Precision achievable with LSM9DS1 gyro ?

[PRebo.1]

In my project I have a LSM9DS1 with the purpose of measuring angular speed on the XY plane.

I'm using for calibration a reference platter spinning at a constant 33rpm, and the code handling the gyro removes a previously measured 0-state offset and denoises the reading using moving averages.

However, whenever I start it, I seldom get the same average readings. They might difer (the average) by +/- 0.02 DPS. Is this the precision limit, or something better can be achieved ?

[Eleon BORLINI · Accepted solution]

Hi @PRebo.1​ ,

the precision achievable with the LSM9DS1 gyro are defined in the datasheet, p.12:

This mean that, from a theoretical point of view, once you have compensated (or filtered) the angular rate typical zero-rate level, you are able to resolve your running angular speed with about 10 mdps resolution in the FS = ±245 dps case. So +/-0.02 dps is not so distant from the actual theoretical limit (about 0.1 dps), and is a probable value, considering additional noise. You can of course average in time, especially increasing the ODR (max = 952 Hz) so that you have more point for averaging in the same time frame.

Please note however that the LSM9DS1 gyroscope is not the best-in-class device as regards the gyroscope: the LSM6DSO(X), for example, is able to run at FS = ±125 dps, reducing the by a factor of 2 the minimum detectable signal, i.e. the precision limit.

-Eleon