ADS124S08EVM: ADS124S08EVM - Internal temperature sensor

Hi Matteo,

Matteo said:

Regarding VREF: so don't I get a more accurate measurement using the internal 2.5 V reference voltage instead using AVDD?

Probably, as the reference is most likely more stable that AVDD. The measurement depends on a number of factors including how quickly you want to make measurements and the analog settling delay that may be required.  I don't know how you plan to use the ADC, so you have to determine the accuracy and timing requirement.

Matteo said:

Is there any advantage using the ADC code instead of voltage for temperature calculation?

As I said originally, there are many ways to compute the temperature.  Minimizing FP calculations is preferred in most systems depending on the math engines that may or may not be available on the micro you are using.

Matteo said:

What is the reason that I measure a temperature about 4 °C lower than with a K-type TC placed on the PCB and also than the ambient temperature?

The initial accuracy is not all that great.  As I said originally, the values given in the datasheet are typical and to get the best accuracy you would need to calibrate.  Also keep in mind that the ADC is measuring the die temperature and not necessarily the PCB or ambient temperature depending on airflow around the ADC.  It will also differ as to which device package is used.  The QFN will be closer to the PCB temperature due to the center pad directly attached to both the die and PCB.  The TQFP version is only thermally connected to the PCB by the pins and the package temperature can be somewhat different.

Best regards,

Bob B

Hi Matteo,

For calibration you could do it many ways.  The simplest would be to find the initial accuracy with a single-point calibration at a known room temperature to set the initial output voltage used in the computation.  Let's say the room temperature is set to 25 deg C and the internal temperature is showing 129.2 mV instead of the typical 129mV as given in the datasheet.  This will give you a better starting point, and you can still use the datasheet coefficient of 403uV / degC.  Using a two-point calibration allows you to more accurately calculate the coefficient itself.

Best regards,

Bob B

Hi Matteo,

How quickly the device will warm up will be based on what is powered on and how much current is running through the device.  Current will increase if the device is actually converting (as opposed to standby), the internal reference is turned on, if the PGA is enabled, and if the IDACs are turned on.

Depending on which package is used it is possible that the die is slightly cooler (PCB temperature as opposed to the air temperature) than the actual ambient.  However it shouldn't take long for the die temperature to stabilize once the desired configuration is set and the device is in conversion mode.

Also, the typical mV specification at 25 deg C can vary in +/-  from the mean.  So it is possible that the initial temperature can appear less than the ambient.

It is not clear to me the use case for the internal temperature sensor.  If you need high initial accuracy TI does make a full line of temperature sensors.

Best regards,

Bob B

Hi Matteo,

You don't need an accurate temperature measurement for this case.  Here you can just use a relative difference.  As far as drift is concerned, offset drift can be virtually eliminated by using global chop.  Gain drift will be combinations of drift of the reference as well as the ADC gain drift.

One thing I will add here is long term drift can be somewhat difficult to measure with the EVM.  The reason is that sometimes the drift is affected by the air currents around the EVM.  We have seen a lot of instability due to lab environments (heating and cooling cycles) as well as air currents from other nearby equipment (power supply fans for example).  Due to these effects, it is best to isolate the EVM from the rest of the environment by placing the EVM in a box or we have even used large ESD bags to limit air flow around the EVM.

One last thing and that is the EVM was not designed to be placed in an oven or temperature chamber.  Although the ratings for the ADS124S08 is industrial temperature range (-50 to 125 deg C), many of the components on the board are in the commercial temperature range (85 deg C or less).  Therefore I would recommend that the maximum temperature applied to the board be 65 deg C or less.

Best regards,

Bob B