Crab Drive Problems

We are doing a crab drive (the wheels swivel). We need to measure the angle that the wheel is at, so we thought we would use a shaft encoder or an unlimited turn potentiometer (non-contact). We can’t find any cheap absolute optical encoders, and we can’t find any unlimited turn potentiometers (that are still manufactured). Any recommendations?
Thanks :slight_smile: .

MA3 Absolute analog encoder, available feom usdigital i believe. We used this the past two year to detect where the wheels were pointing, works quite well

Did you see these continuous rotation potentiometers? they have a 20 degree dead spot…but they are inexpensive

Those look good, thank you for the recommendation!
It appears as if the ones that squirrel recommended are single-turn, though. Wrong link?:confused:

They are single turn, continuous rotation. The mechanical motion is 360 degrees (continuous), electrical is 340 degrees. They do not have end stops.

We used similar pots as encoders a couple years ago on drive motors

Just FYI, we had several failures of the MA3 in the past 2 years where they seemed to get zapped from static discharges. One of these instances cost us a match in the elims in Atlanta.

We’ll never use those encoders again, FWIW.

we had 4 on the 2008 bot, not a single issue, and 1 on the 2007bot no issue. we’ve begun using them instead of pots because of the reliability we’ve had.

We had a particularly dry (as in, not humid) practice area for the last few years which I think accentuated the problem. With the increase in static charge build-up this year, it may become more of an issue. Anyway, if you’re happy with them then great - I’m just putting out a warning that we had a lot of trouble with them.

What we had happen all the time last year is that we’d be driving around, then the robot would get close to a metal part of our field border, we’d see and hear the static discharge (large blue spark was visible), then our encoders would go haywire, sending semi-random but increasing values even though our crab wheels were not turning. This, in turn, caused our feedback loop to incorrectly move the wheels to correct, which would end up spinning them too far and ripping wires out etc. We added a fail-safe to prevent them from spinning for more than a second or two to address the symptoms but never were able to fix the root problem. Most often, power-cycling the encoders would fix the issue, but every now and then they’d be damaged permanently.

thanks for the heads up, we’ll keep an eye on them

It is also possible to drive a pot through a reduction link to allow you to use the pot’s less than full rotation for what ever amount of rotation you wish. If you need to have continuous rotation then this is not for you. However, you may find that a little over 360 degrees will fit right in and a simple 2:1 reduction would use far less than the 270 degrees most pots allow. Please remember to center the pot when the wheel alignment is in the center of it’s rotation.

Take a look at this one. It’s a fairly cheap absolute encoder with 360 degrees of continuous rotation. It’s 8 bit so you have 128 set positions.

Our team is trying crab drive and we’ve got most things ironed out, but we still have a few questions.

The MA3 looks excellent, but the reliability thing is disturbing. Out modules have unlimited spin, so they won’t cause permanent damage to the robot, but they can really mess up a match. Would putting the encoders in a makeshift Faraday cage solve the problem with static discharge?

And second, a more mechanical question. What RPMs should these modules be going at? Our team was shooting for 120RPMs, approximately a 1:2 gearing from the denso window motors, and a similar setup on a globe.

Will there be any issue with using different motors for different modules as long as they are left/right symmetrical and all in independent closed loop control systems?

In previous years we have gone with a 1:1 ratio from a globe motor to our wheel boxes to turn them. Turning faster would have been nice, but we were always limited by the torque needed to turn grippy wheels on carpet.