Is the LSM9DS1 accelerometer sensitivity right for the 16G?

[MSanc.1]

I am using the IMU LSM9DS1. However, when developing the library for communicating with the IMU and reading the accelerometer values, I noticed something odd.

According to the datasheet, the sensitivity of the accelerometer is given by:

 

which normally can be given by the formula: scale / 32768, where scale = 2,4,8,16;

This formula works for all scales except for 16g. In this case, the formula gives 0.488 mg/LSB

So, I wanted to know if the formula that I am using is wrong or if the datasheet is wrong, because according to the formula, the 0.732mg/LSB sensitivity corresponds to 24g. Also when using the sensitivity value of 0.732mg/LSB I can obtain readings of 20g

Thanks in advance,

Best regards

[Eleon BORLINI · Accepted solution]

Hi @MSanc.1​,

It could seem strange, but the 0.732mg/LSB is the correct sensitivity. Basically, the FS is actually bigger than 16g, but near the edge the linearity is no more so good.

You can check the consistency of the 16g-sensitivity by placing the device configured for +-16g FS in steady state with Z axis pointing upwards. You should get something like 1360 LSB; meaning that, being the sensitivity is 0.732mg/LSB, you'll get 1360*0.732 = 995.52mg = 0.996g, practically 1g. i.e. the expected value.

-Eleon 

[MSanc.1]

Hi Eleon,

Thank you for the reply.

I verified that, but then why can the sensor measure up to 20g?

When setting this sensitivity value of 0.732mg/LSB, the accelerometer maxes out at 20 g. Should this happen?

When you say that "near the edge the linearity is no more so good", this means that even if I measure accelerations higher than 16g I should not trust them. Do you confirm this?

-Miguel

[MSanc.1]

Hi Eleon,

Thank you for the reply.

I verified that, but then why can the sensor measure up to 20g?

When setting this sensitivity value of 0.732mg/LSB, the accelerometer maxes out at 20 g. Should this happen?

When you say that "near the edge the linearity is no more so good", this means that even if I measure accelerations higher than 16g I should not trust them. Do you confirm this?

-Miguel

[Eleon BORLINI]

Hi Miguel @MSanc.1​ ,

>> When you say that "near the edge the linearity is no more so good", this means that even if I measure accelerations higher than 16g I should not trust them. Do you confirm this?

Basically yes, above the 16g the value is not guaranteed. If you test, for example, a couple of samples, you may get slightly different values.

Could it be a limitation for your application?

-Eleon

[MSanc.1]

Basically yes, above the 16g the value is not guaranteed. If you test, for example, a couple of samples, you may get slightly different values.

Could it be a limitation for your application?

No it is not for now.

But then I wonder, doesn't this linearity problem applies for the other ranges? So, for example, to 8g is the linearity good near the edge?

-Miguel

[Eleon BORLINI]

Hi Miguel,

No, it is a characteristics only of the 16g full scale. No issue for the +-2g, +-4g and +-8g FSs.

Basically, (all) the sensor(s) is (are) composed by a mechanical part, the MEMS, deputy to transduce a parameter of the world into capacitance variation, and an electronic part, the ASIC, deputy to the (digital) signal processing phase.

The LSM9DS1 is quite an old part number ("old" in the sense of MEMS Sensors product lifetime, i.e. it has been developed some years ago), although it is provided by some unique characteristics (it is a 9-axis device) and it faces some limitations on the accelerometer part of the mechanical sensor for high g (16g FS): when approximates to high g, the MEMS response could deviate a little from a linear response.

It is not of course an irreversible issue, since the devices endure 10000g shock tests during the qualification process, but ok, the 16g edge is a little trick from an accuracy point of view.

In any case, if you would need very high, you can switch to sensors such as the H3LIS331DL device that goes up to 400g.

Let me please know if this explanation clarifies your doubts.

-Eleon

[MSanc.1]

Thanks for your explanation Eleon.

I think that is all

-Miguel