This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS122C04: Internal Vref - Long term drift

Part Number: ADS122C04

Dear TI-Support-Team,

my question is about the 110ppm long term drift of the internal voltage reference.

Is that drift zeroed when device is turned off?
Is that drift zeroed when internal Vref is deselecetd?

I would really appreciate some more information regarding that parameter. 

Thank you very much, best regards!

  • Hi Florian,

    Welcome to the E2E forum!  The internal reference is factory trimmed in final test to be within the accuracy specifications shown in the electrical characteristics table.  However there are factors, such as package stress when the device is soldered, that may affect the reference voltage over time.  Also, as the device self-heats from current running through the device for significant periods of time there can be some drift as the device settles into a more relaxed state.  Generally we see the bulk of the drift within the first 100 hours or so of continuous operation and then the reference settles into a pretty flat response.  The 1000 hour test for the long-term reference drift pertains to 1000 hours of continuous operation at a fixed temperature with the typical drift appearing as 110ppm.  This specification is not to be confused with the temperature drift specification that precedes the long-term drift in the table.  You can think of the long-term drift as the characteristic shift of the reference over time.

    There is no 'zeroing' of the long term drift as it is a characteristic shift.  So once it is stabilized it remains at that voltage.  The more dominating drift is the temperature drift and this will change relative to temperature.  The overall drift specification only pertains to using the internal reference.  When using an external reference, the drift specification would be taken from the datasheet from the selected reference.

    Best regards,

    Bob B

  • Hello Bob,
    thank you for the quick response. I understand.

    So, if we operate the devices for lets say 100 hours after production and do a two point calibration afterwards (f.e. in a thermocouple setup), the characteristic shift is eliminated by the large? 

    Regarding the temperature drift of the internal Ref:

    It's mentioned with typical 5 to 30 ppm max. - is there a device specific variation or dependent of the current operation temperature (-10°C vs 75°C for example)?
    The difference between the typical and maximum value is huge, so (how) can we estimate the relevant factor in our application?

    Thanks for your answer in advance, best regards,


  • Hi Florian,

    Take a look at Figure 25 in the ADS122C04 datasheet which shows the trend of the internal reference over temperature and at two different analog supply voltages.  The large spread from the electrical characteristics table shows the full min/max values for the entire operating range of the ADC.  You can see how the trend would lower based on a more limited temperature range.  The values in the graph would follow the mean of the characterization data.  The standard deviation across the analog supply voltage range is about 3.1ppm.

    If you need a greater performing reference you can always consider using an external reference.  Generally what we have seen is the reference is not the critical issue.  The greater error tends to come from external noise being picked up by the thermocouple wiring and measurement error in the cold-junction compensation.  There can also be computational error as well depending on how the conversion from voltage to temperature is calculated.

    RTD measurements tend to benefit more from calibration due to sensor construction.  Thermocouple measurements also benefit, but here again the calibration is also impacted by the cold-junction measurement.  Achieving a highly accurate system is not a trivial task.  In the end, if your cold-junction measurement is only accurate to 0.5 degree C, the small reference error will be negligible in comparison for most thermocouples. 

    The ADC with respect to the 2.048V reference has an output a code (LSB) relative the reference voltage where a single code using a gain of 1 is approximately 244.1nV.  If the reference voltage changes due to drift to 2.048300V (300ppm), the outcome will still be around 244.1nV for each code.  The level of precision regardless of calibration comes down to resolving a value in the 10's of pV.

    Best regards,

    Bob B