How to Mount a Gyro

[duffany1]

Okay... So we understand how the gyro works but we are having trouble figuring out where to mount it and how to mount it. If anyone can provide a detailed description and pictures possibly that would be wonderful. Thanks in advance!

[bladetech932]

If you are trying to measure the angle of the robot you are going to want to mount it closest to the middle of the robot on a platform that insulates the gyro from anything else (ex. wood,rubber,etc)

Last year we put it in a wood box and mounted the box to the frame.

[DonRotolo]

Mount it firmly to the robot - some vibration is OK but letting it move relative to the chassis is bad. As above mentioned, the center of the chassis base is a good idea, down low is good.

We have used velcro to good effect, allowing for easy mounting and removel when necessary. We have also used small non-metal screws (there are mounting holes) and even zip ties successfully. On a practice bot, we held it down with duct tape.

[WizenedEE]

Why does it matter how close to the center of the robot they are? I've been trying to figure that one out for a while.

[omalleyj]

Originally Posted by WizenedEE

Why does it matter how close to the center of the robot they are? I've been trying to figure that one out for a while.

Think about an extreme example where the gyro very far from the axis of rotation. Its a yaw rate sensor where the amount of rotation is determined by integrating the angular acceleration. You can indirectly compensate for the increased acceleration by changing your mv/deg/s constant, but you will reduce the maximum deg/sec you can measure before you out pace its ability to keep up.

[Ether]

Originally Posted by omalleyj

Think about an extreme example where the gyro very far from the axis of rotation. Its a yaw rate sensor where the amount of rotation is determined by integrating the angular acceleration. You can indirectly compensate for the increased acceleration by changing your mv/deg/s constant, but you will reduce the maximum deg/sec you can measure before you out pace its ability to keep up.

The datasheet for the ADXRS150 says the null has a linear acceleration effect of 0.2 degrees per second per g.

So if the gyro were mounted, say, 38/2 = 19 inches away from the center of rotation of the robot, and the robot were rotating at 180 degrees per second, you could calculate the effect:

180 degrees/sec = 3.14 radians/sec

(3.14^2)*(19) = 187.5 inches/sec^2 = 15.6 ft/sec^2 = 0.488 g's radial acceleration

So (0.2 degrees/sec/g)*(0.488 g) ~= 0.1 deg/sec

[omalleyj]

Originally Posted by Ether

The datasheet for the ADXRS150 says the null has a linear acceleration effect of 0.2 degrees per second per g.

So if the gyro were mounted, say, 38/2 = 19 inches away from the center of rotation of the robot, and the robot were rotating at 180 degrees per second, you could calculate the effect:

180 degrees/sec = 3.14 radians/sec

(3.14^2)*(19) = 187.5 inches/sec^2 = 15.6 ft/sec^2 = 0.488 g's radial acceleration

So (0.2 degrees/sec/g)*(0.488 g) ~= 0.1 deg/sec

Ouch! I had never really considdered how much the rate of spinning dominates the terms as it rises. So if you have a very fast spinning robot (you know who you are ) even mounting the the gyro pretty close to center will lead to more accumulated error than you'd want if trying to go with a field-centric control, say. (and of course you need to either scale the output or go with a gyro that won't hit 5v before you hit maximum spinning speed)
Thanks for the analysis!

[Ether]

Originally Posted by omalleyj

So if you have a very fast spinning robot (you know who you are ) even mounting the the gyro pretty close to center will lead to more accumulated error

The closer the gyro is to the center of rotation, the smaller "r" is, so the radial acceleration becomes negligible.

[Alan Anderson]

Originally Posted by WizenedEE

Why does it matter how close to the center of the robot they are? I've been trying to figure that one out for a while.

For most yaw rate sensors, in most situations we're likely to encounter in FRC, it shouldn't matter. The entire robot is turning at the same rate. But in extreme conditions, linear acceleration can interfere with the ability of the sensor to measure the yaw rate. If you put the sensor far from the center of rotation, centripetal acceleration could result in inaccurate values being provided.

For example, the ADXRS300 datasheet says that a linear acceleration will cause a false response of 0.2 °/s/g. It's not much, but it introduces a rotational bias that can look like gyro drift as the sensor turns first one direction, then the other.