Gyro Drift

[Jayd009]

Hello everyone,

My team (Fernbank LINKS) is trying to use an Analog Devices gyro on our robot for GRITS (October 22). We have been able to attach it to our robot, but every time we test it, the angle increases when the robot isn't moving, which is messing with our autonomous program. I don't know the exact model of the gyro, but I have posted a picture of it below. We have been wiring our gyro in the rate pins, and we have been using the correct cable. Does anyone know how to fix this? If not, does anyone have a different model that they would be willing to sell us?

Thanks,

Jonathan Daniel

[bdaroz]

We don't have that model on our bot (we have the one that came w/ First Choice), but I wonder -- is there a calibration routine that you may not be running?

I know with the FC one we have to have the bot powered up and still for at least 20-30s stationary and then trigger the calibration (which takes a few seconds) and then the drift drops to < 0.1 deg/sec and usually < 1 deg/min. (I can also say if the robot isn't still, or if it's calibrated too quickly the drift is higher, sometimes severely so.)

[Ben Wolsieffer]

This looks like the gyro used to be included in the KOP. We used that gyro in 2014 and 2015, and it worked well enough to be used for field oriented driving throughout the match, but it does drift faster than some of the other options.

When the robot code initializes, the gyro gets calibrated and the robot must not be moving at all during this time.

It is possible that the gyro is damaged, possibly by ESD or impacts. I believe we damaged two of these gyros while crossing defenses in 2016, although I don't know for sure what caused them to fail. They started drifting unpredictably, so we decided use a NavX instead, which worked very well. I have heard good things about the WCP Spartan board as well, but I have not used it.

[adciv]

Originally Posted by Jayd009

Does anyone know how to fix this? If not, does anyone have a different model that they would be willing to sell us?

Please see this thread for an alternative.