IIS3DWB temperature sensor offset

[pb38]

Hello
In the IIS3DWB datasheet, it is said that the embeddeed temperature sensor offset can vary within a [-15°C : +15°C] range.
Does that mean that the reading from OUT_TEMP_H/L can vary within [-256*15 LSB : +256*15 LSB] at 25°C ?
Having tested 2 chips myself, I got a much better result.
I am then wondering whether I was just lucky, and if I really need to calibrate.
Thanks

[Issamos]

Hello @pb38 

I suggest you to perform a one-point calibration to increase the accuracy of your sensor as mentioned in the datasheet. You can take a look at this exemple:

After that, you can see that the accuracy of your system increase.

Best regards.

II

[TDK]

Mind the footnote which explains what min/max means here. 3σ is a wide bin, and most of the time you're going to be in the middle third (so ± 5° C). And 0.3% of samples will be outside of that bin, so no guarantee you'll even be within that range.

Also note that even after calibration, you can have ± 4 °C over the full range. The focus of this chip is accel/gyro, not temperature.

[pb38]

Thanks for you answer,

I did not want to calibrate, but il looks like I will have to do it anyway. At least I will not have to calibrate the sensor sensitivity.

[pb38]

Thanks for your answer.

The 2 chips I have tried are within 0.2°C (probably from a same lot), but I looked closer and their offsets are approximately -2.5°C, and I cannot afford that much.
I think I was in denial, hopping for an error in the datasheet, as ±15°C looked very big to me. It could have been ± 15LSB instead...

I guess I will have to calibrate then, it is a shame ST does not do it for us.

[Issamos]

Is always better to do the calibration by yourself. It can depend on your environmental conditions, your use case,... . So, is not going to be really efficient if it is done in the same conditions for everyone.

If your question is answered. Close the topic by choosing a best answer.

Best regards.

II

[Federica Bossi]

Hi @pb38 ,

If you want an accelerometer more accurate in temperature, I suggest you to use the LIS2DTW12 sensor.

Otherwise, you need to implement an OPC at room temperature (25 °C).

If this helps you, please mark my answer as "Best Answer" by clicking on the "Accept as Solution" button, this can be helpful for Community users to find this solution faster.

[TDK]

If ±2.5 °C is too much, you probably need a dedicated temperature sensor. Note that this will give you temperature at the chip, which is going to be higher than ambient because the chip is heating up.

Standalone temperature sensors will not heat up as much and will be much more accurate.

[pb38]

Thank you, your answer helps.
LIS2DTW12 is not suitable for our application, we need the DC to 6kHz bandwith. Any idea if there is something else in the pipe in this range ?

[Federica Bossi]

Hi @pb38 ,

Our devices with 6kHz bandwidth have a temperature offset min-max of ±15DegC, however, as suggested by @TDK , if ±2.5DegC is too much, you should choose a dedicated temperature sensor.