Digital Audio Interfacing and Networking 1463One way is to use a sample rate converter” (SRC) on
each input to convert the incoming sample rate to the
internal sample rate of the digital audio device. Such a
SRC will add cost to the input, and will in some subtle
ways degrade the quality of the audio. The accuracy
will be better than with an analog interfacing, but the
digital audio will not be transferred with bit accuracy.
Of course, there are different degrees of perfection
available in SRCs at correspondingly different levels of
complexity and cost. Some SRCs will only handle
incoming digital audio that is at a precise and simple
numeric ratio to the internal sample rate. Others will
accept any incoming sample rate over a very wide range
and convert it to the internal sampling rate.
This second sort of SRC is very useful when you
must accept digital audio from multiple sources that
have no common reference, and convert them all to a
common internal sampling rate.
As implied above, the other way to handle inputs
from multiple digital audio sources is to lock all the
devices in the digital audio system to a single common
reference clock rate. In large systems this is the
preferred solution, and the Audio Engineering Society
has developed the AES11 Standard that explains in
detail how to properly implement such a system. Such a
system can have excellent jitter performance since each
device directly receives its sampling rate reference from
a common source. Interconnections between the digital
audio devices can be rearranged freely since we do not
have to be concerned about synchronization and jitter
changes as the signal flow is changed.
39.1.6 AES11
AES11 defines a digital audio reference signal (DARS)
that is merely an accurate AES3 signal used as the
common reference clock for a facility. The DARS may
contain audio signals, but is not required to do so.
There are three basic modes of operation defined in
AES11: use of a DARS, use of the embedded clock in
the AES3 signal, and use of a common master video
reference clock from which a DARS is derived. Use of a
DARS is considered normal studio practice. As
mentioned above cascading AES3 signals through
devices without a DARS can lead to increased jitter.
The only flaw in this scheme is that some digital
audio devices may not have provisions for accepting an
external sampling rate reference. As a result, in many
complex systems, while there may be a master sample
rate clock that most equipment is locked to, there often
is still a need for sample rate convertors to accept the
output of those devices that can’t lock to the master
clock. AES11 acknowledges this limitation.
AES11 specifies two grades of DARS, grade 1 and
grade 2. A DARS that has as its primary purpose studio
synchronization should be identified in byte 4 bits 0–1
of the AES3 channel status. More details are given on
this below.
A grade 1 DARS is the highest quality and is
intended for use in synchronizing either a multiroom
studio complex or a single room. It requires a long term
stability of r1 ppm. Devices producing a grade 1 DARS
are only expected to themselves lock to signals of grade
1 quality. Devices that are only expected to lock to
grade 1 signals are required to lock to signals over a
range of r2 ppm.
A grade 2 DARS is intended for use in synchro-
nizing only within a single room where the added
expense of a grade 1 solution can’t be justified. It
requires a long term stability of r10 ppm as specified in
AES5. Devices expected to lock to grade 2 signals are
required to lock to signals over a range of r50 ppm.
The above information is based on AES11-2003. It is
always advisable to obtain the latest revision of the
standard.39.2 AES3The Audio Engineering Society titled their AES3 Stan-
dard “Serial transmission format for two-channel
linearly represented digital audio data.” Let’s break that
title apart as our first step in examining the AES3
Standard.
This standard sends the data in serial form. In other
words it sends the information to be transmitted as a
sequence of bits down a single transmission path, as
opposed to sending each bit down a separate transmis-
sion path. Each bit of data making up a single sample of
the audio is sent in sequence starting with the least
significant bit on up to the most significant bit. The least
significant bit is the bit that defines the smallest change
in the audio level, while the most significant bit is the
one that defines the largest change in the audio level.
AES3 normally is used to transmit two channels of
audio data down a single transmission path. The data for
channel one of a given audio sample period is sent first,
followed by the data for channel two of the same
sample. This sequence is then repeated for the next
sample period.
Most professional digital audio today is linearly
represented digital audio data. This is also sometimes
called pulse code modulation, or PCM. In such a
scheme for numerically representing audio, each time