LSM9DS1 low pass filters are not workring properly

[HSafy.1]

Hello Everyone,

for the LSM9DS1 )DOF board. I am trying to use the associated low pass filters for the gyroscope and accelerometer and maybe I am doing a wrong confiugration and I want to be sure.

It is not clear for me how to do that specifically for the accelerometer. In page 45 of the Manual ( 7.12 CTRL_REG1_G) it is stated that the frequencies and cutoff frequencies are for LPF1 and LPF2 when both gyroscope and accelerometer are activated. However, the block diagram is only for the gyroscope or should I assume the same block diagram for the acclerometer ? ( Figure 28)

Again , is this only specific for the gyroscope or is it sufficient for both gyroscope and accelerometer ?

My main problem is that I assumed it is not enough and I also edited the accelerometer control registers and then I started getting faulty raw data. To be more specific , I edited CTRL_REG7_XL ( page 52) to operate in HR mode and the FDS to be 1 and this is when I get wrong values for the accelerometer

Also If I want to use LPF1 and LPF2 and ignore HPF , Am I supposed to edit CTRL_REG2_G (page 47) OUT_SEL[1-0] = 10 ?

Thank you,

Hesham

[HSafy.1]

Any answer or help ?

[Eleon BORLINI]

Hi @HSafy.1​ , I believe you are well interpreting the datasheet. The filters configured in the CTRL_REG1_G register refer to the gyro only but are effective when both gyro and axl are enabled. On the other hand, the CTRL_REG7_XL regulates the axl LP filter (try also to not enable the FDS bit). Btw, please consider that the ODR acts as natural LP filter according th the Nyquist theorem.

I suggest you to check the standard C drivers on Github repository for LSM9DS1, especially the lsm9ds1_read_data_polling.c file. In particular, there is a section related to the filter configurations for both accelerometer and gyroscope.

 /* Configure filtering chain - See datasheet for filtering chain details */
  /* Accelerometer filtering chain */
  lsm9ds1_xl_filter_aalias_bandwidth_set(&dev_ctx_imu, LSM9DS1_AUTO);
  lsm9ds1_xl_filter_lp_bandwidth_set(&dev_ctx_imu, LSM9DS1_LP_ODR_DIV_50);
  lsm9ds1_xl_filter_out_path_set(&dev_ctx_imu, LSM9DS1_LP_OUT);
  /* Gyroscope filtering chain */
  lsm9ds1_gy_filter_lp_bandwidth_set(&dev_ctx_imu, LSM9DS1_LP_ULTRA_LIGHT);
  lsm9ds1_gy_filter_hp_bandwidth_set(&dev_ctx_imu, LSM9DS1_HP_MEDIUM);
  lsm9ds1_gy_filter_out_path_set(&dev_ctx_imu, LSM9DS1_LPF1_HPF_LPF2_OUT);

Regards

[HSafy.1]

Hello @Eleon BORLINI​  Thanks for your answer !

Can you please be more illustrative with this sentence " please consider that the ODR acts as natural LP filter according to the Nyquist theorem" ? and if it acts as a natural LP filter, so for example , what would be the associated cutoff frequency for a sampling frequency of 952 Hz ? I am just trying to understand 🙂

Also why would you recommend me not to change the FDS bit ? what could happen if I activate the accelerometer filters ?

I will check the repositories as well. Thanks !

Best Regards,

Hesham

[Eleon BORLINI]

Hi Hesham, I mean that if you acquire data with an ODR 952Hz, you will be able to reconstruct a (sine) signal of maximum ODR/2, i.e. 476Hz (except from an aliasing effect). About the FDS bit, you should enable it to set the accelerometer LP filter, but I understood you are getting wrong values for the accelerometer, so you could try to disable it and check if the data are well. Regards