atan in robotC

[technik3k]

@jpearman

Did you ever read this thread?
Gyro drift mechanical improvements

@FullMetalMentor

My kids recently put a gyro on their robot, which uses mecanum wheels. They screwed the gyro directly to the chassis between the front and back and the drift was 20-50 degrees after moving merely 5 or 6 feet.

They then tried your rubber baby buggy bumper technique :), combined with some added mass (in the form of a C-channel + lots of nylocks upon which the gyro was attached) and the gyro drift was reduced to only about 3 degrees over that same distance. I have no idea if that added mass was necessary, but some sort of mechanical isolation seems to help a lot. I heard MEMS gyros vibrate somewhere between 8KHz to 30KHz, so maybe (I’m just guessing) any kind of acoustic energy in that range of 8KHz to 30KHz could interfere with the gyro’s function. And maybe the whine/buzz/squeal of a motor is in that region.

In any case, isolation seemed to work for my chillun. Maybe it work for yours, too.

@jgraber

I also suggest mounting the gyro to the (massive) battery,
and rubber cushion mount the battery to the frame,
to create a mechanical low-pass filter.

Last year we had cortex and all MC29s mounted on a c-channel and then entire assembly was connected to the chassis through the anti-slip pads. There was no metal to metal contact even with the screws. This had fixed our problem of VEXnet connection frequently dropping out that was plaguing us at the beginning of the season.

When we decided to add a gyro we had mounted it to the cortex assembly on top of another set of anti-slip pads. So, without realizing it, we had the low-pass filter jgraber was mentioning above.

Since the thread mentioned MEMS, I looked up specs for the VEX gyro and turned out they use oscillating element which depends on Coriolis effect and it, indeed, will be very sensitive to the vibrations on its oscillation frequency.

@Collin Stiers

ok, I have done some more testing, the gyro drift is not caused by rotation at all, after carefully tuning the sensor scale for my two gyros, I did 5 rotations, and because the gyros do 3600 units per rotation, the result should be 18000
my results for the gyros were 18004, and 17984, averaging to 17994 for a total error of 6 out of 18000, which is well, well below, a 1% error, however, I can accumulate up to 10 degrees of error, just driving around for 30 seconds and completing no more than one rotation, so some how the error is tied to the robot physically moving and changing its position, not just a result of rotation or motor vibrations, any idea why that might be?

Collin, since you have two gyros, could you mount them in the same plane but 90 deg rotated from each other and run the test? If they use oscillating element it might be more sensitive to the vibrations along the specific direction. Ideally, you would mount gyros on top of each other in center of rotation of your robot which, hopefully, coincides with center of gravity as well.