/sec
3.5" dia Onyx wheel with blue nitrile tread
Azimuth:
NEO BLDC on an 18.15:1 ratio for turning
ma3 absolute encoder for turning
Current ideas for improvements:
fully print the actual module part out of onyx
switch NEO to falcon 500
improved encoder gear hub
I know this isn’t really the purpose of your posting, but could you put me in touch with your programmer? This was our first year also of swerve and although we were lucky enough to compete (Northern Lights) I would be interested in comparing.
Cool! I wanted to design a swerve but most designs I saw used a cnc mill or were made heavily out of onyx, both which we cannot do. We can waterjet so now I have something I can base my design off of. What was the azimuth pulley printed our of?
Are the wheels 1 or 2 piece? From the CAD, it looks like 1 piece, but it seems like you’d need support to print 1 piece which might be problematic.
You mentioned that you didn’t get to compete, but did you have a chance to test the Onyx on steel boundaries? From reports on this forum, boundaries were damaging all sorts of wheels. We destroyed multiple sets of VEX Versawheels on the boundaries at Palmetto while running swerve. We were looking forward to comparing how AndyMark Hi-Grips would perform at Rocket City. We weren’t planning to try aluminum wheels for a variety of reasons including reports that they were denting also and not wanting to transfer the shock more efficiently to the swerve modules.
the wheels are printed as one piece.
we did not get a chance to heavily test with the boundaries, but our guess is that it would have survived, we just would probably have to take it slow
yes, we have considered using that, however we ultimately went with the ma3 because the cancoder option would be more bulky as far as mounting, and our turning was a NEO. ma3’s also are eaiser to mount to in this application
if you are looking to design modules based off of these, keep in mind that when I say its designed to fit our build style, i mean it a lot. if you look at the CAD of our chassis this past year, which is included on the same document linked above, it is almost all water jetted, and not necessarily built for the simple 2x1 frame teams like 2910 use. the reasoning behind this is because we currently have no way of cnc-ing long pieces of 2x1, and the less manual work, the better. So by all means feel free to take inspiration, but keep in mind my team doesn’t build robots like most.
We protected the wheels on our swerves by making “swerve shin guards”. They are ramps that go around each wheel that only allow the tips of the wheel to protrude. The ramps lift the bot before the wheels hit minimizing the impact to the wheel. We used our bot from 2019 to test without the ramps and managed to break a wheel in half on our practice field. The 2020 bot with the ramps had the least damage to the wheels that we have seen in a long time. The ramps could be printed if you have a big enough printer, but we machined them out of HDPE. Our original design was all round so they could be made in a couple of operations on the lathe. For our end design we made them square to allow the mounting surface to have more contact area with the frame of the bot. We milled flats in the ramp area and drilled holes through the flats to allow button head 10-32 bolts to mount the shin guards to the frame and the swerve modules.
We actually were planning on making a simmilar thing, however we decided not to because it would take too long to print, and we didn’t think we needed based on how much we reinforced the wheel.
