A Touch of Class: How to Speak Ampese
Don't look now, but your stereo system is crawling with
amplifiers. Most of the basic functions of electronic components have something to do with
making a signal bigger, even if it's only to compensate for level reductions caused by
other processes, so there are tiny amplifiers in your tape deck, FM tuner, preamp, signal
processor . . . in fact, almost everywhere. But this sort of amplification is hidden
because it is incidental to whatever the particular component is supposed to be doing.
There is, however, one big amplifier: the one that
drives your speakers. Ironically, in many ways this one is almost as obscure to most users
as all those little amps the signal passes through on the way. In its most obvious form,
as a separate component, the power amplifier is basically a "black box" that
takes tiny electrical impulses and boosts them to a point where they can drive an array of
speakers. Anything else that is included on the chassis, such as level controls or meters,
is peripheral to the amplification function itself. But most of us are not even conscious
of our power amplifiers because they are buried within receivers or integrated amplifiers.
All this inconspicuousness makes amplifiers hard to sell,
and it's even harder for an audiophile to choose between them. Unlike a CD player or a
cassette deck, a power amplifier has no real convenience features, and except for its
power output an amp's specifications don't help much because most amplifiers are so good
that the differences between them are inaudible. Even price tends to depend more on the
maker's reputation, or the styling, or added features like meters than on performance.
And yet there are a lot of amplifiers out there, so a
manufacturer has to give the prospective buyer some reason to choose his model over
a competitor's. He is likely to lean on a unit's specs to some extent, of course, but more
than with other components he is apt to talk about an amp's circuit design and
construction.
Whether or not these really matter very much to a consumer
is debatable, but advertising such things has had an important consequence: amps were
pretty good to begin with, but you can't sell that, so audio engineers have advanced the
state of design considerably over the years largely so they could say they had done so in
promotional material. Whatever the motivation, the result has been improvements that, if
nothing else, have brought down the average cost of amplifiers and increased their
stability and reliability.
What are all these improvements? Because
developments in amplifier design have been at the far edge of electronic technology, they
tend to be described in fairly obscure terms -- obscure to the average audio buyer, in any
event. But if you are attempting to make an informed selection of an amplifier on the
basis of its technical characteristics, you have to understand some of the more common
phrases that you will encounter along the way. Then you can decide for yourself which are
important.
The first thing to understand is an amplifier's
"class," and that requires a little background. An audio signal, whether it be
sound traveling through air or its electrical analog in a circuit, has both a positive and
a negative component -- drawn on a graph, the waveform goes above and below the zero line.
But the transistor (and the vacuum tube before it) can only pass electricity in one
direction, which means that if you simply push an audio signal through it
"straight," half of it -- the negative portion -- will be missing.
One solution is to add to the audio a "bias"
signal in the form of direct current (by definition a 0Hz signal). This won't affect the
audio sonically but will shift the whole works into the positive region. The original
positive/negative variations become more-positive/less-positive ones in this
configuration, which is called "Class A." In terms of high fidelity, Class A
amplification has always been considered the ideal because it has the lowest distortion
and the flattest frequency response.
But Class A has some major disadvantages. In the first
place, in this type of amp there is always current passing through the transistors, so
they get very hot and elaborate measures (which means large, heavy, and expensive
measures) have to be used to dissipate the heat. Class A amplifiers are also very
inefficient, which means that they use a lot of power but don't produce a
lot of power. Still, some very fine pure Class A amps do exist, mainly for those
audiophiles willing to put up with their disadvantages in order to obtain the
theoretically best sound.
Early on, however, the problems of Class A were tackled by
audio designers, who came up with Class B. In this scheme, the transistors are allowed to
cut off the negative portion of the audio signal. The positive part is amplified normally,
exactly as it would have been with a Class A circuit but without added bias. The negative
portion is amplified separately by another transistor that thinks this part of the signal
is positive, and the two amplified portions are added together (out of phase, of course)
to re-create the original audio waveform. Class B operation is much more efficient than
Class A, and it also produces a lot less heat because each transistor is off half of the
time.
But Class B is also not without drawbacks. One of the main
ones derives from the fact that transistors are less linear near their zero point, which
is where they are operating most of the time in Class B. Distortion is therefore greater
than with Class A, whose bias pushes the signal up to a flatter part of the circuit's
operating range.
The answer to this distortion problem has been Class AB,
which uses a small amount of bias, so each transistor operates slightly more than half the
cycle but still has the benefit of being off for a significant portion of the time. In
Class AB, each transistor operates in a more linear part of its curve, but it retains most
of the efficiency of pure Class B operation.
But Class B and Class AB share one flaw: as the transistors
switch on and off, they produce sharp spikes than can be audible. This effect is called
switching distortion, and it has been a main area of concern for audio designers over the
years. The solution has generally been to use some residual or variable bias to make the
switching "soft" and therefore less obtrusive. While these designs are still
variations on the Class AB theme, they go by a variety of proprietary names, many of which
manage to imply that they are a form of Class A.
Other classes do exist. In Class C the transistors deal
with less than half of the waveform, which produces massive amounts of distortion;
Class C is therefore unsuitable for audio use, but it does have some radio applications.
The term Class D is sometimes used to describe the digital (or pulse-width-modulation)
output stages that have appeared from time to time. Other "classes" have tended
to be unofficial commercial designations, such as Hitachi's Class G of some years back.
The class designations refer to an amp's fundamental
circuit design. Engineers have always realized that none of these circuits is perfect,
however, so a number of techniques have been developed to correct flaws as they occur. The
most imaginative, and the most common, is "feedback," in which one part of the
signal corrects nonlinearities in another.
The type most commonly used in amplifiers is negative
feedback (NFB), in which a small portion of the amplifier's output is inverted in phase
and fed back to the input. If the amplifier were perfect, this would simply cause a
reduction in overall gain, since the out-of-phase signal would cancel a portion of the
main signal -- a flat signal would be uniformly attenuated across the audio spectrum. But
amps are not perfect; virtually all of them have some nonlinearities, whether these are
resonant peaks or merely gentle deviations from flat response. And the more a signal
deviates from linearity, the more the negative feedback corrects for it -- it's an
automatically self-regulating system.
NFB is almost universally used, but that doesn't mean it
has no problems. For one thing, too much NFB can cause oscillation in the amplifier,
though that's rarely a problem these days because most amps are pretty good even without
correction and thus need relatively little of it. The main drawback of NFB is loss of
efficiency: negative feedback reduces an amplifier's gain across the board, so a bigger
amplifier -- that is, one with greater "open-loop gain," or gain before feedback
-- must be built to obtain a given output.
One solution to this problem is called
"feedforward," in which a comparator circuit continuously mixes a portion of the
input signal with a tiny bit of the output signal, inverted in phase. The mixing is
adjusted so that the two signals cancel each other entirely, leaving only whatever
distortion products have arisen. These products are then amplified separately and
subtracted from the output, a process that takes very little power because distortion is
typically only a tiny fraction of the complete audio signal. In theory, a feedforward
system should enable an amplifier to produce its full open-loop gain, but usually it is
used in conjunction with a moderate amount of negative feedback as well.
Another design wrinkle that sometimes appears is one that
makes an amplifier capable of being "bridged" or "strapped" to mono.
This is simply a method of joining the outputs from the two halves of a stereo amplifier
to produce a single feed. This can't be done with just any stereo amplifier, however; the
capability must be designed in so that the two parts are properly balanced to work in a
true "push-pull" fashion. One remarkable feature of bridged amps is that they
are usually capable of producing more output in mono than the sum of their stereo outputs.
More common is the direct-coupled (or DC) amplifier. A DC
amp has no capacitors in the signal path, particularly in the output stages, so phase
shifts and low-end rolloff are minimized. The amplifier can respond down to 0Hz, a
theoretical ideal. It should be noted, however, that this same capacity means that the amp
can reproduce things like the warp frequencies of vinyl records or the inherent rumble of
recordings made in "live" venues at very high power. As this can easily damage
speakers, most direct-coupled amplifiers include a switchable infrasonic filter to prevent
such low-frequency material passing through.
Much amplifier terminology revolves around the sort of
devices used in the design, rather than the design of the circuits themselves. There are a
few rather exotic exceptions, but the great majority of amplifiers today are solid-state
designs -- that is, they use transistors. There are various sorts of transistors, however,
and their names have found their way into amplifier terminology as particular
characteristics are attributed to them
The original junction transistor was "bipolar,"
which means that the signal could pass through it in two ways: by the negatively charged
electrons in one direction and by the positively-charged "holes" in the other.
The result was the same in terms of the overall polarity of the device -- all transistors
are one-way conductors (or semiconductors). Good amplifiers could be made with bipolar
transistors, but a lot of compensation for their limitations had to be built into the
circuits, and often these compensations resulted in what was known as "transistor
sound."
Eventually a unipolar device came along called a
"field-effect transistor," or FET. Compared with bipolar transistors, FETs more
closely approximate the characteristics of a vacuum tube, as well as offering higher
switching speeds and a greater degree of stability. Audio designers -- many of whom grew
up designing tube amps -- adopted the FET, which is now more or less standard. Various
versions go by different designations, depending on how they are built or which
semiconducting material is used, and all vary slightly in performance, though not enough
to matter very much in an amplifier. The most common sort is the "metal-oxide silicon
field-effect transistor" or MOSFET.
If the actual bits and pieces used inside an amplifier are
really only of interest to the designer, one facet of the design has a direct bearing on
the consumer's interest because it affects an amplifier's size and cost. This is the power
supply.
An amplifier has one basic function: to take the power
available from your friendly neighborhood electric company and modify it by means of an
audio signal so that it can drive the form of motor called a loudspeaker. Much of the
discussion about amplifiers has to do with the audio modification, but how an amp deals
with the electric utility's power in the first place is also extremely important.
A power supply does three things. First it takes the
alternating current supplied by a house's electrical system and converts it by means of a
transformer to a more usable voltage. Then it rectifies the current, changing it from AC
to pulsating direct current. Finally, it filters out the pulses to produce a constant
source of direct current that can be modified by the audio signal.
Traditionally, all three jobs have been done by massive
components, particularly by monstrous transformers, that contribute most of the weight and
size to an amplifier. These components, especially the filter, have to be of very high
quality because even the slightest bit of pulsating DC can creep into the system as hum,
and that makes their cost high. In addition, power supplies are usually far larger than
necessary most of the time. The reason for such "overkill" is that musical peaks
require all the juice an amplifier can muster, even though they occupy only a tiny
fraction of the amp's operating time.
There have been several attempts to get around this
situation, although for the time being we seem to be stuck with massive power supplies.
One approach is to use toroidal transformers, which are ring-shaped devices that are much
more efficient than conventional transformers of the same size. Toroidal transformers
allow amplifiers to be made smaller, but they impose requirements on the circuitry that
increase costs: they operate much better at high frequencies, so a preliminary circuit
must be inserted to push up the operating frequency.
Other approaches include various methods of regulating the
power supply according to the input signal, so that high power levels are only produced
during the small amount of time they are necessary. So far, these methods have only been
tried by a few manufacturers.
Amplifiers, and hence the terms used to describe them, are
at the very heart of audio and thus worth understanding. It's a tribute to amplifier
manufacturers that their products are generally good enough for you to be reasonably safe
buying one even without knowing the jargon. But there's nothing like understanding the
language of a country to make you feel more confident when you enter the territory.
...Ian G. Masters
ian@mastersonaudio.com
|
