Pic: Differential Swerve Module (971)

Maybe I’m looking at something wrong here but I see a final drive of 84/14/4.6 which is 1.3:1. In order to drive the wheel you have to go through that giant overdrive which is not nice to the gear efficiency gods.

Thank you very much for posting this Karthik. Its amazing to see the evolution of a innovation like this over the years. To Ian, 1114, and everyone else that was involved in that build, serious kudos for your ingenuity and creativity.

And to 971, Aren, and Ginger, and your incredible take on this idea, just wow. Can’t wait to see this on an actual bot when the game is suited for it.

I absolutely love how this community leap-frogs over itself year after year, constantly making the competition and themselves better every time. If that’s not inspiring, I am not sure what is.

Turns out with a Haas Mini Mill, a 3/64" ball end, and time, you can make pretty much anything.

CAD will likely get posted when the rest of the module is complete, TBD on that one.

In this case if I were to update the speed it would be reducing the ratio before the 84t gear, it’s just like gearing a tank drive faster, it gets harder to control, but we’ve got a pretty solid crew for that.

Good catch, I intended to have 46:12 listed, missed the 2.

Overall will likely be a ~9:1 reduction to a 12:84 at the module gear, then the final 46:12 upduction to the wheel.
The gear up is a very critical part of having a differential swerve be a feasible thing to control, the other option is to use larger wheels and nobody wants that.

-Aren

Did you have to remove material from the side of the Colson’s to make clearance for the bearings?

In that case, how do you calculate the proper tooth profile for the flat gear to mate with the bevel gears? Or do you just use a “close enough” approximation?

Yes :stuck_out_tongue:
I thought it was pretty funny at the time about how much of the colson wheel we carved out from the sides. I had a good laugh over the ridiculousness of the packaging, but that’s what Aren is best at!

This takes the cake for the most insane thing I’ve seen in FIRST, ever (in a good way :)).

Nice work to all involved. Seeing stuff like this makes me want to do crazy stuff too :cool:.

Aren and 971,
beautiful indeed. I think I recall you making something very compact and similar, around a CIM motor several years back?
If youre willing to share more of the specs, I would love to have our team make this.

Aren, do you plan to re anodize the 84 tooth gears now that you’ve cut bevel gear teeth in them? If not, I wonder how long the teeth will last (even the hard anodized ones rarely last a season for us).

And on a related note, how do you get an adequate amount of grease into this tightly packed differential? :ahh:

I need to contact the plating place and see how the coating already on the 20dp gear will play with going through a hard ano process.

Getting grease in isn’t my current concern, that’s easy enough. Dealing with carpet fuzz etc getting in there is more of my concern as we have blind pocket bevel gear teeth.
We’ll probably get some printed parts in there guarding the gears and making it appear more magical.

-Aren

That’s pretty cool!

I wonder how well it would hold up in a drivetrain, it would be neat to see some tests. With the bevel gear on the wheel axle being so small, you’d have some pretty large forces transmitted at the bevel gear interface. That also puts a large thrust load on the two large gears, deflecting them apart. I wonder if the u-shaped bearing profile is enough to keep the two large gears in place, or if they could deflect enough to wear the teeth out quickly?

It will be interesting to see how the rest of the assembly is designed too, it looks like a fancy little mechanism.

So the little bevel is pretty similar to the 12t steel mcmaster gear I’ve used on many a swerve in the past, and I believe 2451 ran at the final drive for their wheel in 2015, I have made that one out of 4140 regardless. The tooth profile on the aluminum side is also very thick and well supported as one end of the tooth is still attached to the base material.

The bevel profile itself is a custom one I generated in gearteq to ensure fillet radii were big enough to machine, and I also put in a Zerol profile for more rolling contact tooth to tooth.

The SG15U bearings are actually double row angular contact bearings, so I’m hoping they handle it well.

Good to see someone with similar concerns, figured I’d let you know the steps taken to mitigate them (obviously need testing to validate)

-Aren

The real question we need:
Will we see a swerve in 2018 out of 971? (Assuming no stair climbing challenge or anything that requires a terrain-focused drive train?

2015 sure was a tough year on drivetrains. :stuck_out_tongue:

Good info, it sounds like a lot of thought went into the design of the components. I hope you’ll be able to get a full prototype driving to see how it works and record some delicious data.

To make it past the prototype stage to a real robot, we’ll need some tweaks to the CAN stack to hit the timing this thing requires. Unless FIRST makes a game which only needs 2 controlled DOFs and we can run the whole thing off PWM.

Can you elaborate a little? What is the cycle speed you think this needs (*4), and what is the current CAN limit? What’s preventing the 8 drive motors from running on PWM and the remaining mechanism motors from running on CAN?

Who says you need 4 modules? :slight_smile:

1 Like

AFAIK you can adjust the timing on Talons to be 1ms control frames, unless you’re talking about the way CAN is handled inside the RoboRIO.

This thread makes me so happy. The design, machining, and photography from 971 is always cool to see, but pairing them with this awesome design makes it just too cool. :smiley:

Anyways, I was wondering the same thing as Brendan:

What is the cycle speed you think this needs (*4), and what is the current CAN limit? What’s preventing the 8 drive motors from running on PWM and the remaining mechanism motors from running on CAN?

Will we see a swerve in 2018 out of 971? (Assuming no stair climbing challenge or anything that requires a terrain-focused drive train?

You can do swerve and climb stairs. It would be the final step in the transfer of advanced swerves from 1114 to 971:
Ian’s differential swerve to Aren’s, and SimSwerve (liftable modules) to… SpartanSwerve? :smiley: