Swerve Drive Modules

Hi Everyone!
Team 1515 is looking to build a swerve system in the 2011 offseason. Here is a rendering of the CAD so far. http://3.bp.blogspot.com/-lsCjQ1VnEso/TcYLfXDJ0xI/AAAAAAAAAG8/tFmivR_Kg4Y/s1600/photo.jpeg

I’ve been able to source nearly all the parts save for the bevel gears, a crucial aspect of a swerve system. For this design, we need 3/8"bore keyed gears. Does any team (preferably one who has built swerve) know where I might source these?

Also, any comments, or even better, criticisms on the sheet-metal design would be greatly appreciated. I’m considering adding in sides, bent over one side of the swerve, riveted over the next. Is this necessary? Oh, and the vertical shaft is not in this render, but it has all of the snap-ring grooves just as the horizontal shaft does. They’ll both (obviously) be turned on a lathe.


You can get bevel gears from McMaster-Carr. the website can be hard to navigate, but once you get use to it, you will find that they really have everything.

As for the design it looks good. I was the head drivetrain designer this year for FRC 79, and we did a coaxial swerve very similar to this. I really like the sheet metal concept. Our modules this year were open on 2 sides like this, but were made from 1/4 inch 6061 Alu plates. we never had any trouble with them.

If you have any further questions feel free to PM me, and I’ll be happy to help.

Yes. I am familiar with the McMaster bevel gears. my opposition is that they cost upwards of $30, and are relatively large. way too large for our needs.

Boston Gear makes some nice miter gears that might work for you. Check out this PDF.

Every swerve 1625 has had (alot) has used mcmaster part #6529K14
this has a 3/8ths bore, you’d have to broach a key yourself, we’ve had them hex broached typically.

Haven’t killed one yet.

Also advice on the module, take a long look at the path forces travel and what things need to stay perfectly aligned (bevel gears)

It’s going to be tough to get the bearing holes aligned based on the tolerances on the sheet-metal bends. This will be even tougher because your design doesn’t have a flat surface parallel to the bend line that the brake operator can use as a back-stop.

Correct bevel-gear alignment is VERY important in a design like this, especially if you’re using a “small” bevel gear. My advice would be to use the one Aren suggested (good size for this application).

You should consider making your bottom wheel-axle into a structural axle (make it into a standoff with both ends tapped, or (148’s favorite trick) make it 1/2" OD tube with a 1/4" ID, then run a 1/4" bolt through it. Making the axle a structural member in this way will GREATLY stiffen up your module. Right now I’d be concerned that the forces on the module will “unbend” the sheetmetal – you should do something to turn the module into a box structure.

Ohh – I almost forgot: don’t do a swerve drive. :slight_smile: The cost is greater than the benefit in almost ALL situations for almost ALL teams.


I agree with John here. From what I remember, you guys had some pretty big reliability issues with your drive this year; I think simpler is the way to go, not more complicated.

What is the thickness of the sheetmetal?

The concept looks good, but that metal is going to flex at the bottom. My team used 6061 T6 with 3/16 wall in 2008 and we had trouble with stress on the axle bending the modules with a similar design. I would try to reinforce it somehow.

What would you say is a comparable drive train in power and maneuverability like the swerve?

Power is a silly comparison, since it is based purely on motors. a 4-motor 2WD has the same “power” as a 4-motor swerve.

As far as maneuverability – I don’t think sideways motion makes a swerve drive robot more maneuverable in application than any other drivetrain.

Most teams don’t get the software right.
Most drivers can’t control it.
Most swerves tend to fall apart at bad times.

Build a reliable skid-steer drive and you’ll be just as good as a swerve in most games – I bet you’ll spend less effort to make it work, and I bet it is less likely to bite you in the butt at the wrong time.


Thanks everyone. Especially John. I’m going to fix that in the design.
And as to not building swerve, I too have seen it work and fail. Of course, 1717 has an exceptional design that we will not be able to match for at least a year or two.
However, We feel that this is a good way to open our team up to the sheet-metal world. If the swerve works well, we may build our 2012 chassis with SM. If we can form a solid relationship with a SM manufacturing shop, its worth a try.

@AdamHeard, yes we did. I’m going to do my best to get our team away from Mecanum drive.

If you do decide you really love swerve and want to make one, spend an entire offseason just mechanically making it, then set it aside and ignore it for a season. Then spend the entire next offseason making it drivable.

The biggest reason I’ll tell people to not go after swerve, is the control is never quite good enough, it lacks the direct mental connection most tank drives achieve and from that loses effectiveness.

I dont know the efficiency on the bevel gears, but i bet a 4 wheel omni drive with an rs775 and a cim on each wheel would be more maneuverable, and probably push better too. I thought disco-bots did something like this year.

I updated the design a bit to fit everyone’s suggestions. (I did not fix the drive shaft yet, though. I am a little worried that it’ll be difficult to assemble like this, so I may make the dual-triangles on the side into large rectangles with filleted corners.
You may notice that I removed the riveted sprocket from the top due to the fact that it is very obviously ineffective. I am going to eventually replace it with timing-belt pulleys from McMaster. Does anyone have an idea as to how I might attach these to the frame? http://1.bp.blogspot.com/-svOKXD2nI4k/Tcc-KEaziBI/AAAAAAAAAHE/eYkA_unY6Io/s1600/Swerve_03.bmp

Also, I know 1717 uses the main bent SM piece with open sides, in additional to two side plates which rivet on. This makes assembly way easier, but makes repair quite difficult. Of the two designs (one piece, three pieces), which is preferable? Also, would it be a terrible idea to weld the timing-belt pulley to the SM? I know it would warp considerably.

There is no way that you would be able to bend that module in one piece.

What method are you using to hold The output shaft to keep the module standing straight? This tends to be one of the more important things to get right in swerve drive. As it is i don’t see any method in place to hold the module against side force.

I’m not sure what you mean. The output shaft infers the bolt holding the wheel/sprocket, correct? What do you suggest I add to prevent unwanted sheet-metal bending?

I was actually referring to the shaft that the whole module pivots on, as it it i see it being held by one bearing at the top of the module, normally teams will have a second shaft around the drive shaft that the module turns on. Then this shaft can be held by some sort of bearing system that attaches to the robot.

Okay. Well, yes, in this design it is effectively held by one bearing. Where exactly do they place that second shaft? I can also, of course, just change it to a roller bearing, which are longer.