HIFICRITIC ARCHIVE IV

This 1994 essay was tough on the audio industry for its uncritical promotion of the
virtues of balanced operation. Based on contemporary experience, the picture can
change with time and product improvement. See the ARC REF3 evaluation for Issue2
for an idea of progress in balanced preamplifier design.


Balance: Benefit or Bluff?
Martin Colloms, November, 1994

If you read much promotional literature for
recently introduced high-quality equipment,
you'll notice a common theme emerging:
balanced connection. Balanced inputs and
outputs are becoming a must for any audio
equipment that has any claim to quality. The
word itself has promotional value, suggesting
moral superiority over the long-established
"unbalanced" connection (for the purpose of
this discussion, I will call this "normal").
What's my problem with this? Simply this: The
High End could be paying dangerous, costly
lip service to the received wisdom that
balanced operation is the goal for an audio
system.
To give you a flavour of the literature using
balanced operation as a selling point, Audio
Research's literature for their LS5 preamplifier
describes "an imposing array of...professional-
grade XLR connectors...clearly proclaims the
fully balanced design...an uncompromising
approach. While the professional industry and
recording studio used balanced operation and
connectors as a standard for many years, the
consumer electronic industry is only now
gradually adapting its products to this superior
format."
In my view, unwarranted and unwanted
assumptions are being made in the
dissemination of such claims. I am not singling
out Audio Research for criticism at present
this approach is rife.
A state of madness currently exists wherein
several international markets are
discriminating against some excellent audio
equipment simply because it doesn't have
balanced signal connections. This is madness;
there's no reliable proof that an audio system
with balanced connections sounds any better
than a well-designed normal system. In fact,
time and again, top-line components that come
my way for evaluation suggest no advantage
and in some cases, even demonstrate
performance lossfor balanced operation.
Market pressures are so great that, much
against their better judgment, a number of
designers are succumbing to the demand for
balanced equipment, passing the extra cost on
to their customers. For example, I know of a
fine D/A processor that delivers superb sound
quality in its normal form. One of its primary
virtues is its very short signal path following
the DAC chip, which the designer refused to
sully with the usual, compromised, tacked-on
balanced output amplifier. To maintain sound
quality in the market-mandated balanced form,
the designer was driven to replicate the entire
DAC chain, driving the second DAC with a
non-invasive, inverted digital data signal in
order to achieve the pair of audio outputs for
balanced working. It almost doubles the cost of
the decoder, with no perceptible audible
benefit. [However, it does lower DAC noise by
3dB, which some will feel a worthwhile
increase in resolution, while the D/A
processor's overall linearity can also be
improved.Ed.] While this is an extreme
example, there's no dispute that balanced
designs generally cost more.
Many "balanced" products are single-ended
inside with messy, potentially sound-
corrupting conversion buffer/amplifiers at the
inputs and outputs. Some components have
more complex amplification stages in the
minus line than in the plus. Only a few
products are truly balanced from input to
output, and then costs escalate. For example, a
true balanced preamplifier, such as the Audio
Research LS5, needs a four-gang selector
switch and a four-gang volume control, for the
latter each section must be closely matched to
the others.
Normal vs. Balanced
I suspect many of us do not know what
balanced operation is, other than that it
involves a different kind of signal connector
from the familiar RCA plug.
Audiophiles have lived with normal
connectionstwo wires, respectively, for
ground and signal connectionsfor decades
without experiencing significant problems
(figs.1 & 2). Even the requirements of those
delicate low-level signals from moving-coil
cartridges can be handled satisfactorily,
provided that appropriate care is taken with
local hum fields and system grounding. And
with digital sources, the signal levels are so
high that noise and hum are considered wholly
negligible. If a normal, unbalanced connection
can be made to work for a tenth of a
thousandth of a volt signal from a low-output
phono cartridge, it must be cast-iron reliable
for the 1V or 2V signal from a modern digital
source.





Fig.1 An unbalanced signal
connection.






Fig.2 A typical unbalanced amplifier input
stage (which may be either discrete or in the
form of an IC). For balanced working, the
signal's positive-polarity connection (+, or hot)
and negative-polarity connection (-, or cold)
are segregated so that the - is no longer
combined with the earth, shield, or ground
line. Instead, a three-wire connection allows
the ground line its own separate identity. The
opposite-polarity + and - signal lines may then
be said to be balanced, or set equal with
respect to the ground (figs.3 & 4). This balance
is not engineered for aesthetic or philosophical
reasons, but so that the next input in the chain
can be arranged to measure the difference
between the two audio lines. Well-designed
balanced inputs operate as differential circuits,
and any unwanted noise, hum, or interference
common to both lines is therefore cancelled
(fig.4). The recovered signal is the difference
between the hot and cold signals, and is
therefore twice the level of either on its own,
and is why substituting a true balanced
connection boosts the level by 6dB compared
with an unbalanced one, all things being equal.

Fig.3 A balanced signal connection.

Fig.4 A typical all-balanced preamplifier line
stage. The hot signal (H) consists of the
wanted signal (A) plus noise (N); ie, H=A+N.
The wanted cold signal is A in antiphase, -A,
and again the same noise picked up by the
cable, N; ie, C=-A+N. Because the noise is
the same in both hot and cold lines, it is
termed "common-mode" noise. Provided the
amplifier's differential input has total
common-mode rejection, it subtracts the cold
signal from the hot, with an output D =
(A+N)-(-A+N) = A+A+N-N = 2A. Note
complexity compared with fig.1 circuit. In
addition, the resistors marked "R" must be
closely matched between hot and cold signal
polarities to maintain common-mode
rejection and minimize distortion.
This is a powerful and valuable technique
when you need to use it. When a newscaster
needs an audio connection from his or her
mike to the producer's console three rooms
away, the cable carrying the low-level signal
may traverse power and control cable ducting
for the buildingtypically a run of 30m or
more. An effective means of low noise
transmission is vital in this situation. Or, take
a large orchestra at a recording session: local
spotlight mikes may need 100m of cable to get
to a remote mixing desk. Likewise consider
crossed pair mikes hung from the auditorium
ceiling for ambience pickup, these will need a
long cable run. Long runs spell induced noise
pickup.
If the high noise rejection levels possible with
balanced operation are not required, then the
technique's drawbacks come into focusit
adds complication, complexity, and cost to
audio components. Disciples of short-path
design know only too well how each additional
stage can potentially step you down the fidelity
ladder.
So consider the arguments for and against
balanced design for high-quality consumer
audio components:
For:
1) High electrical noise immunity to both
induced signals (Radio Frequency and
Electromagnetic Interference, RFI and EMI)
and also chassis ground-hum loops.
2) Essentially transient free, hands-on cable
connection; with the ground made first, contact
transients are suppressed.
3) Standardized XLR connectors are
mechanically self-locking; make good, gas-
tight contacs to a specified close tolerance; and
generally have good cable strain-relief fittings.
4) With balanced cable, the + and - signal
paths are equal; ie, they use the same
conductor type, which improves the potential
for good cable sound.
5) Effective over very long runsgreater than
60'and/or for very-low-level signals of very
wide dynamic range, such as those from
microphones.
6) Professional and broadcast studio use
implies quality by association.
7) Safety requirements for effective chassis
grounding can be easily met.
8) Encourages good practice for low
electromagnetic radiation and good immunity.
9) For the digital interface, it provides a
beneficially higher signal operating level.
Against:
1) Greater cost.
2) Greater extent and complexity of audio
circuits.
3) May results in matching difficulties with
existing normal single ended equipment.
4) a fairly restricted choice of cables.
5) Restricted choice of connector grades; the
best XLR types are technologically inferior to
the best, albeit costly, RCA plug types, though
the average XLR is undoubtedly superior to
the average RCA
6) Partial disagreement between Europe and
the US over pin code connections.
7) No guarantee of superior sound to well
designed unbalanced systems.
Skirmishes
Although my primary reference system is not
balanced, until recently I've gotten by without
serious trouble. I've had many skirmishes with
balanced equipment, but I've been able to
supplement my few balanced reference audio
components with validated, current-review
loan stock as and when required. However,
when I was asked to review the Audio
Research LS5 line preamplifier and its
matching PH2 phono preamp ( Hi-Fi News &
Record Review, February 1994, Vol.39 No.2,
pp.30-35), I was brought face to face with the
full implications of true balanced operation.

All the usual mixing and matching review
procedures were useless; it was out of the
question to make direct comparisons with
other unit audio components.
The evaluation proved to be a fight every inch
of the way, precisely because these Audio
Research components are of true balanced
design; ie, fully balanced differential circuitry
is present at their every stage, including cross-
coupling between hot and cold signal paths. It
didn't take me long to realize that you can't
mess with true balanced components; fully
balanced connections, in and out, is all that
will be tolerated if satisfactory hum and hiss
performance is to be obtained. With the ARC
equipment, this meant that comparison with
normal phono equalizers and line controllers
was almost impossible. If I didn't know better
(and perhaps I don't), I could be persuaded that
the closed and self-perpetuating circle
generated by wholly balanced systems is a
deliberately constructed barrier raised against
the true comparative assessment of their
subjective quality.
For example, take an Audio Research V-series
power amp with its balanced input (again, I'm
not singling out ARC). In my experience, it
won't perform at its best unless it's driven
balanced. If you want to hear this amplifier in
an unbalanced system, then I strongly advise
you to acquire the relevant
unbalanced/balanced converter, the Audio
Research BL1. Regardless of the latter's
exceptional quality, you've necessarily added a
second set of connectors, cabling, grounding,
and expense.
Yes, it's true that operating the amplifier with a
high-quality balanced control unit, such as the
LS5, facilitates evaluation of the amplifier
that is, assuming that you're totally familiar
with the balanced preamp's quality. If not, you
may have to check it out by turning to your
favourite digital source, which, again, most
likely will have "normal" outputs. You
therefore have to bring another
unbalanced/balanced converter into action,
again resulting in an unwanted step in the
chain.
Cautious words
Enough examples. With no disparagement
intended against the fine Audio Research
products mentioned above, my experience with
balanced components has been that they're not
necessarily better than normal components.
Indeed, they can often sound better via their
subsidiary "normal" signal connections.
Enough equipment has passed through my test
setup that I feel comfortable making a few
cautious generalizations.
True, the balanced condition does result in
lower noise levels and improved immunity to
EMI and RFI, local ham radio, CB operators,
or switching pulses from heating or
refrigeration thermostats. On the debit side,
stereo images may lose absolute focus, stage
width, and depth. Sounding less "locked in" to
the music, some balanced audio components
often exhibit losses in dynamic resolution,
dynamic contrast, and rhythm. Typically
smooth-sounding and laid-back, a balanced
component can be less involving, lending the
music a "downbeat" impression.
This essay shouldn't be seen as a write-off of
all balanced components. The differences I
have discussed are not huge, though the
differentiation of such differences are partly
what the High End is, or should be, about.
Interestingly, the characterizations I mention
are also those associated with less-than-first-
rate connectors. Taking this together with the
necessary additional circuit complexity
required by balanced operation, these factors
together make some sort of sense. Something
as simple and direct as a Conrad-Johnson
Premier Eleven power amplifiera "normal"
input design probably couldn't be built in
balanced form without compromising its
engineering purity and its sonic purity.

Grounding 'n' timing
There's an element of circuit philosophy that's
relevant here: In my experience the most
dynamic sounding, best-timed audio systems
have fundamentally good system grounding. I
don't mean just the chassis ground paths. We
have to consider that the signal ground is
defined as a true reference point, carefully
organized, even to the extent of a hierarchy of
grounds focused at a single star point. These
grounds are scaled and organized according to
their role and the signal section they serve.
Such organization, if successfully effected, can
endow the system with a fine sense of poise
and equilibrium, if I may put it this way, with
the sound set firmly on the ground rather
than precariously "balanced" above it.
Once a fast, well-tuned audio system has been
set up, it's fascinating to observe how quickly
and catastrophically its sound can be made to
fall apart subjectively, piece by piece, as the
main structure of the electrical organization
and grounding is progressively disturbed or
dismembered. In one highly tuned example, a
sophisticated audio system featuring active
crossovers had been set up in every detail save
the AC mains supply. I was worried about its
conspicuous failure to perform properlythe
just "improved" system actually sounded less
musical and entertaining than the simpler
passive implementation of the same
components, used with lower-grade power
amplifiers.
Seeking guidance, I was advised that, in
addition to the centralized, focused grounding
so carefully designed for this system's signal
paths, the AC connections must also be
similarly centralized. Although I was using a
pair of high-quality, dedicated AC lines, the
multiplicity of system power cables
necessitated the use of multi-way socket strips.
Accordingly, I detached all the AC plugs and
compression-bonded all the 10A-rated IEC
power cords to a 30A terminal block, this in
turn was plugged-in to one high-current spur
on the AC mains supply.
In this critical context the sonic transformation
seemed little short of amazing. A whole
spectrum of audible changes occurred. For
example, a previously heard degree of
midrange glare and harshness was sufficiently
reduced that it could now be accepted as a
normal part of the tonal balance. Stereo focus
and depth were improved. Treble grain and
roughness were smoothed out, while bass took
on a new dimension of depth, solidity, and
dynamic slam. Previously, this system had
conspicuously fallen short in rhythm and
dynamics. Now these were to a reference
standard.
A question of balance
At the leading edge of today's high-end audio,
the overall quality of an audio replay system is
known to show a strong sensitivity to overall
design and organization: in particular, issues of
interconnect cable choice, grounding, and
power supplies. In theory, balanced operation
should free audio systems from such
dependencies.
In practice, however, the consequences of such
freedom appear to be losses in absolute sound
quality, particularly in the areas of "foot-
tapping" involvement and dynamics. Perhaps
there's a lack of rigor and critical assessment in
the design of balanced components. Maybe
good performance is taken for granted. The
adaptation to balanced working is unwittingly
used as a problem-solving crutch; the
technological benefit obviously lies in easily
produced, impressive specifications for
signal/noise. But what about sound quality?
We still cannot adequately measure that.
I consider that the industry should take a
critical look at this headlong rush to balanced
systems, weigh the costs, and be honest about
the advantages, if any to the home consumer.
Most importantly, audiophiles must be
objective about changes in sound quality.
Balanced mode may turn out to be a
fashionable whima device to promote
salesand may not add significant musical
value even at the highest quality level.


end