krw said:
Calorimetry is notoriously difficult. Ask Pons and Fleishman. ;-)
Yes, I used an HP true RMS voltmeter when I was in college. It was
a marvelously expensive widget and they didn't like mere students
using it. ;-) It's likely the best way to measure true RMS voltage
at high frequency. I suppose you could read current with the same
meter. ;-)
I worked on some true RMS panel meters many years ago, and they used a
heater and a thermocouple inside a glass bulb. There was a considerable
time delay to get an accurate measurement, and there was some ambient
temperature compensation required, but it was very accurate and worked at
DC to RF (at least to several MHz). But these sensors were rather expensive
and fragile.
A friend and I tried to design a true RMS meter using a lamp and a
photocell, but we found that there was a considerable aging effect on the
output of the lamp, and there was also some ambient temperature error that
needed compensation. I think we were finally able to solve the stability
and aging problems by using two lamp/photocell pairs in a sort of bridge
circuit, where one was driven by the measured signal and the other was
driven by a DC signal that also drove the meter. But the current draw for
both elements was more than the allowed specification, as it was a
self-contained meter with a range of about 4-8 VRMS.
For a bench instrument, such a method would be very practical. You just
need to make two well-matched lamp/photocell pairs, and some signal
conditioning, amplification, and limiting for the lamp, and then use an
op-amp or other means to drive the other sensor so that the photocell
outputs are identical. Then the DC current in the DC sensor matches the
true-RMS current in the other, for any waveform or frequency. But it does
have a rather narrow range of operation. You can't get very close to zero.
Probably 20% to 100% would be possible. The lamp must be driven hard enough
to become incandescent and be sensed by the photocell, which has a limited
spectral range. The heater/thermocouple could go lower, but becomes
relatively insensitive, because heat is proportional to I^2.
Paul
