Posted by Lew on December 21, 2004 at 07:46:09:
In Reply to: Discussion on zero NFB ... posted by Matt Rogers on December 20, 2004 at 16:44:48:
You wrote, "In regards to OPT - this is my philosophy: OPT should have wide bandwidth and no
unsightly ringing when it is operating. NFB is something that designers use to improve that
performance. So to begin with, if the OPT is made well, it should already have wide bandwith and
no ringing. And when NFB is appiled, it improves it even further, and it also allows for lowering
the output impedance and damping factor. It is possible to make a transformer without the need
for NFB to have width bandwidth, low impedance, good damping factor, and no ringing - they
are just hard to make and cost lots of money. So in my view, very well made OPT will work
without any NFB, and NFB should be optional."
I think you are ignoring my simple point: you need to know something about the whole circuit,
not just the OPT, in order to assess the cause of the phenomena you measured or detected on
your 'scope. Without NFB, it may be the circuit itself that is "ringing" and losing bandwidth. For
example, were the amps connected to SL speakers or a power resistor when you made your
observations? If you did observe ringing and narrowed bandwidth at high power output, did you
back off the power to see at what point on the power curve those objectionable phenomena
disappeared? If the amps were connected to a SL, then the ringing you observed could be due to
oscillation created by the capacitative nature of the load, which perhaps would be eliminated or
ameliorated by the application of NFB. (Of course, NFB can make things worse, too.) OPTs are
not active devices; to a large degree they behave in accordance with the nature of the input
signal and the reactance of the driven load.
I disagree with you also in your contention that performance at low power output is not a
significant contributing factor to the overall sonic assessment. I do believe that it is, even with
the demanding load presented by the Sound Lab speakers. But there is no reason to argue this
point, because there is no easy way to prove it one way or the other.
However, I completely agree with you on your fundamental thesis: I don't like the effects of NFB
on music, either, in my system with my amplifiers. However, this is with a circuit (the Atma-
sphere circlotron) that works well and stably without NFB. (But more than that, I don't like OPTs
at all. That's why I prefer OTLs.) All of the above having been said, there are much more
knowledgable people than either of us who would say that a little NFB judiciously applied goes a
long way toward enhancing the performance of an amp that is designed from the ground up to
work well with it. Kevin Covi for example has designed a variant of the Atma-sphere circuit that
improves power output and reduces the amplifier output impedance with just a little NFB (and
other tricks). I have not heard it, but I have considerable respect for his abilities and superior
knowledge about circuit design. We need to keep an open mind.