Feedback On Lazy Susan Swerve Design

Hey all, this season our team is experimenting with swerve. Here is our first attempt using a lazy susan design. We are using the REV absolute encoder for zeroing the steering. Please let me know any feedback you have on the design.

Driving Gear Ratios (4.8:1 reduction; 17.32 fps)
NEO: 3:1 VP Reduction
15T Bevel Gears
18T 9mm Pulley
24T 9mm Pulley
60T Belt

Steering Gear Ratios (7:1 reduction)
NEO: 7:1 VP Reduction
84T Gear (x2)

GrabCad Link

1 Like

Cool! Welcome to CD, and congrats on your design. I have some quick thoughts:

  • You don’t need the encoder on the drive axis. The built in brushless motor encoder will do for auto afaik.
  • I don’t think you have enough steering reduction. The COTS modules that use Neos for steering have about 20:1 reduction. Consider adding another planetary stage, or making the dining pinion smaller (although that would require a different steering encoder)
  • The cast aluminum lazy Susan bearings I’ve seen people use for turrets are not particularly strong. Do you know if they’re up to the job?

I would second being cautious on the lazy susan. Most swerves use 68xx series (thick section) or x-contact bearings. These are much better at taking combined radial and thrust loads, in addition to being stronger than bearings marketed as “lazy susans” tend to be. You can get them fairly cheap from AliExpress or FRC-focused resellers such as The Thrifty Bot.

Bevel gears want to push away from each other, and you have the driving one on a large cantilever which could lead to the teeth grinding or skipping in extreme cases. This setup is specifically not recommended for VersaPlanetary gearboxes, significantly derating them. This also applies to the module rotation gearbox.

Do you have specific capabilities in mind that you are designing for? Does your team have a mill, router, lathe, etc? Knowing this might help us evaluate the manufacturability.

Are you trying to design this to meet a specific goal of your team? Price, weight, size etc? You can probably reduce size quite a bit by using a modern bevel-beside-wheel architecture.

I do like the way you’ve combined small 2D parts to form the module. A clever design solution.

Putting standoffs on the turning VP is not a good idea. A lot of stiffness comes from reaction forces against the plate. Consider using a VersaPlanetary encoder slice instead so you don’t need the standoffs.

Overall I would definitely start looking at some COTS designs to take inspiration from, as they’re near their respective local minima with current technology.

2 Likes

On the topic of using a “lazy susan” bearing, I’m curious if @Andrew_Duerner/Armabot did any stress testing of their “Simple Swerve” design:

image

1 Like

It was mostly a software test bed, we didn’t field or run it for an extensive period of time.

3 Likes

A dual-y swerve? Don’t be ridiculous. The wheel scrub alone would never let it work.

:flushed:

3 Likes