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.

REF2025: Definition and conditions for thermal hysteresis and trial-to-trial offsets

Part Number: REF2025

Hello,

We are using the REF2025 in a test fixture that gets temperature cycled as part of the test.  The test fixture voltage reference is measured with a high precision DMM, then the fixture is removed and temperature cycled some number of times, and the voltage reference is re-measured.  This process may repeat many times.  Our "thermal cycle" is defined as only 25C -> 50C -> 25C.

Measuring the Vref of a given fixture after 50, 60, and 70 cycles has shown a voltage offset of up to 200uV.  This is right around the long-term stability and the thermal hysteresis numbers, but we have not run it for anywhere near 1000 hours, and our temp cycles are much more narrow than the datasheet reference.  We are trying to determine if we are observing an expected behavior of the part ONLY, or an unexpected behavior, which then might be due to some other error in our fixture or measurements.

Looking at the datasheet for the REF2025, I am not certain what the defined behavior of a thermal cycle should be.  Specifically:

One thermal cycle is defined as 25C -> -40C -> 125C -> 25C.  The results of a single cycle are shown to be about 60ppm.  In the characteristics section of the datasheet, "Cycle 2" is listed as having an offset of 35ppm.

Question 1: is that second thermal hysteresis defined with reference to the nominal 2.5V, or with reference to the 1st cycle voltage (i.e., 2.5V+35ppm or 2.50015V+35ppm)

Question 2: does thermal hysteresis eventually settle?  So, after, say, 100 temp cycles, would the result of each cycle be independent of the last (i.e., still 2.5V +/- 60ppm regardless of last point), or would it settle around a particular offset (i.e., cycle 100 is +/- 1 ppm from cycle 99)?

Thank you very much for your time and consideration.

  • Hi Kevin,

    200uV is about 80ppm of shift at 2.5V. If you look at Figure 43 below, from the REF20xx datasheet, the thermal hysteresis can be as high as 100ppm (not guaranteed). The 60ppm and 35ppm values are typical, not max, values. 

    The hysteresis during the first thermal cycle is usually higher then the following cycles, hence the 60ppm and 35ppm values. The 35ppm variation is calculated from the original starting value at 25C (2.5V +/- 0.05%) and can be applied to the 100th cycle as well. In other words, each cycle is independent of the last, and can vary higher then the previous cycle. 

    Thermal hysteresis is characterized from -40C to 125C to cover the extreme cases, however hysteresis variation is not necessarily linear. This means that even though the range might be smaller, the variation might not be. 

    Regards, Diego Lewis

  • Hello,
    Thanks for the info. To add to this question:

    What constitutes a "thermal cycle?" If the hysteresis is non-linear over the temperature range -40 to 125C, then would it be within expectations for a room temperature fluctuation (i.e., increase 1 degree C, then decrease back to start) to induce an offset? If so, how big an offset?

    Is the answer a stacking of the specified maximum temperature drift?
    That is, the datasheet lists 8ppm/degC max. So, from 25C to 26C would be +/-8ppm, then back would be another 8ppm/degC max? What is the right analysis here?
  • Kevin,

    The hysteresis can indeed be stacked with the temperature coefficient; therefore Vout could shift 8ppm due to the temperature coefficient when the temperature is raised 1C and not recover the full 8ppm on the way back due to hysteresis. 

    As mentioned before though, each hysteresis is related to the initial condition. Thermal hysteresis has a much smaller effect then the temperature coefficient though, since a over the 165C range, the temperature coefficient can lead to a variation of up to 1320ppm. 

    Regards, Diego