Tool/software:
Hello!
I have a question regarding the annual temperature drift of the TMP117 sensor. I have read most of the similar posts on here, but want to ask about the interpretation of the drift results between Datasheet Rev. B and Rev. C:
Rev. B : 300 hours @ 150°C --> ±0.03°C
Rev. C : 1000 hours @ 150°C --> ±0.03°C
I was given the task to estimate the annual drift of the TMP117 sensor, based on the information supplied by TI. I interpret these results in such a way that the sensors exposed to 300 hour @ 150°C have experienced a successful "burn-in", where the drift-causing artifacts in the chip (or at least the most significant ones) have been exposed, causing the deviation. Exposing the sensors to additional 700 hours @ 150°C seems to have no effect at all, indicating that the drift has most likely reached a plateau at ±0.03°C. However, I might be wrong, because my conclusion does not take into account any slight differences between the 300 and 1000 hour values, such as hidden digits due to rounding (is the 300 hour number in Rev. B actually ±0.025 and the 1000 hour number in Rev. C actually 0.034°C?).
Is my interpretation of the drift data correct? Can I (given that the typical error graph of a "burn-in" process tends to plateau at some point) assume that the long term stability (annual drift) of the TMP117 is approximately ±0.03°C? If not, how then should the existence of these two highly related numbers be interpreted?
Any help with this is greatly appreciated!
PS. I am well aware of the Arrhenius equation. Plan B would here be to linearly extrapolate and (very simply) estimate the annual drift based on the 1000 hour @ 150°C parameter (which comes out to ± 0.263°C drift per year) and then scale this linear estimate down with the acceleration factor calculated for the operating range that my system will be working within (-20°C to 50°C). This, however, is not an Ideal method since the exact relationship between the long term drift and aging is not known (and it gives me essentially no annual drift at all).
