LSM9DS0 noise issues

[ralfh]

I want to use the LSM9DS0 IMU to measure magnetic field changes and relative motion. The LSM9DS0 specs looked very promising (acceleration sensitivity 0.061 mg/LSB, magnetic sensitivity 0.08 mgauss/LSB), but when I started reading actual data from the sensor I saw that there is a lot of noise (standard deviation of measured acceleration and magnetic values typically 20-40 times the sensitivity). This is using the Adafruit sensor and LSM9DS0 libraries to read the sensor data and configure the sensor to the narrowest measurement range (best sensitivity).

Based on the observed issues, my questions are:

1. Is it plausible that the Adafruit sensor and LSM9DS0 libraries fail to correctly set the sensor measurement ranges or is there another explanation for the high noise levels?

2. How do Adafruit sensor and LSM9DS0 libraries deal with the additional sensor configuration parameters (as mentioned in the LSM9DS0 specs), in particular the high-pass filter and the continuous/single conversion mode?

Thanks for any help/clarification,

Ralf.

[adafruit_support_mike]

1. Is it plausible that the Adafruit sensor and LSM9DS0 libraries fail to correctly set the sensor measurement ranges or is there another explanation for the high noise levels?

It's possible, but would be hard to prove one way or the other.

I've been through the LSM9DS0 datasheet three times, plus searching the app notes and related documents at the ST website, and can't find any mention of typical accuracy or noise levels. Even the basic characteristics have a footnote that says, "typical values not guaranteed" which has a suspiciously absent comma in the middle.

Overall, I get the impression of a datasheet where the marketing department didn't let the science and engineering teams say very much. As a matter of pure speculation, I wouldn't be surprised to see something like that for a device based on new technology that's reached the point of being useful but still has plenty of room to mature.

2. How do Adafruit sensor and LSM9DS0 libraries deal with the additional sensor configuration parameters (as mentioned in the LSM9DS0 specs), in particular the high-pass filter and the continuous/single conversion mode?

Broadly speaking, they don't.

There are enumerated constants for all the registers, plus read/write routines for basic communication with the chip, but they don't contain code for the full gamut of chip features. In this case, given the frequency, bandwidth, filtering, FIFO, and interrupt options, a complete library could end up being too big to fit in an Arduino's Flash memory.

[adafruit_support_bill]

MEMs accelerometers are very sensitive devices and are often used to measure vibration. Vibration is a form of oscillating accelerations and is often mistaken for 'noise' when attempting to measure linear acceleration.

[ralfh]

Thanks Bill & Mike. Yes, it looks as if the performance of the sensor has been exaggerated quite a bit. I'll try to use filters to get something useful out of this.