TDM Audio 32CL-2 v.2 - Manual (Page 5)

© 1996 TDM Design, Inc. Compressor/Limiter Owner’s Manual Page 5
compressor adjust its gain to achieve an exact 2:1 compression ratio? Well, the answer is that it
can’t. The way compressors work in the real world is with a threshold. The next term to familiar-
ize yourself with is threshold.
The threshold is your compressor’s way of letting you tell it what the softest part of the program
will be. With this knowledge, the compressor can accomplish its goal. Take the 2:1 example again.
A compressor with a 2:1 compression ratio reduces the gain by 1 dB for every 2 dB the input
signal is over the threshold. If the threshold is -60 dBm, and the input signal is 20 dBm, which is
80 dB above the threshold, gain reduction will be 40 dB so the resulting output signal is -20 dBm.
-20 dBm is 40 dB above the threshold so the dynamic range of the input is reduced by half. If you
were to set the output gain to be exactly ½ of your expected maximum gain reduction (in this case
¼ of your expected input dynamic range), you would achieve true 2:1 compression. In such a
setup, the loudest sound (at +20 dBm) would output 0 dBm, and the softest sound (at -60 dBm)
would output -40 dBm. From -40 to 0 is a 40 dB dynamic range compressed equally at either half
of the spectrum. It rarely really works this way, though.
One problem with the above scenario is that it raises the noise floor by 20 dB which is usually
unacceptable.
Another problem is that such a scheme usually doesn’t sound the way you would want it to.
Generally, there is a range of sounds near the bottom end of the dynamic range of the program
that you want to leave uncompressed in order to let the quietest parts of the program decay natu-
rally without a lot of “breathing.” Breathing describes the sound of the compressor bringing up
the gain as things get quieter. You can hear the program noise increase, and the program decays
unnaturally.
Another problem is that if the sound source is a microphone being reproduced through a loud-
speaker, low thresholds cause feedback problems. The problem is that the microphone gain is
much hotter when there is no sound than when there is. When a person speaking into such a
microphone pauses, the compressor does the equivalent of cranking the volume of the micro-
phone way up. This often results in feedback.
Low thresholds also cause unnatural sounding attacks. Suppose the threshold is at -60 dBm and
the compressor is set to a 2:1 compression ratio. If the nominal level is 0 dBm, the signal gain is
30 dB hotter when there is no signal than when the nominal level is present. When the signal first
hits, the initial output is really loud, and then you can hear the gain being reduced by the compres-
sor. This is because it takes time to reduce the gain. In an ideal compressor, this gain reduction
would happen instantaneously, but in the real world it never does.
The solution to these problems is to raise the threshold. In most setups, the compressor doesn’t
actually start compressing until the level is much higher than -60 dBm. This leaves a large portion
of the dynamic range uncompressed, eliminates breathing, reduces output noise, provides for
natural sounding attacks, and just generally works better. What it also means, though, is that the
ratio no longer really represents the actual compression of the dynamic range. If the threshold is
set at -20 dBm, and the loudest signal is +20 dBm, then only the portion of the range from -20
dBm to +20 dBm is compressed. For 2:1 compression, the gain reduction at +20 dB is 20 dB. If
the dynamic range of the input signal ranges from -50 dBm to +20 dBm, the range is compressed
from 70 to 50 dB. This is really only a 7:5 ratio, and compression only happens in the high 40 dB
of the 70 dB range. Although this fact might be mathematically offensive, it results in much better
sound quality.