Amplifier Design 717at the individual load devices be of high quality. Poor
quality transformers with either high insertion losses
and/or poor impedance characteristics will completely
defeat the advantages offered by the constant-voltage
distribution technique.
Many of the present-day power amplifiers intended
for professional use are produced in two-channel
versions even when the ultimate employment is to be
with monaural program material. When preceded by
active or passive crossover networks, such amplifiers
can provide biamplification by devoting each amplifier
channel to a separate part of the audio spectrum. This
technique when properly employed may offer level
adjustment, distortion, and loudspeaker damping advan-
tages over the full-spectrum approach employing an
individual amplifying channel. Additionally, such ampli-
fiers may be employed with a balanced bridge output
driven by both channels, which doubles the output
voltage swing but requires a load impedance that is twice
as large as a single channel alone. This technique can
drive a 70.7 or even higher voltage balanced distribution
line without a transformer at the amplifier but the ground
isolation previously mentioned is lost in the process.
This may furnish the user with a difficult choice.
Audio power amplifiers are designed in reverse,
which means that the output stage is designed first
followed by the design of the output driver stage that, in
turn, is followed by the required intermediate stage or
stages and then finally the input stage. Depending on the
power, distortion, and efficiency requirements the class
of operation of the output device or devices has tradi-
tionally been restricted to A, AB, B, or D. The most
recent developments have widened the choice somewhat
in that some current designs involve changing the supply
voltage to the output stage under dynamic conditions. It
appears that one may look forward to an entire alphabet
of classes of operation. When a single device is
employed in the output stage, class A operation, in
which current exists in the active device throughout a
complete cycle of signal swing, is the only acceptable
class of operation. Class A is inherently the most linear
class of operation. If pairs of output devices are
employed in push-pull in the output stage, then classes
A, B, AB, and AB plus B (at least two pair of devices)
are distinct possibilities. Other than A, the other classes
are in general more efficient but inherently are not as
linear as class A. In class B operation each member of a
push-pull pair is active over only one-half of a complete
sinusoidal signal cycle. Class AB is intermediate in this
regard between A and B. In AB plus B a pair of devices
operates push-pull in class AB while a second pair of
devices in push-pull operates nearly in class B. Class D
is the designation given to the mode of operation
wherein the output devices are operated in a switching
mode. This means that the output devices are conducting
as heavily as possible or not at all. This mode of opera-
tion offers efficiencies bordering on 90% but introduces
a host of other problems with regard to radio-frequency
interference as well as requiring specialized active
devices, drive circuitry, and design techniques.
The advent of bipolar complementary symmetry
transistors introduced the possibilities of many new
circuit topologies in power amplifier design and the
development of complementary symmetry power field
effect transistors has opened up even more exciting
avenues for truly superb amplifier developments.
Fig. 20-19 is a rather basic complementary
symmetry bipolar transistor output stage for operation
in classes A, AB, and B. The class of operation is
dictated by the details of the biasing and drive arrange-
ments.The currents in Q 1 and Q 2 are equal at quiescence
and there is no current in the load. If the bases of Q 1 and
Q 2 are driven with a positive-going signal, Q 1 conducts
more heavily while the current in Q 2 decreases, thus
producing a net current in the load directed from left to
right such that the left end of the load assumes a posi-
tive voltage relative to ground. On the other hand, if the
bases of Q 1 and Q 2 are driven with a negative-going
signal, Q 1 conducts less heavily while Q 2 conducts
more, thus producing a net current in the load directed
from right to left such that the left end of the load
assumes a negative voltage relative to ground. The load
in effect is connected in the emitter circuits of both tran-
sistors that consequently operate as common collector
transistors. As is well known, the voltage gain of a
common collector amplifier is slightly less than one and
without polarity inversion. Hence, the driving circuitry
must be able to produce a signal swing in excess of the
swing to be expected across the load.Figure 20-19. Complementary symmetry output stage.LoadQ 1Q 2Bias and driveFeedbackBias and drive++