MSP430FR2433: MSP430FR2433IRGET temperature sensor accuracy

Part Number: MSP430FR2433

Hi to All,

What is the absolute error and accuracy of the Temperature reading when using the internal temperature sensor (in a MSP430FR2433)?  

It based on the ADC specific channel included in the MSP430FR2433,

If it is acquired through a sequence of readings exploiting the multiply channel sample method, that automatically switches the analogue mux, then can be affecting this the temperature readings by the signal present on the other analogue channels?

We didn't find out any accuracy information on the data sheet for this temperature sensor, except the registers to read  and the formula to apply to convert the ADC reading into a °C or °F.

We missed something published by TI?

Based on our first evaluation (with about 10 samples) we found errors between -3°C to +3°C from the ambient temperature (22°C), with an overall chip consumption below 2mW in average (measured by the Energy Trace Technology).

Is it conforming the Chip specification?

Thanks

Best Regards

  • Hello Aslessandro,

    There is no spec for the temp sensor, therefore it is not test in production.  To achieve the best accuracy, perform the 2-point calibration mentioned in the datasheet.  Did you do the 2-point calibration?

  • thanks Dennis,

    no! we didn't performed any calibration task but simply we used the registers found with the default voltage on ADC when at 30C and at 85C, they are:

    CALADC_15V_30C and CALADC_15V_85C (because we use the ADC reference at 1.5V)

    writing them it seems necessary a procedure to unlock them first. 

    New Question: if we modify them for calibrating the ADC temp sensor , can them be back to the default values at certain conditions?

    for example at reset or when the battery is lost or removed?

    the calibration is effective also against the VCC supply voltage? 

    thanks 

    Best Regards

    Alessandro

  • Hi Alessandro,

    Since you can't write to the CALADC_15V_30C and CALADC_15V_85C ROM memory, you can set aside some memory locations in INFO memory for the calibration points you derive from the 2-point calibration.  Since INFO memory is FRAM, the values you write will not change through a reset or power cycle.  If you want to retain these values during a programming cycle, you can setup the programmer to not erase INFO memory, only MAIN memory.  

    Also, the temperature sensor should be independent of the VCC because it is bias by an internal reference

  • Thanks Dennis,

    we solved this following your answer partially, by calibrating the offset error only, so saving the offset into a FRAM location, because the two point calibrations is really dangerous for many other reasons: when performing them into a oven with a batch of PCB boards, to be calibrated each. We need the accuracy of the board temperature in a short range between 20 to 45°C, so a single point calibration in the middle region is enough for our applications and more safe. 

    Thanks and Best Regards

  • Hi Alessandro,

    Ok, I understand the challenge of calibrating at the hotter termperatures.  So it sounds like you are good with a single pt calibration.

**Attention** This is a public forum