pic: GBX-127, bevel-beside-wheel swerve drive



A swerve drive similar to 1323’s this year, only using a delrin-groove bearing instead of a Silverthin or other large bearing.
Weighs 5.3lbs, but it could be made lighter by using a thinner top plate.
Geared for 17.7fps with a 2.5" wheel. The wheel is 3 1.9" diameter 1/4" thick waterjetted a mill-finished plates sandwiched together with 0.3" thick blue nitrile tread on it. Currently the design depends on locktiting and/or pressing the wheel to the bevel gear hub (0.35" wide) instead of using positive engagement like a key to save on machining. Before using this in competition we would do some torque tests on it. It’s possible to instead convert the gear hub to a D-style shaft, key it, or set screw the wheel to it.
CAD files are available in the 115 GrabCAD cad release (as soon as I update it). PM me for a .stp (STEP) file.

Can you better explain how the encoder is mounted? It looks like there’s some kind of polycarbonate bracket going to the CIM, but it’s hard to tell what it’s doing or where it’s mounting.

I think he’s going for something like a surgical tubing mount. That is the encoder shaft is connected to the shaft by a piece of surgical tubing. This method allows for some axle misalignment and is useful for low precision parts like the poly mount

Yup, that’s it.

Why is it geared so fast? 17 fps for a swerve seems like it would be way faster than you’d want in most games.

I ran it through a drivetrain calculator, and I found that 17.6fpsish is the best time-to-distance for 2ft, which is perfect for getting around opponents. 17.7 is a little fast, especially given inconsistencies which the model doesn’t catch, but it’s easy to change the design depending on the game.
That’s free speed, not adjusted; it should be between 14-15 adjusted.

Is this designed to be made with resources available to your team?

If so, does it use so many resources that it takes away from other systems?

Yes, the method of waterjetting + milling is one that was specifically tailored to our needs; materjetting is cheap, and using the DRO on the Bridgeport makes the tolerances for bores and holes spacings above what you can get with a waterjet. GBX-130 requires even less machining due to the use of a Banebots gearbox in place of a custom turning gearbox, and would be used if the ETA on our manipulator prototypes was very short.

Of course it takes away resources from other systems; a WCD would take a little less work than this particular model. A KOP chassis takes less work too. Hovever, due to various things within our team using a swerve drive is advantageous, even irrelevant of the game.

I meant what is the encoder itself mounted to, not how is it connected to the shaft.

Is your Ma3 1:1 in relation to the steering portion of the module? It looks that its part of the first stage of the rs gearing.

Edit:

  1. If your pressing the bevel gear into the wheel, why not put bearings into the wheel and make the assembly dead vs live? It would make the bottom of your module way more rigid. Also much easier to do wheel swaps, we were able to do wheel swaps in under 10 seconds on our modules this year.

  2. I would recommend milling the side of your forks and adding a bearing to support the smaller bevel gear.

Looks pretty cool - if you’re worried about the cantilevered bevel gear, I’d look at the AndyMark swerve module. They cantilever their bevel gear too, and I haven’t heard of any issues with their setup.

Where do you plan on getting the bevel gears from?

A bent piece of polycarbonate.

That’s an error on my part. It should be an S4 incremental encoder, but I’m used to putting in an MA3 and forgot.

I tried 3/8" and 5/16" bearings on a dead axle, and basically I either had to use a 1/4"-20 axle, which is possible but not something I’ve had experience with, use a very weak bearing, or bore out the bevel gear hub more than I’m comfortable with. If a 1/4"-20 would work as an axle, then I would switch to that plus an 8mm OD spacer as a dead axle. 8mm because I would already be ordering 8mm/16mmm bearings for this.

I would add a bearing on the forks, but it would basically be a teeny tiny bearing with at most a 1/4" outer diameter. The thrust bearing cantilevers the bevel gear a lot more than I would like, but I couldn’t find a thrust bearing thinner than the one I’m using right now.

The Andymark Swerve & Steer, as well as most coaxial swerve drives, all cantilever their bevel gear. But to handle thrust loads they sometimes just use thrust washers, and none of them cantilever the bearing out as much as I do. I think I’ll change to a thrust washer instead of a thrust roller bearing.

how is that connected to the module? it looks like the bent polycarb goes to the CIM housing in the picture, but that doesn’t make much sense.

We would just velcro the polycarb to a convenient spot, in this case, the housing of the CIM. We’ve found that that works so for mounting encoders, however it is possible to mount it with a tapped plate and standoffs instead. IRL we would not use the CIM for velcro, rather we would mount it to a piece of the chassis or something; the polycarb is only in this model for weight and ascetic purposes.

Gotcha,

If you are using a s4, how are you zero’n it?

http://puu.sh/jxgeU/4eb20e2530.png

Hopefully this shows how we did it, we used the tube axle by VEXpro and used r6-5zz bearings. The forks also had a recess for the tube axle to sit into and it was held in with a 1/4" bolt. We liked this layout but we will be updating the use of the R6-5zz bearings to a metric flanged bearing to retain the wheel better. Atm the wheel can slide off the bevel gear, while it never happened during season it did happen to one of our pbot wheels.

Unsure how thick your forks are as well, we did .5" thick and had room for an 11mm OD bearing. Also used a thrust washer this year, will most likely be switching to a thrust roller bearing.

-RC

I was thinking of a hall effect sensor for zeroing, but we haven’t tested the precision on that yet so we need to do that first. I will put the magnet and stuff into the CAD as well.

I can see that; I didn’t think of putting the bearings in the face of the bevel gear rather than in the hub! A 1/4" bolt, spacer and flanged bearings should work if you used them and had success too.
Then the only question is how easy it is to bore out the gear rather than the side plates, and we just need to test that once we get our new lathe.
I don’t want to move to 1/2" plates for weight and waterjetting purposes (we’d have to waterjet a single 1/2" plate for the whole bot), although I went up to 3/8" already due to the way the bevel gear is set up. I can try putting in a 1/4" OD bushing or something, but switching to a thrust washer should remove the need for one by reducing cantilever. I need to test this out on a module to be sure; it’s always possible to upgrade just the forks. I love this design; it’s very modular. :smiley:
Did you experience any issues with the thrust washer, or did you just see the beginning of issues like minor wear? I used a McMaster 7806K63, but it’s very thick and increases the cantilever on the gear by a whole lot.

EDIT: Actually putting the bearing in the bevel gear in the hub would make it impossible to machine for us… good thing a face mount is possible.