I have some weird behaviour that I am observing on an ICM-45686 gyro, where the Z Axis gyro readings are giving me some data that I do not understand.
Initially, when just looking at the long term behaviour of the gyro outputs over a ~8 hour span to perform some allan variance testing.
However, the Z Axis outputs puzzle me. I observe the bias wandering around quite a lot compared to the X and Y Axis. Additionally, the noise behaviour is very weird. I attached a histogram of the recorded gyro readings of the different axis, and while X and Y are somewhat approximating a Gaussian, the Z axis has some very weird distinct peaks and valleys.
My setup is fairly simple, just a PCB sitting on my desk while I am away with the ICM soldered on it, sending out the data through uart to my PC. I am running the ICM-45686 in low noise mode, at 800Hz sampling rate.
The layout of the PCB is done like in the datasheet, with 0.1 uF caps at the Vdd and Vio in lines. These are connected as well, as there's only one 3.3V supply on my PCB. Could noise on the power line or soldering (done ourselves, no pick n place, etc) have this effect?
The setup is not temperature controlled or anything, so that might influence things. But I am curious about why I see this behaviour on only the Z Axis and not the others. Is that one more sensitive to things like these?
Hi,
First of all, having your PC, fans, USB cables, or other equipment on the same desk can introduce mechanical and electrical disturbances that show up in long-duration gyro measurements. Also, Z resonator can be more susceptible to certain mechanical stresses or board bending. If your board layout or attachment applies uneven mechanical stress, one axis can be biased more.