General Swerve advice

Hi, I’m working on designing a swerve module for my team and I’m looking for some tips from teams who have experience with swerve drive who could share any of the knowledge on what works and what doesn’t.

Here are some resources me and my team have put together:

Design Basics

Thread based off of above presentation
Design and Programming

2018 Modules

1640 is also a great source

If you need any advice on your designs feel free to email me at: [email protected]

1 Like

I’ve designed, built and drove a few different swerve designs, I’ll try my best to consolidate some starter information.

There are really two ways you can go for deciding what you’re going to use for your translation. Either Cim/Mini-Cim, or 775 pro. The 775 pro route will add some more variables in the equation and somewhat complicates some things when trying to get your mechanism up and running for competition. I would recommend starting with a Mini-Cim or Cim swerve even though it will add weight to the module, for the simplicity it gives, it’s worth it for your first design. The challenge with swerve is transmitting the power from your motor, to a rotating wheel. By far the most popular way to do this is with bevel gears. This is achieved by having a bevel gear centered directly above the rotational point of the module, and another on a shaft that will rotate around the middle one. It is one of the easier ways to transmit the power, both for design and manufacturing. Vex sells bevel gears you may be able to use.

The main motors that have been used for rotation are the 9015 and Bag motors. Both of these are good choices, and would do well geared around 100:1. They are simple, and the 9015 has flexibility with multiple different gearbox choices you can use from either vex or AndyMark. The more choices the better, as it adds flexibility to what you end up doing in your mechanism. There are a couple ways to do this, some teams will put a big gear around their wheel that connects to a gear the motor is attached to to transmit the rotational power. Another option is to use a 775 pro for rotation. If you can handle the reduction needed, and feel capable with maintaining the motors, it’s an option. But once again, for simplicity 9015 and Bags are your best bet in most scenarios.

I would recommend Colson wheels, they have varying thicknesses which will allow for more flexibility, and choices when determining thickness & surface area to improve autonomous accuracy. The VersaWheels are also a popular choice for some swerves, although wear much more quickly, and need replaced often. Wider wheels are preferred by some teams as more grip removes some uncertainty on the carpeted surface, but it is harder to account for while designing. If you have an easy swap wheel module, then it wouldn’t be an issue, but if you want to not have to worry about it as often, definitely go with the Colson option.

You are going to want at least 1 encoder on each module, for your rotation. This encoder should be geared 1:1 with the physical rotation of your module to simplify all guessing on orientation of the wheel. The less needed but suggested encoder is for translation. Adding one on each module will give you better distance measurements than time, and will assist once you want to get into more complex auto routines. This encoder can be geared to whatever ratio, as long as the max rpm the encoder is rated for is higher than what it will be spinning at on the module.

No one designs anything entirely on their own. Inspiration for swerve is everywhere, CD is a great place to start, some have websites with more information. Look over swerves that teams have used in season and have posted materials on them.

Jack in the bot

Bomb Squad

Triple Strange

Sab-BOT-age

Stryke Force

There are totally more… hopefully someone will feel like helping me out and adding to this list. Looking at their materials, and what worked for them will help when designing your own.

I’m open to questions, i’ve built and driven swerves from pros to cims, slip ring to unicorn, and beyond. Post your designs, or message me on CD, both me and the CD community will be willing to help out.

I recommend using versaplanetarys for turning, 1640 has said the andymark gearboxes have a lot of backlash. If you are driving with a 775 pro, current limit it at 40 amps (this is what 2767 does).

There are a bunch of design tips in the presentations I linked above.

General swerve advice:
Don’t try it for the first time during build season. Right now, it’s the offseason, so knock yourself out, but in January if you haven’t built and programmed a swerve drive prototype yet, that’s going to be a really good time to set that drive aside and work on something else for a couple months. You can always use another offseason to finish…
Of course, that’s the standard advice for just about any drivetrain more complicated than a skid steer, and there are exceptions. But it can take a couple of years of practice to get up to the level of being able to build a swerve and whatever goes on top of it in under the build season time limit.

I see you are from MN. Check with 2169.

Otherwise, look to Michigan - some of those teams have it down pat.

Now, beyond that, please note that developing swerve will take some time and considerable resources. However, if done correctly, it is an incredible drive train. Seeing that your team has had some success in recent years, it may be something to try - but don’t put your eggs in one basket. The fabrication of the modules is one thing, but programming it is another. I wish you the best.

Owen’s links are awesome.

Some more reference material.

https://www.chiefdelphi.com/media/papers/3468

Contains a history of our swerve along with software link and CAD link.

You are going to want at least 1 encoder on each module, for your rotation. This encoder should be geared 1:1 with the physical rotation of your module to simplify all guessing on orientation of the wheel.

How would you recommend doing this? The way I have the rotation set up the module is turned by a pulley riding freely on the translation shaft.
If my description didn’t make sense, here is a link to the file: https://cad.onshape.com/documents/6d9a22f2266beb1ba70c1a2f/w/81a3ba0e55a7cf68196ec968/e/0b340b98aa8c3091e25a4201 It isn’t finished. The large pulley is fixed to the module.

Also, Thanks for all the advice!

2767 is the current God of swerve

1323 (wish they would share more, but hearing their are going COTs!) and 2910 are runners up

honorable mentions - always to 16 (swerve hall of fame member - along with others: HOF members: such as a 1717 (pause and bow down) 111, 118, 148, 1640,2767, ?

Based off your design I see two potential options:

  1. After the versa-planetary gearbox use don’t have any reduction like in the design I linked above and just use a versa-encoder
  2. Make custom gears at the same ratio as what is between the two belts used for turning that are attached to the encoder (like 1533)

On a side note, it seems to me if you attached the turning motor directly to the lower of the two plates with a 70:1 ish (thats just what we use) ratio you can then have the belts 1:1 and use a versa-encoder, which is probably the easiest option. It would also get rid of that set of gears, which would reduce a lot of weight.

Agreed

2 stage versa planetary with the encoder. 1:1 belt. Azimuth spur gears go away.

I don’t feel comfortable with using just the drive shaft to support the module. Might bend with an impact. We have bent non hardened steel ones in the past and they weren’t supporting anything. Refer to how others put a large diameter bearing or a bushing to support the module.

by drive shaft do you mean the vertical powered shaft, or the horizontal one with the wheels on it. We’re using thrust bearings on the vertical shaft. If you mean something else may you please clarify? I looked at some, but didn’t see what you meant.

One of the major design decisions that you will need to make is how you plan to allow the rotation of the module. Our module design has incorporated a double row of ball bearings above and below the main fixed plate. We cut the bearing grooves into our plates with a lathe which is something we have the machining capability to do. Other designs have incorporated a large diameter COTS bearing into their design. This is also a good design and should provide ample load carrying capability allowing the module to rotate freely when loaded.

Another major design decision is whether you will keep the drive motor onto the stationary part of the module and transmit it’s torque into the rotating part of the module through gearing (i.e. the bevel gears that were mentioned earlier), or whether you will put the drive motor onto the rotating part of the module and then transfer the electrical power to the motor through a slip ring. I have seen successful designs both ways. Our preference is the bevel gear approach, but we have toyed with the other approach a bit (at least in CAD).

When you study the ‘prior art’ designs, pay attention to these two features of the designs. The rest of the design is generally dictated by these two decisions.

The vertical one. If the wheels hit a bump or get pushed sideways the vertical shaft will try to bend in the area between the lower plate and the 42 tooth pulley ( a little cantilever action going on). Gut feeling is the vertical shaft is suspect to getting bent from an impact. Say… an impact that might happen when running over the electrical wire protector that powers the scale. All the designs in Owen’s presentation have a large diameter thrust surface or bearing between the mounting plate and module.

This is absolutely key. I tried to make a big deal out of this in my presentations because this is a common mistake for newcomers to swerve. Hereis a picture of a cross section of what 2767 does. You can see they use the two large bushings and the drive shaft goes through the inner most bushing.

Side note, with this design, you need something that prevents the inner bushing (grey in image above) from falling out of the outer bushing (white). What 2767 does I believer is just a c clip on the inner bushing with a washer underneath.

*If the link doesn’t work the image is in the design basics presentation

You mentioned a 70:1 gearbox, just wondering what motor you were using for the rotation. I’ve been going with a 775 motor. I also was having it at ~20 Amps, planning on using a 30 Amp breaker (plus a safety factor) so as not to use up all of the 40 Amp slots, is this reasonable?

We use a bag without a current limit, and have for a few years, they have worked flawlessly for us. A 775 is probably overkill for just the turning motor, but I know there are people who say the extra power is needed. I can’t off the top of my head remember which slot on the PDP it is plugged into, but I can check tomorrow and edit the post.

I was talking with the mentors of Team 16 at Houston Champs last year. They said it took them 5 years of refinement to make their design effective for use in competition.

You mentioned using a 70:1 ratio for the rotation. I was planing on using a 775 motor on a 30 amp breaker. I was planning on much larger ratio. I was wondering what type of motor you are using, and if it is reasonable to use a 30 amp breaker. My reasoning behind this was so that the drive train didn’t take up all the 40 amp breakers.

As I mentioned above, we are using a bag. You could go up to a 100:1 easily but even with a 775 pro you shouldn’t need a higher ratio than that. (General rule of thumb is 150-250 rpm on the turning motor). I can’t speak to whether or not a 40 amp breaker is needed.