9 - 16
Traditionally audio limiters have two time constants, an attack, the time is takes the
limiter to respond to a signal above the threshold and a decay or release which is the
time is takes to respond to a drop in level. In a traditional audio limiter the attack time is
usually set to somewhere in the region of a few milliseconds and the decay time
considerably longer at somewhere in the hundreds of milliseconds. This is not the most
optimum solution because transients that last only a few milliseconds will reduce the
level of the waveform for hundreds of milliseconds, reducing loudness and creating
audible pumping effects.
The solution is multiple time constants where one set of time constants can be set to
handle the fast peaks and another to handle the average level of limiting. Fast transients
will release in a faster less noticeable way and won't punch holes in the sound in a way
that single time constant limiters can. The secondary slower time constant circuit will not
have much effect on the audio waveform when hit with a transient because the higher
attack time, generally in the hundreds of milliseconds will not allow a build-up of energy.
In the case of a sustained envelope of audio above the threshold the multiple time
constant will attack as normal with the peak time constant but the sustained energy will
also charge the secondary slower circuit.
When the audio energy falls away and the circuit goes into release the peak decay will
dominate until it reaches a point where it hands over to the slower secondary time
constant for a slower rate of decay. The illustrations show this to good effect, where
transients have a fast release but multiple or sustained transients build up energy in the
secondary circuit which acts as a platform for the peak to release to. The secondary
circuit's platform can be thought of as the average level of limiting. Having this fast peak
responding circuit ride on top of the average circuit creates many advantages, limiter
transparency, less chance of pumping and greater loudness. By setting the time
constants appropriately we can have the multiple time constant based detectors work
as peak handling, average handling or the optimum setting of a balance of the two.
The peak attack time should be set to the desired attack time required from that limiter.
The range is 1-10 which corresponds to 1 to 00mS on an exponential scale. The peak
decay time should be set to the desired peak decay time required for transients. The
range is 1-10 which corresponds to a decay time of 10 to 1000mS.
The average attack time is perhaps the most important control in the dual time constant
detector as it sets the balance between peak and average energy in the detector. With
smaller numbers more energy is transferred into the average circuit and a higher
platform level is created so more time will be spent releasing at the slower average rate.
Higher numbers offer slower attack times for the averaging part of the detector and this
