LSM6DSL temperature drift?

[THofr]

We have a problem reading acceleration data from the LSM6DSL sensor.

The sensor is placed next to a linear controller, which heats the sensor a little. After switching on, the acceleration value drifts away according to the temperature (see Screenshot).

 

The sensor is configured as follows:

CTRL2_G=0x62 //ODR_G = 416Hz, FS_G = ±125dps

CTRL1_XL=0x20 //ODR_XL = 26Hz, FS_XL = ±2g

CTRL3_C=0x40 //BDE enabled

The sensor is polled every 10ms and the bit XLDA is evaluated.

What could be the problem?

[Eleon BORLINI]

Hi @THofr​ ,

The drift excursion is indeed very high, considering that the temperature drift for the LSM6DSL should be in the order of ±0.01 %/°C (datasheet p. 21)...

Are you facing the same issue on all the 3 accelerometer axis (or just on Z), and also on the gyroscope?

This to understand whether the issue is related to temperature drift, time drift or effective noise coming from the environment.

Consider in general that, to test the , you could run the self test procedure, as described in the AN5040 application note, p. 96. A sample code in C for this test is available on Github --> lsm6dsl_self_test.c.

-Eleon

[THofr]

Hi Eleon,

thank you for the answer.

I have recorded the course of the axes:

I will implement the test procedure next week.

Could the problem perhaps provoked by the soldering process?

Greetings

Torsten

[Eleon BORLINI]

Hi Torsten @THofr​ ,

thank you for sharing your data.

I see that the Z-axis (i.e. the out-of-plane axis) and the pitch are the ones affected.

It could be that the soldering process induces mechanical stresses on that axis, resulting in a risk of damage.

Can you please share the soldering profile for a check?

-Eleon

[THofr · Accepted solution]

Hi Eleon,

we have only assembled two prototypes with an undefined soldering profile,

where some of the soldering was done by hand.

Therefore, we now assume that the problem was caused by soldering.

We will soon assemble more prototypes with a defined soldering profile.

I will then look at the measured values on these.

If we continue to have problems with this, I will get back to you.

Thanks for the help and have a nice weekend.

Torsten

[Eleon BORLINI]

Have a nice weekend you too, Torsten

-Eleon