A typical full-featured gate will have the following controls:
threshold, attack, hold, decay and range. Additionally, there may be
high- and low-frequency controls. Less expensive gates may not include
one or more of these controls. There are usually LED indicators which
tell whether the gate is open or closed. Some gates use an LED bar-graph
display to tell you how much attenuation the gate provides.
I find it easiest to start when all of the gate’s controls are set
for "no gating." Start with the Threshold set to Minimum
(largest negative number!), and attack time at minimum, (shortest time),
hold time at maximum (longest), decay time at maximum (longest) and the
range to maximum (largest negative number). Set the low-frequency
control to minimum, and the high-frequency control to maximum.
As the drummer slowly bangs on something, raise the threshold. At some
point, the gate will close and no sound will pass through the gate. The
threshold control "looks" at (the envelope of) the input
signal, and if the input signal is below the threshold, the gate is
closed. The gate opens when the signal is above the threshold. So, the
trick is to find the point where the drum opens the gate, but the
ambient noise on the stage, or the next drum, doesn't. With a little
practice, you'll find it's easy to narrow that down.
The threshold control is calibrated in decibels. 0 dB refers to the
gate's nominal input level. This means if your gate has +4 dBm I/O, then
if you set the threshold to 0 dB, the gate will open when the box input
exceeds +4 dBm. If the threshold is set to -15 dB, then the gate opens
when the input is above -11 dBm. Likewise, if your box has -10 dB I/O,
then the 0 dB mark means -10 dB.
Next is the attack time. The attack time control sets the time
(usually in microseconds, or milliseconds) the gate takes to go from
closed (maximum attenuation, as set by the range control) to open (zero
attenuation). The gate doesn't "wait" the attack time before
snapping open; rather, it smoothly ramps the attenuation from max to
zero in the attack time. This is analogous to starting with the channel
fader on minimum, and fading up to unity, in microseconds.
If you're playing along at home, you may notice that if the threshold
is set so that it's barely below the drum input level, the gate will
"click" as it opens. You can mitigate that click with the
attack time control. Here's what happens: when the signal goes above
threshold, the gate is told to open. If the attack time is too fast, the
gate output wants to switch instantly between 0V and some non-zero value
-- maybe a couple of volts. Now, if you use the attack time to slow down
the attack, that voltage doesn't change "instantly," but
rather smoothly ramps up.
So, the trick is to set the attack time fast enough to capture the
drum's transient, but not so fast that the gate clicks.
The hold time control is obvious -- it's simply how long (in
milliseconds) the gate remains open once it's fully open. Too short a
hold time, and you clip the end of the drum's ring. Too long, and the
gate may not close, or you'll get excessive ring, or what have you. You
can also get weird clacking noises from the gate as it chatters -- too
short a hold time and you'll find the gate might try to open and close
and open and close quickly.
The decay time is the same idea as the attack time, except it
determines how quickly the attenuation increases once the signal goes
back below threshold. Too quick a decay time and you'll clip off the
drum sound tails. Too slow and it may not be closed before the next drum
hit.
The range control is what sets how "closed" the gate is.
When set to max attenuation (say, -80dB; negative dB gain is
attenuation!), when the gate is closed, there is 80 dB attenuation from
the input to the output. That's closed! Now, as to why you would want to
control the amount of attenuation: imagine putting a gate on, say,
background vocals. When the person isn't singing, the gate is closed and
the background noise disappears. When (s)he sings, the gate opens, and
the background noise is present. It's a bit odd sounding, so by setting
the range to, say, -10, the difference between open and closed isn't so
startling. It's as if you pulled the channel fader down to -10. If you
set the range to 0 dB, there is no gating action at all. Gates usually
don’t have a “positive” range, since they do not add gain.
Now, for the frequency controls. These controls let you tune the
frequency range the gate responds to. For example, say you're gating the
rack toms, and there's a loud crash cymbal right above them. With the
frequency controls "wide open" (low at min, high at max), the
gate may respond to the cymbals as well as the drum, which is not what
you want. So, you use the frequency controls to set the passband that
the gate responds to...so, for a rack tom, set the LF to something like
50 Hz, and the HF to something like 500 Hz (or whatever). Since the
majority of the cymbal's energy is not in that passband, the gate won't
respond to it. But the rack tom does have a lot of energy there, and the
gate opens when the tom is hit.
It is important to remember that the frequency controls do not affect
the tonality of the audio! Also, some gates have slightly-different
controls for setting the gate passband. It’s similar to a parametric
EQ – there is a center frequency knob and a Q control. The Q sets the
width of the band centered on the center frequency. Some gates may have
just the center-frequency control and a fixed Q.
Now, for some technical details: The gate is based on a device called
a Voltage-Controlled Amplifier (or VCA). This is a device whose gain
(and attenuation) are set by a Control Voltage. The audio signal through
the VCA is usually called the audio path.
The Control Voltage is derived by something called the sidechain,
which is parallel to the audio path. The gate’s input is split into
two outputs; one drives the audio path and the VCA, the other drives the
sidechain. The sidechain uses a circuit called a detector to
"look" at the audio and generate a DC voltage that corresponds
to the level of the audio. This voltage is modified by the threshold,
attack, hold, decay and range controls to come up with the proper
Control Voltage to set the VCA to do the right thing. You should note
that none of the gate’s controls affect the audio directly – they
affect the control voltage to the VCA, which affects the audio level. A
gate has no affect on the tonality of the audio.
Note that the frequency controls come before the detector. Therefore,
they affect what the detector responds to. Also, side-chain inserts come
before the detector.
The detector is probably the most important part of the gate
circuitry. A poor choice of detector can make a gate built around the
world’s best VCA chip a piece of junk!
A noise gate is really a special case of a device called a downward
expander. (Aside: an expander affects signals both above and below a
threshold; a downward expander only affects signals below a threshold.
That’s an important distinction.) Instead of having a range control,
an expander has a ratio control. Ratios are expressed as in : out, where
in is the input level in dB, and out is the output level in dB. Ratios
range from 1:1 (meaning no expansion) to 1:10 and higher (lots of
expansion).
When the signal is below threshold on an expander, it is attenuated
by an amount determined by the ratio control. So, for example, if you
set your downward expander ratio to 1:4, that means that for every 1 dB
below threshold, the VCA will attenuate the signal by 4 dB. A signal at
–4 dB, then, would be attenuated by 16 dB. A gate is nothing more than
a downward expander with a very high ratio, in much the same way that a
limiter is a compressor with a very high ratio.
Finally, regarding gate usage: Gates are a double-edged sword, and
they can be immensely frustrating devices if the drummer is either very
uneven with his playing, or if he's got an excellent sense of dynamics.
Too much gating and the good drummer's quieter parts get clipped off
(unless you're riding them, which you have to do). Not enough gating,
and the lousy tone of the bad drummer rings forever.