Quote:
Originally Posted by AdamHeard
I would not recommend trying anything other than absolute feedback geared one to one with the wheel.
Why?
|
Because we've tested robots that violated that, and they caused lots of trouble.
Early in our swerve testing, we didn't store the offset from a known direction. So our absolute feedback was really incremental. Every time you changed batteries, rebooted, pushed code, or suffered a brown out, you had to make sure the wheels where pointed in a specific direction. That gets old really quick. Also different people would set the wheels slightly differently, or worse incorrectly.
We had another test chassis with the feedback geared 2:1 to the direction. On this one, you still had to start the robot with the wheels pointed close to straight. The robot always knew the axis of the wheel, but not which direction it would go. It was a real joy, because we tried to count the 2.5v crossings of an analog 0-5v signal. Sometimes it would miss count and the wheel would start dragging.
I cringed when I saw Andymark release their swerve modules this year with non1:1 gearing on the feedback.
So I always recommend to teams thinking about a swerve drive:
Use absolute feedback and gear it 1:1 to steering rotation.
Analog or digital signaling shouldn't really matter much. But the analog sensors quoted above are the cheapest I've seen.
To Bryce2471:
We do scale the analog voltage from the datasheet to 0-5v. It still twitches, but I haven't really tried to fix it. I agree it should be an easy fix, unless maybe the sensor has trouble pulling the signal quick enough.