[Trevor Wilson[_2_]]

"keithr" wrote in message
...

"Trevor Wilson" wrote in message
...

"keithr" wrote in message
...

"Trevor Wilson" wrote in message
...

Good transistor amps function perfectly well with wider tolerances
than
you say because of the abundance of
local and other NFB.

**"Good" is your term. not mine. I happen to feel that amplifiers which
use lots of Golbal NFB are not "good". They are average.

Trevor Wilson

Speaking as one who has never seen, heard, or even smelled an ME amp
perhaps you can clear something up for me. As I understand it the ME is
made up of several modules in the signal path. Each of these modules has
it's own feedback loop.

**Not quite. The amplifier is not all that complex. It is pretty
standard, full complementary symmetry thoughout, dual differential
inputs, with the obligatory active current sources. The Voltage amplifier
stage uses around 15dB of Global NFB to stabilise Voltage gain to the
desired level. The Voltage amp feeds the current amplifier, which uses
local feedback only (no loop, from output back to input). The output
devices are coupled to the load via the collectors, rather than the more
usual emitter follower configuration. So, no, each module does not have
it's own feedback loop. Only the Voltage amp has a feedback loop.

Assuming that the module and it's feedback path are correctly
designed and thus pass an undistorted signal to the next downstream
module, why does this require such tight tolerancing in the output
stage?

**Because there is no loop feedback around the output stage.
Additionally, the feedback that does operate, is lightly applied.

I do
assume that the ouput stage module has it's own feedback loop and is
being fed an undistorted signal.

**The output stage uses no feedback loop. All feedback is local only.
IOW: Degenerative feedback only.

Maybe a little feedback around the output stage would eliminate the need
for such close matching without changing the sound, but I suppose that in
audio, everybody has to have their own gimmick to differentiate their
product.

**There's the rub. Peter Stein did build some sample amps which used some
Global NFB. Distortion was reduced (from what was already an inaudible
level) and it would have reduced the dependence on matched devices. HOWEVER,
blind testing showed that most listeners preferred the zero Global NFB
approach. I was one of those listeners. I received two, otherwise identical,
amplifiers. I was asked to judge which I preferred over a period of several
weeks. I was not told what differences there were, nor what to expect.
Whilst the differences were not earth-shattering, they were obvious.

Trevor Wilson