Cheap Swerve Attempt 3

Seems like the night before kickoff is “everyone release their pre-season swerve design” night! Here’s what I’ve been working on in December.


Building off of my previous, attempts to design a “Cheap” swerve module that teams can build at home, I was inspired by the fact that the Thrifty Bot Swerve does not use an X-contact bearing, but instead uses a lubricated aluminum plate sliding on delrin to take the thrust and radial loads. Using that idea, I came up with the design below. It has some advantages:

  • The delrin thrust bearing saves the cost of a $35 X-contact bearing. (It could be redesigned to use a KA045XP0 if that made people more comfortable. I just haven’t found a cheap supplier for the larger size.)
  • The bearing is down at the wheel axle, where the bending forces are minimized. Having the bearing there also makes the wheel forks simpler.
  • All components are either 2D plates (can be routered, Wjet or lasered), low-load 3D prints, or COTS components.
  • The bevel hex shaft is simple to make, even with a file.
  • The gear train is totally enclosed.
  • It uses a 4" wheel so it can handle regular field obstacles like the BOUNDARIES. In fact, it can use any COTS swerve wheel you like, including the SDS, WCP wheels. (even the TTB 3" wheel if you can modify it for a dead axle).
  • I’ve designed it for the promised 4" ID Lamprey absolute encoder, but it could be easily adapted for other encoders or homing switches.

Overall specs are about the same as my previous design.

  • Size is 6" wide, x 7" long x 8" tall.
  • Weight is 5.1 lb
  • Cost is roughly $250 USD with the UP gearbox, but without motors or costing the custom parts. With everything it’s closer to $450. That’s pretty competitive with other modules, but I’m actually disappointed - I thought sure I could get this down below $200. More thought is obviously required.

As always, feedback is appreciated. If you decide to build all or part of this design, please share your results!

Bill of Materials:
CX20-swerve3-a002 BOM 2.xlsx (17.9 KB)

CAD:
cx20-swerve3-a002.pdf (273.7 KB)
STEP and Solidworks files.

Images:


13 Likes

I took a look at your BOM to see what your big cost driver are:

$75 of your $250 (30%) is in the bevel gear set. Not much you can do about that unless you are willing to completely change your architecture or source your bevel gearsets.

You have $22 (9% of your BOM) in spacers. You can bring that down a lot. You are quoting $12 for Qty 4 of you 1.25" spacer. Maybe the price is higher up in Canada, but assuming you bought 16 (enough for 4 modules) they would be priced at $1.31 each or $5.24 per module not $12. But there are a lot cheaper options for these spacers. For example, 91780A044 is a lot cheaper than the 92510A434 for the same length and comes pre-tapped for 10-32. You should be able to source suitable alternate spacers for the 8 spacers that you have in the BOM for less than $5. ($17 BOM savings)

You can source the 80T HDT belt from V-belt guys.com for $2 ($7 BOM savings)

If you have room to change your 8mm x 19mm x 6mm flanged bearings to 8mm x 22mm x 7mm non-flanged bearings, then you can buy skateboard bearings for these which you can get really cheap (plenty of good options under $1 per bearing on Amazon). If you have pocket rather than a hole, you can trap the bearing without the need for the flange. ($7 BOM savings)

Looking at the pictures you posted, It looks like you have added a bearing to either support the motor shaft or support your 3DP housing. I don’t think that is necessary. The motor shaft should have adequate support without that bearing for the gear mesh loads and that housing is going to be plenty stiff to resist flexing without the bearing support. You should be able to eliminate that bearing from your BOM (I’m not sure which one it is).

On the flip side, I don’t think you will be happy with bushings rather than a bearing for your azimuth rotation. I think you can integrate grooves for delrin bearing balls into your azimuth plate, bottom plate and top plate to create your own azimuth bearing without a huge re-design to your module. We cut our grooves on a lathe, and you are clearly planning on using at least a CNC Mill, so it should be within your machining capabilities to add that. We use 1/4" balls for our home meade bearing, but you might be able to get away with something a bit smaller if you wanted to. You will need to make the OD of your 3DP housing below the belt pulley feature where it attaches to your module plate a bit smaller to make room for the bearing race groove in your plates, but it looks like that would be possible with your design (although you might need to lower your wheel just a bit).

I’ll keep studying this and see if I can spot any other areas with obvious cost savings. If I do, I will add them to this post.

1 Like

Good insights! Thanks especially for the deep dive into costs, since that’s what I really want to optimize.

The bearing between the motor and the gear housing was added because I was unsure about slop from the Delrin thrust bushing causing misalignment between gears. With some experience, you could probably delete that part yeah.

1 Like

We have been experimenting with making our 60t bevel gears from light weight materials. So far natural white Delrin is working very well. I have posted some our photos and the Fusion 360 files to this thread.

2 Likes

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.