-linear, Williamson, full complemen-
tary, quasi-complementary, quasi-linear, class-G, class-
H, grounded-bridge, class-D, etc.
One common thread in all of the above is the use of
push-pull circuitry. Loy did not invent push-pull circuitry.
Class-A push-pull amplifiers were around before 1931
and were used whenever larger output powers were
needed than could be derived from a single device
vacuum tube output stage. Loy was the first to describe
class-B push-pull amplifiers which he developed to
power both the large audio modulators of AM broadcast
stations and the output stages of home radios. Both
applications had a common need, the need to produce
more high-quality power output with less electricity and
natural resources.
In 1931 Loy Barton published High Audio Output from
Relatively Small Tubes in the Institute of Radio Engi-
neers proceedings. The very theme of the article is in
harmony with the goals of every designer who has ever
wrestled with the power amplifier problem of wanting
unlimited output power from a small box of affordable
cost. There are some things which do not change.
One thing has changed since 1931 and that is the vari-
ety of electronic devices which are available to imple-
ment circuits. The original electronic power devices were
vacuum tubes which were characterized by large out-
put impedances and high saturation resistances. They
made poor power switches and were most useful when
used with transformers to match their output impedances
to lower impedance loads such as loudspeakers. Todays
solid-state devices such as power MOSFETs offer char-
acteristics which are most appropriate to make high-
speed switches, not linear output stages as practiced
by Loy.
Loys genius was to operate the two tubes of his class-B
output stage in strict time alternation. To produce one
polarity of output current he would turn on one tube; to
produce the other polarity of output current he would
turn on the other tube. Previously with class-A designs,
both tubes were always turned on and even at no signal
were dissipating large amounts of quiescent power. By
careful selection of the class-B bias point, he was able
to produce essentially undistorted output without hav-
ing a massive quiescent power loss. This greatly in-
creased the power output that could be obtained from a
pair of tubes and reduced the wastage of electricity.
While many variations on this basic theme have been
developed since 1931, Loys class-B paradigm has sur-
vived unchallenged. Operation of the push-pull power
devices in time alternation has been part of all high per-
formance designs for the last 66 years. Even when the
devices became class-D PWM (Pulse Width Modulation)
switches, they were operated using the class-B para-
digm, first one on and then the other, in strict time alter-
nation.
While switching and PWM methods are the methods of
choice to all modern power electronics engineers, PWM
amplifiers have remained relatively useless for precision
power amplification. Ironically the class-B paradigm lies
at the heart of the problem.
To produce a class-D PWM amplifier with low amounts
of distortion near zero output current, it has been neces-
sary to operate the time alternating power switches with
very precise sequencing of the two switches. If the
switches have any dead time (no switch on) between
their activation large amounts of distortion will form. If
they overlap, the circuitry would self-destruct with large
amounts of shoot-through current. The circuitry has been
designed around the paradigm and is therefore not tol-
erant of any violation of time alternation.
So pervasive has been the paradigm that it has gone
unchallenged until now. While Loys class-B paradigm
has served us all well for most of a century, its days are
numbered.
With a marked bifurcation in design concept, the para-
digm for the next century uses simultaneous activation
of its push-pull switches and has been appropriately
dubbed a Balanced Current Amplifier. This is the very
antithesis of the time alternation paradigm.
In the Crown BCA design, when there is no intended
output signal, the power switches are being turned on
and off simultaneously with a 50% duty cycle. The result
is the formation of two balanced and canceling high-
frequency output currents with no output at the no-sig-
nal condition.
To produce an output signal the output of one of the
switches is increased in duty while the remaining switch
1 ] Balanced Current Amplifier
is decreased by the same amount. Both pulses remain
centered on each other or balanced in time. The result
is that the difference ripple current has a minimum fre-
quency which is twice the operating frequency of the
individual switches.
The frequency doubling character of the output is re-
markable and further allows advancement towards Loy
Bartons goal of more from less. The switching losses
are effectively halved by this property as it is only nec-
essary to switch at 250KHz to make a 500KHz amplifier!
The result is that the operating frequency is taken to its
theoretical maximum of N (the number of switches) x fs
(the switching frequency). This is a full factor of two faster
than any known previous design.
The modulation process makes two decisions per switch-
ing cycle for each switch, as both the turn-on time and
the turn-off time are independently controlled by the
modulator. A 250KHz Crown BCA design thus has one
million switch decisions made each second. This is what
is required for full bandwidth audio operation. Previous
to the BCA the conventional wisdom correctly held that
any full-bandwidth audio amplifier would need operate
at 500KHz. Low quality or limited bandwidth PWM de-
signs have operated at lesser frequencies.
The result of the new paradigm is a convection cooled
2.5KW amplifier which mounts in two rack spaces. This
is approximately an order of magnitude larger amplifier
than could have been built previously in the same space
without any cooling fan. With no fan there is no need for
filter maintenance, no fan noise and no contamination of
the unit resulting from normal use.
The Crown K2 amplifier has all of the nearly ideal power
converter attributes of class-D PWM amplifiers in that
reactive loads such as loudspeakers are easily
driven. The reactive energy returned from the
load to the amplifier is reabsorbed and
reoutput with little loss. Non-switching
amplifiers are forced to dissipate all
of the returned energy plus much
more (the latter ratio is a function of
the topology used) and is typically
three fold or more.
Difficult loads are driven with grace
and ease. Cur rent overload is
smooth and sonically identical to
voltage overload. Thermal overload
is rendered a thing of the past as it
is difficult to produce large amounts
of heat. Conventional amplifiers
tackling the same difficult loads be-
come overloaded within minutes
and become sonically dysfunctional with either large
amounts of distortion or shutting off entirely. The result is
that a BCA output Watt is operationally larger than that
of previous amplifier designs.
Real-world high power operation of most large amplifi-
ers reveals that rated bench Watts and distortion ratings
often bear little if any relationship to what can be sus-
tained under normal field conditions by the typical user
using loudspeakers and music.
In critical studio environments the K2 is sonically flaw-
less and will outperform the best large studio amplifiers
in that it does not have the one sonic flaw that any unit
with a low-speed fan has fan noise. The K2s over 100dB
of electrical signal to noise (A-weighted) is not rendered
superfluous by fan noise.
With a low-frequency damping factor of over 10,000 and
low distortion (<0.1%THD), the K2 is ready to give your
music the quality of presentation that it deserves.
One final footnote: One (the larger) of Loy Bartons 1931
design examples was a 2.5KW amplifier. Ironically there
are some things that do not change, no matter which
centurys paradigms are in force.
PUSH-PULL
The basic concept of push-pull amplification is quite old
(1920s) and can be described as an amplifier in which
there are two similar signal branch circuits operating in
phase opposition and whose outputs are combined in a
difference (summing) circuit to produce an increased
power output.
2 ] Balanced Current Amplifier
The simplest combining method is to join the output sig-
nals at a single circuit node. This is the method used in
all power stages that are referred to in the jargon as a
totem-pole, half-bridge or single-ended push-pull de-
sign. While combining at a node is the simplest method,
it was not the method first used to produce push-pull
power amplifiers.
The second and original method of combining the push-
pull output signals was to use a magnetic device, a trans-
former with a center-tapped primary, to perform the
differencing. Transformers had been in use previously
to adapt the high output impedance of vacuum tubes to
lower impedance loads. Power output was obtained at
such a high cost that it was rarely permissible to oper-
ate a power stage with impedance mismatching.
Push-pull operation using a transformer with a center-
tapped primary as the combiner was particularly attrac-
tive as it also solved a problem implicit to transformer-
coupled designs. It was now possible to minimize (can-
cel) the DC magnetizing force produced by the quies-
cent bias currents in the primary. The output transfo