What is the noise of the accelerometer in g/sqrt(Hz), of the magnetometer in gauss/sqrt(Hz), and the gyro in dps/sqrt(Hz)? The offsets for the gyro and the compass in the data sheet seem unreasonable. Should the units be mdps and mgauss?

[Eleon BORLINI]

Hi @NathanCrow​ , which part number are you referring? LSM9DS1? And are you complaining for the unit of measures or for the noise/offset values? Btw please note that the noise (RMS) has a "g", "dps", "gauss! and refers typically to a BW of ODR/2, while the noise (spectral) density, which is the noise power per unit of bandwidth, has a /sqrt(Hz) unit. Regards

[NathanCrow]

I am asking about the noise specs for the LSM9DS1. The datasheet does not mention noise at all. I am planning to use each of the sensors on their most sensitive settings (+/-2g, +/- 4 guass, and +/245 dps). Knowing where the noise floor is would be very helpful.

I would be happy with either an RMS value and a bandwith (for example milli-gauss at an ODR of 238) or a spectral number (eg milli-guass per root Hz).

I was also confused about the offset numbers. The datasheet lists +/- 30 dps as a typical offset for the gyro. This seems really high (even on the +/-2000 dps scale). The datasheet also lists +/- 1 gauss as a typical offset for the magnetometer. This also seems really high. Should these numbers real be milli-dps and milli-gauss?

[Eleon BORLINI]

You can consider the noise as the minimum detectable signal minus "epsilon", where epsilon is a very little number. The minimum detectable signal is the LSB sensitivity, that is for example 0.061mg in case of FS +-2g (accelerometer), 8.75mdps in case of FS +-245dps (gyroscope), and  0.14mgauss in case of +-4gauss (magnetometer). This can be considered as a white noise independent on the ODR (of course, the noise spectral density is dependent on the BW and refers to ODR/2).This is the best case.

On top of that, you have to consider the fact that the offset is not strictly equal to 0 (zero-g level, zero-rate level and hard iron parameters, that have to be taken into account but that can be removed in post processing or setting the appropriate offset registers).

I agree with you that this product has little specifications on noise performances (spectral density), but it is not one of the newest and most performing ST motion MEMS in terms of noise... Btw, are these values enough for your application?

Regards