Here are some pictures of a swerve drive that I’ve been working on. All of the plates are 1/4" thick, but I think that I could get away with making the top one 3/16".
The gears are from an AM cimple box, the hex bearings/sprockets are Vex Pro parts, and the rest are McMaster parts. The miter gears are boston gear l110y, although mcmaster sells almost identical ones for less. The steering motor is a rs550 with a 100:1 versaplanetary gearbox. My main concern is that the bearing set up for the module will not be strong enough. I have a thrust bearing on top of the module to hold the robots weight (can be seen in the last picture), a huge bearing (blue colored) on the lower plate, and a bronze bushing on top. Does anybody know of a team that has used this setup successfully in the past?
Feedback and constructive criticism are appreciated!
Can you show an exploded view or side view of the bearing assembly that supports the wheel? I’m having a hard time visualizing the assembly and how it all goes together. It would help if you also included the part numbers (McMaster) for the bearings.
Are the three mounting bolts threaded with a spacer tube in between the two 1/4" plates to keep the height level across them?
How does this mount to a chassis?
Overall, it looks like you had a good project this summer!
There are three bearings which support the wheel holder. There is a 1" diameter tube that is welded to the top of the wheel holder.
On the top plate, there is a bronze bushing (colored brown), which is mcmaster part number 2934T26.
On the bottom plate, there is another bearing (mcmaster 6384k373).
On the top of the wheel holder there is a thrust bearing that has an ID bigger than the tube that come out the top. It sits in a groove cut into the top of the module. It is mcmaster part number 6655k250.
There were a couple things I forgot to make visible before I took the screenshot, including the spacers on the 1/4-20’s and the hex shaft for the 56t gear.
I’ll add an exploded view when I’m at my computer with Inventor.
This is a cool setup. As far as your bearing system is concerned. I think it is more than strong enough. I also designed and built a swerve drive system this summer. It had two plastic sleeve bearings, and a 50 thousandths thick steel washer on top of the caster box. This option saves money and weight, and is more than strong enough.
Some suggestions: plan for multiple frame and mounting options. Include any encoders that you plan on using in your design. If possible Plan on your chains stretching, and put tensioning mechanisms in the design.
I’m not ever going to use this design, so I don’t have to worry about the chains stretching. Also, all the chains are exact center to center and are pretty short, so I think I should be fine.
As for mounting, I would probably just make the three bolts that hold the two plates together longer and use these.
Just personal preference, but for ease of maintenance I’d probably change the orientation of the steering motor. With it hanging below the module, it’s going to be a pain to get at, especially with the wheel right next to it; I’d probably flip it.
I’m also a strong advocate for belt over chain wherever possible, but that’s really minor for short distances like this.
I agree, it would be easier to change out the steering motor if it was upside down.
As for the belts, I had originally left out the gears and I had a 4 to 1 reduction with belts. The problem was that I couldn’t make a layout using pulleys and belts from sdp/si or mcmaster that had enough wrap around all the pulleys.
Do you know what type of belts FRC teams usually use? I looked at the GT2 series from gates, but I couldn’t find any info on the lengths that actually exist or where you could order from.
I think that I could probably get away with chains in this case because they are pretty short and probably won’t stretch that much.
I generally use HTD 5mm belts, which seem to be pretty standard; you can buy them from B&B Manufacturing (http://www.bbman.com), or alternatively just use their very nice catalog for designing purposes and see if you can buy the parts from a cheaper supplier once you have all the part numbers.
You can order GT2 belts from SDP/SI. That’s what we did this last year. They come in a ton of different sizes (as should be expected) and widths (they carry 5mm, 9mm, and 15mm, if I remember correctly).
We were very happy this last season with our belts ordered from SDP/SI. There are some changes we’d make to how we drove wheels and such, but compared to our use of HTD belts last year (in a low-torque, low speed usage), we were much happier with the GT2 profile. It seems, at least from my team’s experience, a little more forgiving.
As of right now, SPD/SI does not list any HTD/GT2 profile belts in their online catalog, but an older PDF of their catalog shows that they used to carry them, so I’m unsure if they are availible. I’m not planning to actually use this design, but I at least wanted to make it possible to build with parts that I could buy.
SDP/SI has a really, really obnoxious catalog, which is why I recommend using B&B Manufacturing instead, at least to find the part number. Most of the stuff is manufactured by Gates, so the part numbers are universal; you can hunt around for the supplier with the cheapest price once you’ve found your specific part.
Honestly, we used those wheels last season for our robot. By the end of the season, we went through 2 wheels and the third was down to the rim. The tread material is really easy to chip and break. If you want a good wheel, just go with the self-treading wheels. Everything else looks good. Make sure you are ok with tolerences; our team forgot about that and we got some parts that we had to redrill. Good luck with the swerve drive!
We ran (are running) 4" HiGrips on our swerve this year. They’ve been great to us. The 'bot has well over 100 matches on it from 2 Districts, MAR Champs, Newton, Einstein, IRI and 3 off-seasons, and the 2-3 replacements each still make all the HiGrips cost less than the Plactions would. For us, switching the wheels is basically as easy as switching the treads would be. 3 more events left, and maybe one more switch (but probably not).
We ran Plactions for years and liked them as well, but between performance, cost and ease of use, HiGrips won out. (Ran Colsons for a bit too, but the prep time especially just wasn’t worth it for us.) They do chip, but I can hardly blame them the way we drive our swerve. We did some (not-quite-official) friction tests tilt-tests last summer, and HiGrips beat Plactions on our carpet (also not-quite-official) basically until the stage at which we replace them. YMMV.
I’d give them a shot–preferably try Plactions too, but at $26/ea vs $6/ea, I know which I’d try first. See how they behave for you.
However, we have 2 different sets of modules, so in terms of effective wear-to-switch, you’d be looking at 8+ wheels. (We do drive our practice 'bot, but not as much as we’d like.)
This actually gets sort of interesting when we’re running a mix of newer and older wheels. We have no proof that this has any distinct effect, but there’s some superstition around it with regards to pushing matches.
Also, all the chains are exact center to center and are pretty short, so I think I should be fine.
