You are here

Gain Structure

Exploration By Paul White
Published October 1995

Do your recordings sound like they've been made through a white noise generator? Well, don't just blame that vintage synth or battered effects unit — it could be your gain structure that's at fault. Paul White offers some tips on getting it right...

From the dropping of a pin to the exploding of a tank shell, the nature of sound is such that it covers a vast dynamic range. Reproducing this range using an equivalent analogue electronic signal results in a signal level that might vary from just a few microvolts (millionths of a volt), up to 10 volts or so. In a typical studio, the signal from a microphone is amplified to bring it up to what we call line level (a vague term meaning closer to a volt than a millivolt), and is then passed through a whole chain of circuitry, where it is equalised, routed, mixed, and effected, before being recorded onto the final medium.

Noise Annoys

The main problem when processing analogue signals is that each and every piece of circuitry adds noise to the signal — there's no such thing as noise‑free circuitry. This noise is actually due to the random movement of electrons, and until we find some way around the limitations set by quantum mechanics, we're stuck with it.

Fortunately, a well‑designed circuit adds only a tiny amount of noise, and this noise is largely constant. It is therefore pretty obvious that if you feed a very low level signal through the circuit, the ratio of the noise to the wanted signal is going to be worse than if you feed a strong signal. As I've pointed out on many occasions, the way to minimise the noise contamination is to make sure that you're passing as high a signal level as possible through the circuit, but there is a limit to how hot your signal can be — if it's too high in level, it will cause the circuitry to clip, and you'll hear distortion (see the 'Red Light District' box for more on this). Most analogue circuits don't suddenly clip when the level gets too high — instead, the amount of distortion rises gradually as the last few dBs of headroom are used up, then hard clipping occurs.

Digital circuits have very similar limitations to analogue ones. If the the maximum numerical value the system can handle is exceeded, the signal will still clip. However, there's no safety margin, or area of progressive distortion, as there is with analogue — one moment the signal is perfect, but push it up another dB, and you're into clipping. For this reason, the nominal operating level for digital equipment is usually chosen around 12dB below the actual 0VU, or clipping point.

If too small a signal is fed into a digital system, it is represented by fewer bits, which, in practical terms, means that the signal suffers from quantisation distortion, and this sounds very much like noise. In other words, digital circuitry doesn't mean you don't have to worry about noise — you still have to feed it the right level.

Gain Structure

Just as you'd never dream of making a tape recording without setting up the input level properly, you should apply as much care to matching the signal levels when two or more pieces of audio processing equipment are connected together. Once noise is added to a signal, it's amplified by circuits further along the line in exactly the same way as the wanted signal, so there's no way to get rid of it.

The whole idea behind optimising gain structure is to ensure that every circuit in your studio is running at its optimum signal level. When it comes to the inside of your mixer, you don't have too much control over what goes on, but the designers will have done their best to ensure that the internal gain structure is correct. However, this carefully planned gain structure can be compromised if you, for example, plug something into a channel insert point that significantly reduces the level of the signal passing through it — a compressor with the output gain set too low is a good example. Now the signal fed back into the mixer is too low, so you have to restore the level by turning up a gain control somewhere further on in the signal path, and as you might expect, when you turn the level up, the level of the background noise rises too. The golden rule when using insert points is that your peak signal level should be the same, whether the external piece of gear is connected or not.

Start As You Mean To Go On...

Perhaps the most important place to get the gain structure right is at the start of the audio chain, especially if you're using microphones. Always use the PFL buttons on your mixer to help you set the Input Gain Trim control, so that the peak signal is just going into the red on the console PFL meters. It takes a few minutes to check all the mics individually, but it's something that really has to be done religiously if you're to stand any chance of making a clean recording.

The most vulnerable level‑matching stage is the mic amplifier, because it's here that the tiny signal from the microphone is brought up to line level before its journey through the rest of the mixer. If you take the time to optimise your mic levels, you're far more likely to end up with a professional‑sounding recording.

Unsound Effects

Another potential source of trouble is the external effects unit. If you have your console Effects Send controls turned nearly right down, and then turn up the input stage of your effects unit to compensate, you'll end up with more noise than if you had the console levels set higher, and the effects input level set lower. The best scenario where line level signals are concerned is to maintain a unity gain situation wherever possible — in other words, the signal stays at nominally the same level, instead of being constantly built up and then knocked down again, or vice versa. On most mixers, a send setting of around seven corresponds to unity gain, and the same is true of the Master Send control. If you make sure that the channel with the most effect on it has a send setting of around seven, then you won't go too far wrong. Of course, it helps if the unity gain setting is marked on the mixer controls, and some manufacturers actually do this for you.

If your effects unit has an Output Level control, this should also be set at around seven, or to the unity gain position, which means that the actual level of effect is determined by the Effects Return control on your desk. If you set the gain structure first, and then use the effects return to fine‑tune the amount of effect on the most heavily processed channel, you'll end up with the best possible signal‑to‑noise ratio.

Turn It Off And Throw It Out!

Finally, don't let anything redundant feed noise into your system. For example, if you have an effects unit patched in that isn't being used, make sure the appropriate mixer inputs or returns are turned right down, and similarly, if you're a MIDI user, don't forget to turn down any channels handling synths that aren't being utilised, as some of these put out far more background noise than is decent for a well‑designed piece of audio equipment.

Disconnecting unwanted items of equipment also applies to sections of your mixer, so if you have any unused channels, don't just mute them, make sure they're not routed anywhere either, as even a muted channel can add noise. The same applies to aux sends, and while most semi‑pro desks don't allow you to unroute your sends, you may have the option of routing them to an unused alternative mix buss, which has exactly the same effect. You may be surprised to know that a lot of the noise you blame your effects unit for adding comes from the aux send mix buss, and not from your effects unit at all. So, if you have an effect that you only need to add to one channel, consider feeding the effects unit from either the insert send point, or the direct channel output. You'll be surprised at how much cleaner it sounds.

The Red Light District: Distortion

Distortion is defined as any difference, other than in level, between what you put into a circuit and what you get out of it, and clipping distortion is about as severe as it gets. The secret of avoiding distortion is to find a happy medium (Doris Stokes on acid?), where the signal level is kept reasonably high, but where a little safety margin, or headroom, still remains to accommodate any unexpected signal peaks. If you visualise the VU meters on an analogue tape recorder, this whole concept makes more sense. The nominal operating level is where the signal peaks at around the 0VU mark (where the red area starts), and the safety headroom is how far you can push the level into the red before you hear distortion.