Skip to main content
Nta
Associate
August 31, 2020
Question

Accelerometer z-axis on LSM6DSM jumps too much!

  • August 31, 2020
  • 1 reply
  • 1177 views

I am using an LSM6DSM but I got weird data on the z-axis for the accelerometer, as you can see in the attached picture z-axis jump too much in comparison with other axises, the device was steady in the whole sampling time.

So does anybody have any idea what could be the issue?

Thank you for your time in advance.

    This topic has been closed for replies.

    1 reply

    Ozone
    Principal
    August 31, 2020

    The x-axis signal seems to have much more low-frequency components which apparently cause the larger deviations.

    I suppose they are actual vibrations in your environment.

    You could do the following:

    • Try to turn it by 90°, and see if the signal characteristics follow to the y- and z-axis;
    • place it on some dampening material (rubber mat);
    • compare with a refererence accelerometer (known output);
    Nta
    NtaAuthor
    Associate
    August 31, 2020

    Thank you for your fast answer,

    I already tried to turn it by 90° but it always influences the z-axis not matter in which direction, please check the attached picture.0693W000002ll2GQAQ.png

    I also have a STREF-MKI28V6 but the output data on the reference board is good enough in the same condition.

    I soldered the IMU by hand, can it be the issue?! I also changed the LSM6DSM chip on my board with a new one but it has the same behavior.

    Eleon BORLINI
    ST Employee
    September 14, 2020

    Hi @Nta​ ,

    if LSM6DSM sensor on the reference board is ok, cannot exclude it is a soldering-related issue for the other two samples, affecting the Z component of the MEMS structure... I suggest you to check the Surface mounting guidelines for MEMS sensors in an LGA package for more information.

    At application level, if you don't need the DC and the low frequency components of the acceleration, you could activate the accelerometer high-pass filter acting on HPCF_XL[1:0] bits of CTRL8_XL (17h) register (p.75 of the datasheet).

    Pls let us know if you'll do some progress.

    -Eleon