K7ITM wrote:
I fully agree with Roy's comment about the AGC-derived S-meter output.
As he says, the AGC characteristics are temperature dependent, and they
are also dependent on the particular set of active devices (and to some
extent the passives too) in the signal path.

If you have access to a spectrum analyzer, chances are decent that it
will have a well-calibrated amplitude readout. The one I use is
accurate in a relative sense to a fraction of a dB over more than an
80dB range, and would certainly be sensitive enough for antenna
comparisons. I suppose both those would be true of most modern
spectrum analyzers. In addition, some are quite good at determining
the total power in a specified frequency range, and if you can find
such a range with no signals, you can get a better reading on noise
than you're likely able to do with an S meter, or even a narrow band
spectral measurement.

One of the most complex and difficult parts of a spectrum analyzer to
design is the log amp which provides this stable and precisely
logarithmic response over a wide dynamic range. There's an incredible
amount of really ingenious work on the part of some extremely talented
engineers in those circuits. In relatively recent times, Barrie Gilbert
and his folks at Analog Devices have done some equally clever work in
the design of IC log amps. It's not a trivial task by any means.

Roy Lewallen, W7EL