Page 7
Once it was recognized that excessive use of negative feedback was creating problems with
the sound, several designers addressed the problem by simply reducing the amount of
feedback and regaining the performance by paying more attention to the character of the
amplifying circuit itself. Feedback stopped being a “something for nothing” idea, and became
more like a credit card, which is OK to use as long as you can afford to pay when the
statement arrives. In this case, the ability to pay involves the intrinsic quality of the amplifier
circuit. The paradox is that feedback is best applied around circuits that need it the least.
One of the alternatives is the use of “no feedback”, or more accurately what is known as only
local feedback. I say this because purists might argue that local feedback is still feedback. In
point of fact, there is always some amount of feedback locally around any gain device by the
nature of the device. So I will state here and now that I consider “no feedback” to be where
feedback does not extend further than a single gain device or stage, so that circuits having
“local feedback” are still considered “no feedback”. Anybody disagreeing with this should send
me a diagram of a “true no feedback” circuit, and I will try to point out the hidden feedback.
On the push-pull front, a major improvement was offered by Class A operation, not a new
concept, which delivered significantly better performance by sending a much larger amount of
current idling through the gain devices. This lowered the distortion of the gain devices
dramatically, but at the cost of high heat dissipation. Operating an amplifier in Class A mode
was, and remains, an expensive proposition compared to conventional designs, not
necessarily so much in wasted energy, but in the cost of the heavier hardware needed to
deliver and dissipate the additional heat.
One of the important potential advantages of Class A operation is the possibility for simplified
circuitry requiring little or no feedback because of the much more linear performance of gain
devices biased to a high current. By the mid 1970’s the marketplace began to see high end
solid state amplifiers offering varying degrees of Class A operation in their output devices,
although as far as I can tell, at the time none of them took advantage of Class A operation to
create simpler circuits with less feedback. Mine didn’t, in any case.
Also about this time Matti Ottala introduced the concept of Transient Intermodulation Distortion
(TIM), in which the overuse of feedback, coupled with slow amplifier circuits was identified as
the major culprit in bad sounding amplifiers. It was all the rage for a while, but is no longer
touted with such enthusiasm. The solution to TIM is low amounts of feedback coupled with
fast amplification (high slew rate).
In retrospect, the idea was at least half right, but I believe not completely for the following
reasons: First, it presumed that there was really fast signal in music. Research conducted
independently by Peter Walker and myself showed conclusively that real music contained very
little signal with appreciable slew rate, therefore slew rate limiting on the order proposed by
Ottala was pretty unlikely. Further, all those good sounding tube amplifiers had terrible slew
rate figures.
However, while slew rate limitations of an amplifier might not be the cause of bad sound, it did
correlate to sonic performance in the following manner. It turns out that there are two ways to
make faster amplifiers, the first way being to make the circuit more complex. The second is to
make it simpler. Video amplifiers, which must be very fast, are very simple. Tube circuits tend
to be very simple also.
