Effect of the temperature on BMI270 readings

[Myzhar]

Hi,
testing the effect of the temperature on BMI270 readings I noticed a strange behavior. I expected to see a drift on readings according to increasing temperature, instead I see many different steps, like if there is something acting internally to compensate the changing temperature.
I searched on the datasheet, but I have not found any explanation.

You can appreciate what I'm saying in the attached 30' plot of the Gyro value for the Z axis where also the temperature value is reported.

Any idea about what can be the cause of those steps?

PS the sensor was steady and I verified that nothing was acting on it while testing.

[Myzhar]

Hi,
I performed the same test with the BMI160 sensor and it is not affected by the same problem, the Gyro data drift according to temperature changing, but continuously, without any kind of steps.

The BMI270 datasheet says that the Temperature Sensor is automatically enabled when the Gyro is enabled... this seems to confirm my hypothesis about an automatic temperature compensation enabled by default.

This behavior is impacting the calibration of the sensor because there is not a continuous relation between temperature and sensor data drift

[o_o]

Hi Myzhar,

This is expected temperature compensation behavior and looks within specs.

The effect is greatly amplified by the scale of your graph, as you can see the variations are below the noise threshold.

Indeed if your application is sensitive to these effects the BMI160 would be better.

That being said, if you require a high temperature stability, we strongly recommend one of our IMUs with a closed loop gyroscope, such as BMI085.

o_o

[Myzhar]

Thank you.

I needed indeed this kind of confirmation to be sure that I'm not using a wrong configuration.
The automatic enabling of the temperature sensor when gyro is enabled was a good clue about the casues of this effect.

[Myzhar]

Hi,
I made a comparison test in similar conditions between BMI160 and BMI270 and I attached the plot of one of the axes of the Gyro (X).
The behavior of BMI160 is predictable and the effect on temperature on readings can be compensated by software.
The behavior of BMI270 is instead cahotic and there is no way to apply an algorithm to compensate the temperature effect... the temperature compensation internal algorithm does not effectively compensate the effect of the temperature as you can see be the attached plots, but introduces instead discontinuity on the data.

Can you confirm me that this is the real expected behavior of BMI270 and that there is not a bug in the temperature compensation internal algorithm?
Isn't there a method to disable this feature in BMI270?

Thank you very much for your help
Walter