Does a gyro need to be placed at the center of rotation?
I just checked this by "thinking" about that and noticed, that for an ideal rigid body (!) it doesn't matter where you place it.
I just imagined that like this:
- You have a gyro at each side of a stick (axes of the gyros parallel to each other).
- You rotated the stick about one end (e.g. x axis by 90°).
- Observation: the other axes is also rotated 90° about the same axes, same direction, seen from center of the other gyro.
Theory vs. Practical "Rigid Body":
- You won't have an ideal rigid body. Your material is at least slightly flexible...
- If you want to combine data from 2 sensors, e.g. accelerometer / gyro the axis of the two sensors should be pointing in the same direction (or you need even more sophisticated math). And this coupling should be very tight, e.g. not flexible or swinging (else you'll measure a lot of garbage); Rule: the longer the distance the more influence of those effects.
You possibly want to measure the orientation/position of something (e.g. wings of a plane or the "body" of quad copter). So the gyro should be fixed to exactly that.
PS: Thank's for pointing me at this issue. This helps me a lot (need those sensors for research).
Yes, there is a reason for this. For example, consider pitch of an airplane. If the plane pitches up 20 degrees, then the angle at the center of the plane is 20 degrees. If the gyro is mounted in the nose...still 20 degrees. The entire plane is pitched up 20 degrees, so it doesn't matter where you measure this. So this leads many people to assert that the mounting point doesn't matter.
HOWEVER, notice that when the gyro is in the center, it is experiencing 1g all the time, independent of the pitch-up. If the gyro is in the nose, it will experience positive g force while pitching up, and negative g force when the plane is pitching down. Typically gyros will have some sensitivity to both g force and vibration, due to minor asymmetries.
Consider for example this application note from Analog Devices: http://www.analog.com/media/en/technical-documentation/technical-articles/MS-2158.pdf
According to this, the g-sensitivity for cheap gyro (and anything in a model airplane will be "cheap"), can be about 0.3degrees/sec/G. Well, that is not a lot. If your plane nose experiences 3g during a 20-degree up-pitch in 0.5sec (that seems like a rather violent bump), then the rotation is 40deg/sec, and the error term due to g sensitivity would be 3g * 0.3 or about 1 deg/sec. That is only 2.5% error. Well, that is not a lot, but not insignificant either. And as you can see from the application note, other components have less G sensitivity.
So the bottom line is - yes there is some theory behind it. If you want to be conservative, mount it in the center. But I doubt you would notice the difference in typical practice. Also, different MEMS components are different, and it would take some research and calculation to figure out the precise significance in your application.