"Sander deWaal" wrote in message
...
"Robert Morein" said:

The Hafler box does not compare two amps. However, it does compare the
amplifier to the input, ie., to the mythical "straight wire with gain."
It
will extract every change the amplifier has made to the input, and
display
just the change.


An experiment I sometimes do with a just finished amp: connect a dummy
load consisting of a resistor and capacitor in parallel, and playing a
CD.

Then another amp of known reputation connected to the feedback signal,
reproduces the error signal that feedback loop nr. 1 creates.

Very revealing!

Of course, a double trace oscilliscope and a low distortion generator
will yield results that are better suited to produce on paper.

The Hafler box does the same IIRC, but with a HF compensation that has
to be nulled out or something (have to go through my stash of
magazines, I know it was described somewhere).

You are thinking of the HF compensation on the Excelinear amp design, an
optimization adjustment.
The Hafler test box has no adjustment except for a level matching pot on the
voltage divider, so that the bridge can properly compute the difference.

OK, you have the feedback loop signal, but the actual correspondence of the
loop signal to the error in the output depends upon the loop gain. Do you
feel the observation gives as accurate an assessement of the error as a
nulled comparison of input to output?