The term “gain structuring” refers to operating each stage of an audio chain at its optimum level to get the best signal-to-noise ratio (S/N) and the least distortion. It’s a balancing act, in a literal sense, and sometimes requires compromises, and, in live situations, some “best guesses.”
There are several reasons that it is important. Every device has a “noise floor.” This is a small amount of noise that is unavoidably added by all the components in that stage. Better quality equipment adds less noise than cheaper equipment, but it is still present. It is the “N” in a S/N (Signal to Noise) spec. The noise floor is measured as the number of dB below a reference level (+4 is pro-quality equipment). So, a device with an S/N ratio of 64dBm means that the “noise floor” is 60dB below the reference level of +4dbM. If you pass a signal through the device at 0dBm, it is 60dB above the noise floor. During quiet passages, this will be noticeable, but probably not obnoxious.
Now, let’s say that with this same device that has a noise floor of -60dBm, you pass a signal that is -30dBm. That means that your signal to noise is only 30dB. In other words, the noise is only 30dB below the music. This will be noticeable AND obnoxious. In a live setting, where the loudest passages of music are 110dB, the noise, when nothing else is going through the system, will be 80dB. That’s LOUD and quite unacceptable.
If you increase the gain of a stage 20dB, you increase the volume of that stage’s noise 20dB. You also amplify the noise from all the previous stages by 20dB. Conversely, if you run the signal ‘hot’ through all of the early stages, and then, since it’s too loud, you reduce the gain (attenuate) in a later stage, you attenuate the noise from that stage, and all earlier stages.
What are the potential pitfalls from running the signal very hot? Well, consider the dynamic range of the various instruments. Dynamic range is the range from the loudest to the softest sounds from an instrument (or a total mix). So, if you are setting the trim on the input channel for a bass, as an example, and set the input gain so the normal sustained note is at -3dB, as soon as the bassist picks louder, or “pops” the strings, he will generate a sound about 20dB louder than what you used to set the gain. This would result in a signal level of +17dBm (-3dB + 20dB). On a lot of boards, this is going to result in some pretty annoying distortion, clipping, and a very unmusical “compression.”
This is the “balancing act” I mentioned earlier. The trick is to run the signal hot enough so that you avoid amplifying the noise floor, while leaving enough headroom to avoid distortion and clipping. Each piece of gear reacts differently, and has different characteristics. Some have huge amounts of headroom, while others may start to produce very audible distortion well below 0 on the VU meters. It’s all in getting to know your equipment.