Turning Motor / Gear Ratio for swerve drive

I am working on a early design concept for our team’s offseason swerve drive. I am wonder if anyone has had success with a certain motor / gear ratio config.

I’ve heard 100-200rpm as a bench mark speed, but don’t have any experience with that myself. As far as motor selection goes, the popular ones are 775pro, BAG, or 9015. There’s a lot of debate as to which is best, but really any of them will work if the rest of the design (gear ratio, stall protection, etc.) is sound.

We have used a bag and a 70:1 for the past two years. In the past we used a 100:1, but swapped to 70:1 so it was easier to turn by hand.

We have had good success with 9015s with about 180:1 gear ratio (including the final ratio between the gearbox output shaft and the turret). That would put our steering speed at around 80 RPM (if you use the free speed of the motor). The steering seems plenty fast. We have been using the same gear ratio for the past couple of years (as far as I know) based on what has worked in the past. We have really not done a lot of experimenting to find that sweet spot. But based on some recent discussions here on CD, I’m thinking of proposing a summer project to the team to run a few experiments with faster (and maybe slower) turning speeds to see if there is a better gear ratio to use.

The trade-off (at least as we see it) is between speed and accuracy. My understanding is that if you get the turning speed too high, control becomes an issue because of the inherent computational lag between the steering encoder response and the input to the motor. So, if the speed is too high, the control gets jittery with a lot of overshoot on the positional accuracy.

The problem with high gear ratios on the other hand is that you get a lot of backlash in the gears so that introduces lag due to taking out the backlash before you get any movement of the turret.

So the sweet spot is somewhere in between. I think the general consensus of the FIRST community is in the 100-200 RPM range, and that is probably where I would start from a design perspective.

The goal of the experimenting will be to find the highest rotational speed while providing stable response for the control system with no jitter and minimal if any overshoot. Minimizing power consumption would be a secondary objective. Since we use versaplanetaries for our steering gearbox, it should be relatively easy to swap out gear ratios to find the best response.

We ran 9015s with 100:1 planetaries and liked that setup this year.

BAG motor with a 90:1 reduction worked well for us in 2017. We based our design on a Robostangs swerve module which used a BAG at 50:1.

The sensor sample rate, loop rate of the PID controller, noise in the system, and delays in commands, and various other non-linearities all impact the maximum speed you can run the motor at without getting system instability. Something something nyquist theorem something…

In general, I think you’d want to have a gear ratio & motor combo such that full motor power is just barely under this maximum controllable speed. Then you’re getting the most out of the sensor/controller system (which probably has to stay somewhat fixed).

If your gear ratio/motor combo max speed is faster than controllable, that’s sorta ok, just don’t run the motor faster than you can control. And make some other assumptions.

However, if these are mismatched by a lot, it means you’re only using say the bottom 30% (or less?) of the mechanical system’s range. This is inefficient. Also, due to static & kinetic friction and slack, the system tends to get non-linear toward the low-power ranges, which can cause issue.

We ran 65.45:1 ratio off 550 (AM-9015) this past season. The motors did get hot after extended use, and it seems to steer plenty fast, and we think faster then it needs to. We are looking at adjusting to a larger ratio similar to the ballpark 2767 is in (100:1 off 550). Also looking at potentially switching to a BAG or 775 to reduce temperature buildup, plus the 9015s are discontinued now.

I think this application is a perfect example of something that could benefit from a 550pro type motor. I think if that motor existed, FRC robots long term would probably become all 775pro and 550pro.

I have added the 550pro to the COTS parts suggestion thread. Hopefully it becomes a thing in the next few years.

So much this. I’ve been wanting a 550pro motor for a long time, maybe even with a 5mm output shaft so that all the 775pro stuff fits on the shaft, because for a lot of things, 775pros are grossly overkill. We didn’t need two 775pros on our intake this year, for instance. A 550pro would definitely allow us to minimize weight while also using just the right amount of power.

Aiming for 80-100 RPMs of the wheel is a good range of controllable and producing a smooth driving output. 3419 has always used Banebots 550s or AndyMark 9015s, but I just noticed on the AndyMark site that the 9015 was discontinued, which was a bit of a surprise: http://www.andymark.com/Motor-p/am-0912.htm

775pros are overkill for this usage, and they take up extra weight and space. BAG motors weigh more, and you don’t need the extra thermal capacity in this usage - you aren’t working these very hard in a typical game, and they should really never stall.

We used Banesbot RS550 at 132:1 one year and have used the AndyMark 9015 at this ratio in recent years. This year we used a 9015 at 100:1. I feel that the 9015 at 100:1 is the sweet spot for our design.

Would a PG71 Gearmotor be too low speed (75 free) and too much torque(16.5ft-lbs)?

Andymark’s COTS swerve uses this and actually gears it down slightly more 48:40 after the gearmotor, so I think it would be fine (insert 148 this is fine meme here)

Actually it’s geared up 40:48. That gives you 90rpm free speed.

9015, 100:1, motion magic. Now that the 9015 is gone, RS550, maybe 70:1 but probably 100:1, motion magic.

Putting too much motor here (RS775) makes it harder to tune.

Motion magic was new for us this year for steering (Bag @ 70:1) and we loved it, Real tight.

(Prior years was BAG or 9015 @ 70:1 or 100:1 on Roborio PID) and was fine, but not as tight!.

I have to think about why higher torque 775 motor vs BAG makes it harder to control? (while generally deferring to 2767 experience) not sure I get this.

I’m not the control guru, but I’m told it has to do with rotor inertia and how fast the motor responds, in this case, too fast. Its like trying to put a 60 mph cruise control on a top fuel dragster. We had 2 teams contact us with tuning problems using 775s and the quick solution is to voltage limit the motor which, coincidentally, makes the motors look more like a BAG. I believe both teams changed the motors to lessor ones. I’m curious if others had issues and, if any, how they handled it.

This was NOT using motion magic.

I think this is what I was thinking - just reduce max voltage and go

I have been a Bag - and 9015 advocate but recently getting convinced on 775pro at reduced voltage - but appreciate your team’s thoughts.