Consoles 907which are those generally most predominant and are so
controlling of the gain reduction.
Major differences between limiting and compres-
sion are in the nature of the side chains, the level detec-
tors, and in particular typically applied time constants.
Limiters almost invariably have peak detectors, such
that the peak of a waveform is detected and, time
constants allowing, protected from overload by the
limiter; this is fine for the protection mandate of
limiters. Compressors, on the other hand, tend to have
much more relaxed attack and release times, such that
they are less intense sounding than the typically frenetic
limiter settings. Similarly, since the concern with
compressors is less peak level than loudness to the ear,
which tends to gauge by overall signal energy or power
rather than peak values, the detectors are typically
average or power sensing. The slower time constants go
a long way toward this by essentially ignoring peaks
and responding more to an average over the time
imposed by the attack and release times. Deliberate
averaging detection (as opposed to the more or less
accidental) sounds much more even and unobtrusive
than peak detection; taking it a step further, power
detection by means of a root mean square (rms) level
detector is better yet. In reality, though, there’s little to
choose between average and rms detection, since
although they give significantly different answers under
test waveform circumstances, dancing around in the
heat of audio battle they are quite difficult to tell apart.
Occasionally either one or the other will be fooled by a
difficult piece of program material.
The threshold tends to extend farther down for a
range typically of 30 to +10 dBu, and the ratio adjusts
from 1:1 (straight) usually to infinity:1 or close (limit).
Some commercial units extend the ratio beyond infinity
to negative values; that is, if a signal progressively
exceeds the threshold it gets progressively further atten-
uated! Although on first glance it seems a bit pointless,
it does allow fairly nice sounding level control for a
compound signal; it permits looser (longer) attack times
than would be possible on an ordinary limiter, with the
resultant overshoot merely propelling the signal farther
downward away from possible overload. It is also good
for some pretty silly effects on individual instruments.
25.12.4 Expansion
An expander increases the dynamic range of its output
signal in relation to the input signal. Its ratio determines
how much: a 1:3 expander renders a 4 dB level shift in
input signal, which results in a 12 dB difference at the
output.
As mentioned under compression, true full-range
expansion is a rarity and is generally only found as a
complement to a compressor in a double-ended noise
reduction system. In these circumstances they are nearly
always of 1:2 ratio with an axis point (the level at which
the input signal is the same as the output signal level) of
around 0 dBu.
Practical expanders come with a threshold setting,
above which they leave the signal alone and below
which gain reduction sets in. Sounds a bit like a gate? A
gate can be emulated by an expansion with a ratio of
1:infinity; any signal below the threshold gets attenu-
ated away completely—the one exception is that
expanders usually don’t have a depth setting. The
purposes of expansion are very similar to those of a
gate, only generally it can sometimes do a better, less
noticeable, job. A relatively gentle expanding slope (say
1:2 or 1:3) can provide the same degree of noise reduc-
tion as a gate with less abrupt changes in gain; since the
signal is audible still (but quieter) and doesn’t have to
be resurrected with a start to normal level, fairly gentle
(slower than a gate) attack times do not have as notice-
able a softening effect on the required leading edge.
Expansion side-chain time constants are similar to
those for a gate, as is the threshold range. Ratio, as with
compressors, is usually 1:1 to 1:-infinity, although often
“classic” implementations have a fixed ratio of some-
thing close to 1:2. Expansion is used as subtle gating in
much the same way as compression is a gentler substi-
tution for hard limiting.25.12.5 Feed-Forward VCA-Style DynamicsThe feed-forward class of dynamics owes itself to the
development of VCAs and similar log/antilog
processing; it is exemplified by the classic dBx160
series. As far as consoles go, the mere existence of a
VCA in the channel for fader automation begs for this
style of processing to be incorporated. (VCAs are
further discussed under Consoles and Computers, later.)
Figure 25-84 shows such a processor in block diagram-
matic form.
Key to VCA dynamics is the inherent exponential
(logarithmic) control, which relies on reasonably simply
implemented basic transistor behavior (base voltage
versus current). Gain (or gain reduction) of a VCA is as
good as linear dB/v, which can lend to a deterministic
design approach (meaning one can pretty well predict
what the circuit will do within narrow limits, without a
servo loop to help). Simple log/antiloging lends itself to
another typical feature of VCA dynamics sections.