Gearing of Swerve Drive

I’m designing a swerve drive system and have gotten to the part where I must decide what gear ratio to use. I originally used a gearing system that made the drive train faster than the KoP drive train. I then considered the fact that having a drive train with more torque would give a it higher max acceleration which might allow a robot with such a drive train to be able to “juke out” other robots. I’ve discussed it with the design team but we haven’t reached a decision. Looked online but couldn’t find anything helpful. I figured I would ask here for any advice on this decision. Thanks for any help

1 Like

The general consensus is gear for speed, not torque, on a swerve. Within reason, of course. ~16fps seems reasonable to me, but others with more experience can chime in to confirm/deny.

We ran a gear ratio that resulted in 16.3 feet/sec free speed. That was with a full weight robot. I wouldn’t recommend going any more aggressive with gearing unless your robot weighs less. We heavily relied on SRX current limiting to prevent from browning out. Pushing force wasn’t good, but speed and maneuverability were both great. In our experience, around 13 to 16 feet/sec theoretical free speed is the sweet spot for swerve gearing.


I’m going to turn this around on you and ask what your requirements are. This will help you decide what*** you*** need, not what others have done.

1> what speed do you wish to go?
2> what size wheels do you wish to use?
3> what motor do you wish to use?
4> Do you wish to be torque limited, or traction limited?

With the above questions answered, you’ll be able to answer the gearing question yourself.

1 Like

What does a typical “cycle” look like for this robot? How far does it “sprint” between field elements on a regular basis? What is the most time sensitive driving conditions it will experience (autonomous vs. tele-op)?

If you know the answers to those questions, you can get a much better idea of how you should gear your drivetrain (any drivetrain, not just swerve). Swerve will have some ramifications upon this (namely that most conventional swerve designs will be limited to 4 drive motors, rather than the 6 or 8 motor possibilities for tank drive), which can put some limitations upon how much current you draw on a per-breaker basis. Additionally, the geometry of the swerve modules may limit the ratios that are able to be packaged conveniently (namely the final stage of reduction to your wheel, where the gear/pulley/sprocket can’t be larger than your wheel or else it will scrape carpet).

I’ve never heard this, and I’ve never built a swerve robot, but can someone tell me the reason(s) for this?

Seems to me, the best speed and sprint distance to fit a particular game will be the same no matter which drive train.

We ran at the following ratio:
12:100 32DP first stage reduction off 775Pro(s)
16:26 20DP second stage reduction
15:15 VEX Bevel Gear third stage.

Total reduction of 1:13.54

We used a 3" Colson wheel for our 2018 swerve.

Theoretical free speed of ~18.1fps

While this is all true, there really is no reason that “conventional” drive trains couldn’t have 6 drive motors across all the modules. We ran 2x 775pro drive motors per module for the back two while running 1x 775pro drive motor per module on the front and did not see any issues with the imbalance. And because of having 6 drive motors overall, we were able to gear for a high top speed while not losing on acceleration. We also did not run any kind of current control or voltage compensation for our 775pro drive motors, just ran them open with no issues all season.

We’ll be posting a more in depth whitepaper on our swerve that we developed last offseason and ran for the 2018 season and will go more in depth on all of these kind of topics.

1 Like

Short answer: if you’re in a pushing match, all you’re doing is wasting your time.

If you have a swerve drive, it’s a much better play to just skip around any defenders in your way rather than to push them around. If you want to defend, swerve is not the answer.

Tell that to 16 in 2017.

Defense != pushing. Pushing is generally not the best defense anyway. Swerves aren’t the best at pushing, but that doesn’t disqualify them from all sorts of defensive roles. Many swerves over the years have used their drivetrains to play defense that tank drives could not.

That said, you aren’t as durable as a tank drive, so be sure your swerve can handle the loading caused by being pushed, as a lot of swerve defense is about putting you in between the opponent and their destination.

1 Like

We had more time to play defense in the Semis when we were only going for 2 rotors…

This is fantastic advice. OP, identifying YOUR requirements then designing to those is a much better method of engineering.

We made ours fast, with the intent of using the extra degrees of freedom in the base to out maneuver our opposition, and to not lose cycle time by brawling it out. The idea of doing an ‘unconventional’ system only to use ‘conventional’ strategy didn’t make much sense to us.

On to gearing:

We geared for 20fps free speed on 1.875" colson’s. The wheels wore down extremely fast. We were at the edge of, or browning at out at all times. It was awesome, but a bit too fast.

Something closer to 17-18ish might have kept us free from browning, and given us better acceleration. Once we were at speed, we were pretty immune to heavy defense. We spent most of Ventura last year sprinting around defense.

May I ask how many/what kind of motors did you use for your driving and what the rationale was behind using the 1.625" wheels?

Looks like my memory isn’t the best right now. We used 2.5" Colson wheels, my mistake. When I get home I’ll add some images of the modules, and try to link our gearing sheet. Our team also has a “lessons learned” doc floating around somewhere, I’ll see what I can dig up.

Gonna revive this thread to clarify something. I’m using JVN Design Calc, and was wondering what the difference between Drivetrain Free Speed and Drive Train Adjusted Speed is.

From what i’ve heard, free speed would be off the ground spinning in the air no friction. I would guess adjusted speed would be the speed observed when there is friction.
Teach this to me if i’m wrong?

This is generally correct. Keep in mind that the “adjusted speed” shown in the JVN calc is an estimate, and by no means guaranteed to be accurate of how fast the robot will actual move in reality.

I generally like to talk about speeds and such in free speed; everyone seems to have their own parameters to decide what the “adjusted” speed is so using free speed is a good way to normalize things.


The “adjusted speed” is just the calculated free speed multiplied by an arbitrary reduction factor. It’s supposed to account for the fact that when the robot is driving, the motors are under load therefore spinning slower. The default of 89% is supposed to be empirically found to give a good estimate.

As long as you revived this thread, we have learned a bit since last summer. Our 2018 swerve drive was geared for a theoretical free speed of 17 ft/sec using mimCIM drive motors (6:1 gear ratio with 4" wheels). We switched to NEOs this year with the same gear ratio and wheel size (drop in replacement on last year’s design) which slightly reduces the theoretical free speed. But in side by side testing, the actual top speed is better than with the miniCIMs.

The consensus of our drive team is that with this gearing we were leaving some performance on the table. Their reasoning was that the NEOs have much better acceleration performance than the miniCIMs and therefore we were now geared too much for acceleration and not enough for top speed. From watching the matches as well as the testing, I can see where they were coming from. In addition, we won several pushing matches against WCD defenders (I know we are not supposed to get into pushing matches with swerve, but they did). So, this adds to their assertion that we were geared too low.

Unfortunately, our gearing uses a 3:1 VP off the drive motor and then a 2:1 from the bevel shaft to the live axle. Vex does not sell lower gear ratios than their 3:1 for the VP, so if we want to lower our gear ratio we either need to remove the VP gearbox and replace it with spur gears, or we need to change the 2:1 ratio after the bevel gear. They have chosen to pursue the latter and have CADed a 1.5:1 final drive ratio using belts. This would bring our theoretical free speed up to 22 fps. We are going to do some testing with this over the summer to see how it performs. The best part is that the belt and pulleys for this have nearly identical C-C distance as the spur gears, so we can swap this out on our current modules without needing to make any new parts.

Bottom line is that the NEO motors seem to have shifted our design point a little because the tradeoff between top speed and torque for acceleration is different for the NEO than for the miniCIM. This was not a result we were expecting but it is sort of exciting and we can’t wait to see how this little experiment works out.

1 Like

This match shows very good defense by 2910 using a swerve drive