[time-nuts] temperature sensor

[Bill Hawkins]

Group,

I worked for Rosemount, a manufacturer of precision resistance
thermometers, for many years.
Platinum does have a well-known formula for temperature versus
resistance, with second order corrections.

But a sensor is not enough. You need to convert its physical property to
a signal that is useful.
This is done by some sort of temperature transmitter or conversion
device. The device and the RTD are a system.

The temperature system is calibrated as you would any other such system,
by comparing it to a standard that is 10 times more accurate.
So the question is, how accurate do you want it to be, just as it is for
time and frequency standards.

Boiling water with an ambient pressure correction is fine for some
systems. More accuracy requires more purified water and a better
pressure measurement. Similarly, the triple point of ice, water, and
vapor depends on purity and knowledge of ambient conditions, as well as
the heating effect of the stirrer. And that only gives you two points,
with no knowledge of nonlinearity in between or outside them.

There is plenty of literature on the subject, but it is not in the scope
of precision time and frequency measurement.

It is, however, summertime.

Bill Hawkins