Is there a typo in sensitivity units (datasheet of LIS2DH12, page 10: Mechanical characteristics)?

[ARema.1]

Hello,

I am still new to accelerometers, and I am trying to understand the logique of how to decode the raw data using the sensitivity.

In the datasheet of the LIS2DH12 accelerometer, in page 10: Mechanical characteristics, the unit of the sensitivity is (mg/digit), isn't it supposed to be the inverse ? (digit/mg)

Because when I used another accelerometer (LIS3DSH), the granularity (or the scale) of the sensitivity is less than 1 with a unit of (mg/digit).

For example, if I take the FS bit set to 10, High-resolution mode case (Measurement range = 😎 from the LIS2DH12, which has a sensitivity of 4, if I take the inverse of it ( 1/4 ), I get 0.25, which is quite close to LIS3DSH's sensitivity in the case of FS bit set to 011 (Measurement range = 😎 that is equal to 0.24.

So I am a little bit confused, if someone could clear this for me.

Thank you

[Eleon BORLINI]

Hi @ARema.1​ ,

both LIS3DSH and LIS2DH12 sensitivity are expressed in milli-g / digit, so at least the two datasheet are coherent 😉

The fact is that, taking for example the FS equal to 2g and the high-resolution mode for both accelerometer, the sensitivity is (almost) equal to FS / 2^(nbits-1) in both cases, but the nominal nbits is different (-1 since you have to consider the sign bit).

For the LIS2DH12, it is 12, so the resolution is 2 g / 2^11 bits = 0.96mg/digit (almost 1 mg/digit as shown in the datasheet), while for the LIS3DSH the value is 16, i.e. 2/2^15 = 0.06 mg/digit, coherently with the other datasheet.

That is, the dataout is defined on the 12 MSbits or 16 MSbits ("left shifted").

You can also refer to the C drivers on Github for both sensors:

• LIS2DH12 case --> lis2dh12_reg.c

float_t lis2dh12_from_fs2_hr_to_mg(int16_t lsb)
{
  return ( (float_t)lsb / 16.0f ) * 1.0f;
}

• LIS3DSH case --> lis3dsh_reg.c

float_t lis3dsh_from_fs2_to_mg(int16_t lsb)
{
  return ((float_t)lsb) * 0.06f;
}

The calculations include the int16 casting for the two's complement conversion and the eventual bit shifts in terms of divisions.

Let me know if this can dissipate your doubts.

-Eleon