y
oned
't
al
riving
th a 10
e that
rs
ure.
ot
Not always. I got a comment published in the Review of Scientific Instrumen ts pointing out that paper that was claiming +/-0.1K absolute accuracy prob ably had about 0.25K self-heating in their thermistor. The authors' reply i ncluded their measurement of the self-heating - 0.2K - and went on to agree that a little less would have been a good idea.
That was a nit worth picking.
The real problem with self-heating in a negative temperature coefficient de vice is that you eventually get to the point where you form a hot channel, but you can get funny effects a lot earlier.
I was able to report a weird event from my (then) recent industrial past wh ere perfectly good thermistors gave a (very slightly - 5th decimal digit) u nstable resistance when you dissipated a just a little too much heat in the m while measuring the resistance.