Consoles 915tage that large amounts of the bus can be run in shielded
cable. The extra capacitance here does not have the
awful consequences it does with long virtual-earth
summing amplifiers.
For consistency—if this approach is taken—all buses
should be run devolved. This means the submix facili-
ties for the PFL buses, effect sends, foldbacks, the main
stereo/monitor mixer, and analog subgroups (if used).
Also, provisions must be made to arrange the master
mixer for each of those at the grouping end.
25.13.7 Balanced Mixing
The earliest form of signal mixing consisted of directly
paralleling the sources, which were generally
medium-impedance (nominal 600:) and balanced.
This form of passive balanced mixing persisted until
semiconductor electronics and its readily achieved zero
impedance transpired. The balancing was done entirely
by transformers; again, things that have fallen at least
partially by the wayside. As a technique it was simple
(for the technology at the time) and maintained all the
advantages balanced systems have in general—princi-
pally a welcome robustness and immunity to interfer-
ences, induced noise, or crosstalk.
Balanced or differential mixing has became prac-
tical again with falling component costs and the devel-
opment of simple electronic differential and floating
balanced input and output circuits (see Sections 25.9.6.1
to 25.9.6.4). Fig. 25-90 shows how differential sources
of the trivial kind (straight and inverted) can be mixed
onto a balanced virtual-earth mixing bus, created and
sensed by a superbal input stage.
Although requiring a comparatively large number of
parts, the performance of such an arrangement in the
context of a large multitrack console is truly staggering,
especially noise, head room, electromagnetic field
rejection, and crosstalk. The noise improves in two
respects:
- No longer is the mix-amp amplifying the noise on
its reference ground. It is referenced to itself,
effectively. - Square law noise summation—twice the signal
(coherent) means 6 dB gain, two lots of incoherent
noise 3 dB gain, bingo, 3 dB noise advantage.
Head room, by virtue of two signal paths carrying
the same information differentially, is 6 dB higher.
(Naturally the noise and head room are interrelated;
whichever is more pressing in a given circumstance
necessarily takes precedence in the level architecture.)
The RF field and crosstalk rejection improvements are
dramatic, but they really ought to be expected from the
naturally self-canceling nature of balanced systems.
All the problems of keeping virtual-earth mixers tidy
and stable apply twofold here; of course, bus buffering
is strongly recommended, mostly to allow the band-
width definition around the superbal to be effective.
Passive balanced mix-amps can be arranged around
integrated mic-amp devices such as the THAT 1510;
being single-ended output doesn’t lend them to dynami-
cally generating differential virtual–zero impedance mix
buses, but does allow the choice of mix resistor values
versus mix width to optimize the parts mix-noise
contribution. It is presently difficult to consider any
serious large console design that doesn’t use balanced
mix buses.25.13.8 Pan PotsAs outlined earlier pan pots are a means of positioning a
monophonic image somewhere within a stereophonic
image plane. About the simplest pan pot is shown in
Fig. 25-91A where a pair of linear potentiometer tracks
are complementarily wired; one goes up, the other goes
down. All well and good and even the sums work out
nicely; if the L and R outputs are subsequently remo-
noed the summed signal remains at the same amplitude
regardless of pan pot position—center, either end, or
anywhere in between. Subjectively, though, the image
seems too loud at the peripheries (i.e., extreme left and
right) and subdued in the middle.
Replacing the linear pots with a ganged log/anti-log
pot (the log section wired upside down) performs much
the same function but with a different law, Fig. 25-91B.
If a signal is panned steadily right, the left hand output
is steadily attenuated, leaving the right output fairly
steady in level (in practice it shifts about 1 dB). The
center position sees both L and R only attenuated
slightly (<1 dB) with respect to the starting mono
signal. Not surprisingly this has the opposite subjective
effect to linear pots: the image seems louder in the
middle than at the peripheries. Despite that, often this
law is more appropriate, particularly where the pan pot
is used as part of multitrack odd/even panning or in use
as a correctional offset control. In these cases there is
virtue in leaving at least one side fairly unscathed.
Somewhere between these two extremes (if extremes
is the right expression for a 6 dB difference) should lie a
happy medium at which the signal keeps an even
subjective level panning across the image plane and also
tracks well (i.e., has good correlation between control
position and image position). Easy? This has been the