AudioBanter.com - View Single Post

Svante wrote:

(Dick Pierce) wrote in message . com...
John Fields wrote in message . ..
On 16 Jan 2004 06:40:38 -0800, (Svante)
wrote:

Harmonic distorsion is expressed as the ratio between the distorsion
components and the fundamental. What surprises me is that it is the
VOLTAGES that are compared (in the electrical case) not the POWERS. So
if we have a second harmonic 40 dB down, the second harmonic
distorsion is 1 %, not 0.01 %.
(In this case the voltage of the harmonic is 1% of the fundamental,
and its power is 0.01% of the fundamental)

What is the reason for this convention? I'd think that power would be
more logical.

---
It is _much_ easier to notch out the fundamental(s) and measure the
voltage of the remaining component(s) than it is to measure their power.

Then, knowing the voltage of the offending component(s) appearing across
the load and the impedance of the load at the various frequencies at
which distortion rears its ugly head, their contribution to the power
being dissipated in/by the load can be easily calculated.

Precisely. ALL of the THD meters in existance are essentially
voltmeters with filters. Going from the ancient and venerable
General Radio 1936 or Hewlett-Packard 330, through the HP 334,
through the Sound Technology 1700, Amber, Marconi and so on,
they are ALL voltmeters that have a narrow-band notch filter
in them.

Hmm, just curious, did they have circuitry to measure RMS voltage,
rather than average voltage? I guess so, because that would be crucial
to end up with a correct result. The point here is that the only
reasonable way to add up harmonics is to add their powers. That is why
the *RMS* voltage of the (sum of all) harmonics would have to be
measured.

In the standard distortion analyzers, like the HP 339, Soundtech 1700,
etc., they either use a rms circuit or an average detector to measure
(a) the level of the signal under test, and (b) the level of that signal
after the fundamental has been nulled out. No one uses a power sensor to
measure distortion or signal power.

You can also use a spectrum analyzer to measure distortion. In that
case, a narrow-band measurement is made at each harmonic frequency, but
the detector used is still either a rms detector or an average detector.

So, in all cases, you are still measuring voltages, not powers, for the
simple reason that you are supplying a signal in the form of *voltage*.