Consoles 919monitors and idiots being used briefly to make sure
nothing’s gone amiss or become overblown.25.14.3 Related Crosstalk
In a program sense, two forms of crosstalk are relevant.
The first, related crosstalk, is a signal bleeding over into
another signal path that is carrying a musically and
temporally related signal (e.g., between the left and
right of a stereo pair or between adjacent tracks of a
multitrack recorder). It happens quite often and is fortu-
nately not often subjectively obvious or embarrassing;
usually they’re playing the same song!
Crosstalk within multitrack recording systems is
usually little short of horrifying. As a result of the large
physical size of the console, ground paths are unavoid-
ably long and ground currents generate (and cross-inject
into other paths) crosstalk voltages across the resultant
ground impedances. Capacitance between intercon-
necting cabling, looms, modules, buses, indeed every-
thing, results in a reasonably suspect electrical overall
crosstalk performance. Naturally, the better the design
and construction, the better a console tends to be in this
respect. One typically gets what one pays for.
This was overshadowed and mitigated by analog
multitrack tape machine crosstalk between tracks—a
safe order of magnitude worse than even a horrid
console could be. These tape machines not only had the
same electrical problems as consoles but also had many
magnetic heads in very close proximity, all dealing with
a tape medium not notable for magnetic isolation
anyway. It was all tolerable and usable simply because
all the crosstalk was related and blended in unnoticeably.
25.14.4 Unrelated Crosstalk
Unrelated crosstalk is the clashing and cross-bleeding of
signals that have nothing whatsoever to do with each
other and are a mutual embarrassment.
In console monitoring a hostile signal (i.e. a delayed
replay B check of a master) can be screaming about in
uncomfortable proximity to the main stereo mix paths.
Broadcasters face this same problem all the time. All
their sources are hostile unless brought up on air.
This is unrelated crosstalk, where the bleeding signal
is totally dissimilar and irrelevant to the interfered
signal. Basically, if any unrelated crosstalk is audible
above system background noise, it will be noticed.
A fairly recent and insidious sort of unrelated cross-
talk comes in the forms of assorted chirps, buzzes, and
sizzles stemming from the relentless march of digits
into console design and operations. The Society of
Motion Picture and Television Engineers (SMPTE) time
codes and automation codes were bad enough, but
trying to get computer clock droning and vdu squeaks
out of the mixing buses and audio paths is not one of
life’s most enjoyable tasks.
Designing it out in the first place is the only way to
deal with computer noise:- Make sure all the logic grounds and analog grounds
interrelation makes sense or are tailed back sepa-
rately and never meet. - Scrutinize printed circuit layouts to make sure there
are no digital signals adjacent to or on the direct
opposite side of the board to any analog signal. - Intersperse lots of ground traces.
- Screen high-current high-speed digital signals.
- Try to allow only static digital control lines onto
analog boards—this means decoding digital buses
elsewhere other than on audio boards. - Ground-plane everywhere there is board space.
- Choose logic families—or at least interface
devices—that are low current and devoid of large
power-rail gulps. CMOS is just fine. - Decouple everything for all signals—decouple
digital for AF and analog for RF. - Work on your karma.
25.14.5 Quantifying Crosstalk“If you can hear it or measure it, it’s failed.” Such is the
empirical crosstalk test. A more formal test was origi-
nally the test for interchannel crosstalk (i.e., between
any channels in a console); it’s also used for any dissim-
ilar path crosstalk measurements. In short, it asks for
better than 60 dB of isolation of 6 kHz between the
paths, measured with a standard peak program meter
(PPM) with a CCIR 468 weighting filter in line. Since
this CCIR curve has 12 dB of gain at its crest (at
6 kHz), the specification is actually calling for better
than 72 dB of isolation at 6 kHz, which is neither easy
nor often realistic. Such a figure is occasionally not far
above system noise floors. Remember, it’s a peak
measurement; an rms measurement would be some
7–10 dB lower. Nobody said it was going to be easy.
Crosstalk’s a tough problem.25.14.6 MetersSome indication to the operator of the signal levels
running through the console and, most importantly, the
levels that are being sent to other places is necessary. In
Fig. 25-92 a pair of level meter feeds are taken from the