LSM6DS3TR-C behaviour

[KevinSR]

Hello...

I'm currently testing the LSM6DS3TR-C sensor output before trying to proceed to calibration.

I already run the internal self-test on the accelerometer and gyroscope, both results were between the acceptable range.

After that I did my own tests on both sensors (change over temperature, change over orientation, noise calculation, time for stability, etc), I didn't get any problems while testing the acelerometer, but in the case of the gyroscope I didn't get "reliable" data.

First of all, the time of stability (the real output when the sensor is still after turning the sensor ON) is pretty slow compare to the accelerometer, only after 1 minute and 25 seconds (for X and Y axis) and 6 minutes and 10 seconds (for Z axis) I can consider the output to be stable.

The thing that is really troublesome for calibration purposes is the repetability of the gyroscope, after the output of the gyroscope is "stable", it is not the same after turning the sensor ON again. For example, if the "stable" output of the X axis of the gyroscope is -200, after turning the sensor OFF and ON again, the "stable" output is now 10.

So... this are my questions:

1. Is the time of stability of the gyroscope normal? (there's no reference in any datasheet I've found)

2. Is the gyroscope Zero Offset error suppose to change each time the sensor is turn ON? 

For reference this are the order/value of the registers I wrote to:

CTRL1_XL = 0x88

CTRL6_C = 0x03

CTRL4_C = 0x06

CTRL2_G = 0x70

CTRL5_C = 0x00

CTRL8_XL = 0x00

[Federica Bossi]

Hi @KevinSR ,

Welcome to ST Community!

Can you log also the internal temperature sensor (reg 20-21h)? And share the plot with the gyroscope's ones?

Thanks.

[KevinSR]

I had to acquire two new sets of data.

FIRST SET: (Long time after turn ON)

These are the raw values of each gyroscope axis and its relation with the temperature after the time of stability:

SECOND SET: (Immediately after turn ON)

These are the raw values of each gyroscope axis and its relation with the temperature before the time of stability:

As it can be seen, before the time of stability the relationship of the gyroscope data and the temperature isn't lineal. Another thing to point out is that the total change of the gyroscope raw data of the second data set is bigger than the one of the first data set, even though the range of temperature is the same in both cases.

This is the temperature plot over time related to the last data set:

Finally, the gyroscope data over time of the last set:

:

 

[Federica Bossi]

Hi @KevinSR ,

I don't see any anomalous behavior in these graphs.

If the temperature is not stable, the outputs will not be stable. Please, be always sure that the temperature is stable before starting a new acquisition.

[KevinSR]

Unfortunately the stability time can't be totally related to temperature changes. In that case a temperature compensation strategy can assure a stable output from the beginning. That's not the case for (at least) the first 2 minutes after turning ON the sensor (taking a look at the graphs: the x axis output should have a step down direction as temperature rises, that's not the behaviour at start up).

Anyway, for the calibration procedure there's no problem in waiting some minutes for the sensor to get stable results. What is actually troublesome is the non repeteability of the sensor... after returning from a change of aprox. 3 [°C] (trusting the temperature sensor) there are changes of more than 1.5 [degrees/s] in the output of the gyroscope.

Isn't the sensor output suppose to have a fairly lineal relationship as temperature rises, or at least, a repetable relationship?

(take a look at the graphs, sometimes the relationship is exponential, other times lineal, other other times the tendency change direction from positive to negative. This makes impossible a temperature compensation strategy )

[KevinSR]

This might be a more representative set since is manly focus on temperature changes and not on start up behaviour:

temperature over time

gyroscope value over time:

Six points are specified on all graphs. This points corresponds to changes in heating/cooling.

Hopefully it's evident that increasing/decreasing the temperature multiple times doesn't provide the same effect... rather, it seems pretty random how a big decrease in temperature provides smaller changes than a lower one.

Another rare thing is the step like behaviour right at the start of heating or cooling.