Power derating of NTC

[thexeno]

Hi all!
I was studying some thermistor (NTC) datasheets and I notices the derating curve. May be a simple thing under my nose, but I don't understand why should I derate at temperatures below 0°C.

Here what I mean, from the NTCLE203, under "DERATING AND LONG TERM STABILITY".

I was thinking something related to an accumulative self heating, so some kind of drift, but this drift depends on how the NTC is biased (and dependent on at which resistance is the maximum power transfer to the NTC), while in the datasheets is always specified below 0°C. So it must be due to something else.

 

[mikerj]

Because at low temperatures (high resistance) you will reach the voltage limit of the part before you reach it's power dissipation limit.

[thexeno]

Ok, that is actually clear.
I tried to make some verifications on another sensor, like the NTCLE100, same graph and a maximum voltage defined to be no more than 50V.

At 55°C, at 50V I will dissipate a bit more than Pmax of 500mW (something around 770mW), so make sense to have a less voltage, since the 50V maximum is a top limit that shall never be reached.
But at 0°C, the Pmax allowed is still 100%. Making a trial, for the 10k part, at 0°C has 32554\$\Omega\$. With 3.9mA is dissipating 500mW but to do that, a supply of 126V is needed. So why are they putting 100% of Pmax? Is for other parts of lower resistance in the same datasheet? Are they not so precise because normally you use those NTC for measurement and are never used with such power?

[thexeno]

Up