986 Chapter 25
25.25.1 Moving Audio—small-scale
25.25.1.1 AES-3
Stereo pairs (or pairs of monos) have been catered for by
the venerable AES-3 standard; this is a Manchester
encoded stream of two (up to) 32 bit audio words, some
informational tags for link status and format, and a
number of user bits that may be used for anything from
turning stuff on and off remotely to serially carrying
metadata (program-specific information) or more
complex real-time control data. It was designed to be
robust, simple, and as usable as possible in the predomi-
nantly analog world into which it was born, even down
to using the familiar 3 pin XLR connectors in its usual
implementation. With minor updates, mostly concerning
connection variants and data rate (it now handles the
once-unthinkable 96 kHz with ease), it still serves well.
It is a very close cousin (indeed the underpinnings are
chemically indistinguishable, and use the same chip
sets) to the domestic S/Pdif (Sony/Philips Digital Inter-
Face). The audio is treated identically, but the format
and informational tags differ. It is common for
AES-3/-S/PDIF receivers to be set up to strip these off
so as to allow universal connection, but obviously this is
at the expense of any metadata that may accompany the
audio stream, and, if this is of the least concern, any
digital rights mismanagement flags.
A performance downside to AES-3, particularly with
early implementations, was recovered clock jitter. Best
performance is achieved by reclocking at the receive end,
either by SRC (Sample Rate Conversion) or the use of
very good flywheel phase-lock loops to reestablish solid,
quiet clocking. If the facility is homogenous with every-
thing running off a master clock this is less important;
“bits is bits” and as long as they arrive within the same
framing period (e.g., 20.8μs at 48 kHz) and sample clock
period, and any D/A is done with the same pristine clocks
as any A/D, transmission jitter is irrelevant.
25.25.1.2 AES-42
As will been seen in the later mention of USB micro-
phones, there is a drive to push digital as close to the
source as possible, in that case for simplicity’s sake, in
proaudio for performance. The concept of putting mic
preamplifier, A/D converter, and processing inside the
microphone itself is at one and the same time seductive
and puzzling. The idea of simply taking a digital stream
(possibly in AES-3 format) straight from a microphone
into a digital system holds strong sway; reflection
shows that this—in any meaningful system—means
either the addition of a plethora of hitherto unknown
knobs and switches on the microphone itself or the
means of remote-controlling all those functions and
takes the shine off the idea somewhat, particularly to
those to whom a microphone is something one simply
plugs in and uses.
As has been made clear, there is little that binds a
particular function to a particular physical location or
piece of system hardware or software. Given that, some
mouse-and-screen GUI widgets to control the micro-
phone parameters, or indeed a physical set of hardware
knobs and switches to do the same, don’t care whether
the target is in the same box, another processor, or even
on the same continent. That, in this instance, the target
is on the top of a shiny microphone stand in the studio is
irrelevant. So, not only is a means of getting digital
audio from the microphone necessary, but means of
getting the control parameters or coefficients up to the
microphone, as well as a synchronizing reference clock
so that the microphone’s pristine audio doesn’t have to
suffer the immediate indignity of a sample-rate conver-
sion to match the rest of the system. And, of course, a
means of powering all this.
And so was born AES-42, in an effort to standardize
all this before multiple incompatible approaches dissi-
pated the concept’s appeal. Fig. 25-153 shows in outline
form its scope.
Many hitherto console functions have found their
way into microphone control via AES-42. Although the
scheme is not limited to these, the Neumann
TLM-103-D digital microphone, for example, allows
gain, microphone pattern, absolute phase, high-pass
filter, an in-built compressor/limiter/de-esser, and a
peak limiter’s parameters to be controlled. It’s easy to
see where that’s headed; no need for console channels
as we’ve known them.
The normal connectorization is via the old familiar
XLR, although the XLD is suggested for circumstances
where confusion with other XLR-using systems could
potentially result in damage. As would be expected,
signal formatting owing much to the familiar AES-3 is
used to retrieve the audio, which ordinarily comes
differentially down a shielded pair; user bits in the data
stream relay fixed data such as the microphone’s manu-
facturer, model number, and available controls; vari-
able data such as instantaneous parameter value are also
available by this means. Now the fun begins—power is
sent phantom style (common mode and with reference
to the shield) back up the line; instead of its merely
being regulated down to power the microphone and its
electronics, it is also modulated with control data and a
synchronizing word clock, which are filtered off and