If terms such as gain structure, impedance matching, and headroom give you a headache, you're in good company. Most people would rather their gear work perfectly right out of the box than have to tweak it into compliance. Nevertheless, when it comes to setting up and operating a P.A. system, a working knowledge of gain structuring (and a few related concepts) will help you get the best possible performance from whatever equipment you use.
In short, gain structuring has to do with the relative levels of audio signals going into and out of two or more connected audio circuits. Audio gear has a range of input and output signal levels within which it sounds good. Going outside of that range results in problems such as hiss, distortion, reduced fidelity (especially when dealing with digital gear), and lowered power output. This article will provide you with a basic understanding of gain structure in a live sound system and will show you some ways to optimize it for each piece of gear in the signal chain. As an aid to understanding, see the diagram of a signal chain in Fig. 1. Some sound systems will be simpler than the one depicted in the diagram, and others will be more complex, but the basic principles apply to any configuration.
FIG. 1: In a typical P.A. setup''s signal chain, one of two input sources is connected to an input channel on a mixer, with a compressor patched in to the channel insert and an external effects unit patched in to the auxiliary loop. The channel signal is routed internally to an equalizer in the console''s master section, and the mixer''s master output bus feeds a power amp''s input.
All audio gear has a peak maximum signal level (above which the signal begins clipping) and what is referred to as its noise floor: the natural noise of the electronics when no input signal is present (see Fig. 2). The difference between the two extremes is called the dynamic range, which is expressed in decibels. For example, if the peak maximum signal level of a device is +24 dBu and the noise floor is -60 dBu, the device has a dynamic range of 84 dB. The difference between the noise floor and the nominal level at which gear operates (zero on its meter) is called the signal-to-noise ratio (S/N). Finally, the difference between the nominal level and the maximum peak level is referred to as headroom.
FIG. 2: Maximum peak level, nominal level, and noise floor level determine how much headroom you have as well as what your signal-to-noise ratio is.
Why is all that important, and what does it have to do with getting optimal performance from your audio gear? As a rule, you want to drive the inputs of a piece of gear at as high of a level as possible without inducing distortion. So if the nominal level is +4 dB and the peak maximum level is +24 dB, theoretically you have 20 dB of headroom to work with. That means you'll probably want to set the input level so that the loudest peaks fall a few decibels short of the maximum, say around +18 dB. (I say probably because practice doesn't always follow theory — use your ears.) Conversely, input signals that fall below the nominal level become increasingly noisy as they approach the noise floor.
Setting the optimal output level is even trickier, because output characteristics can vary wildly from one piece of gear to another. For example, some begin distorting at relatively low settings, and others sound best when they're wide open. That's one of the reasons it is important to become familiar with each piece of gear in the signal chain. If one or more pieces of digital gear are in the chain, additional considerations will apply (see the sidebar, “Digital Matters”).
That may seem plain enough, and in general it is, but a number of additional variables must be considered at each link in the chain, and only by recognizing and dealing with those variables can you get optimal performance from your system. Take a closer look.
When structuring gain relationships, you should always start at the beginning of the signal chain, which is, not surprisingly, the sound source. In the case of direct line feeds from sources like instrument amplifiers, electronic keyboards, and personal mixers, a sound engineer can do little more than request that the signals arrive at the mixer inputs at optimal levels. But the engineer does have control over at least two sources: condenser microphones with built-in pads and direct injection (DI) boxes with selectable output levels.
Most condenser microphones have a built-in pad (input attenuator) that reduces the signal between the capsule and the output electronics by 10 dB or so. Generally, you won't need to engage the pad unless the mic is used on an especially loud sound source. For example, if an AKG C 535 condenser mic is used on a loud vocalist, then its 14 dB pad should be engaged. If a condenser mic is used on acoustic strings or as an overhead on a drum kit, the pad can stay off. Leaving the pad engaged for soft sound sources raises the noise floor of the capsule to a point at which it can be noticeable during quiet passages, so use the pad only when necessary. The general rule of thumb is: if you hear something that sounds like clipping or limiting from a condenser mic itself, activate the mic pad.
Many DI boxes have selectable output levels. For instance, my DOD 265 Stagehand DI boxes have selectable 20 dB and 40 dB output pads, and my Whirlwind DI boxes have 20 dB pads. Because most consoles I use have pads on the input strips, I like to send as hot a signal as possible from a DI box without clipping the output of the box itself (something that usually happens only on active DI boxes). That lets me trim back the signal to something usable at the console input while keeping the signal as hot as possible for its trip through the signal snake to the console. Doing so helps attenuate any ground-loop problems that might exist on that line due to its interaction with, say, a stage amplifier.
Also, regarding ground loops: if a passive DI box has a metal XLR jack and the mic cable's shield isn't connected to the XLR shell properly, then it's impossible to break a ground loop using the ground-lift switch on the DI box. In that case, you'll need to replace the mic cable with one that has the shell properly floated, or use a short XLR-female-to-XLR-male adapter with the shield disconnected. You won't believe how much grief that can save you.
IN THE CHANNEL
Once you optimize the levels coming from the various sound sources, you are ready to connect them to individual channel-strip inputs on your mixer. Nearly every mixing console has a trim (or gain) control on each channel strip, and many consoles also include a pad switch, sometimes labeled mic/line. The two controls are used individually or in combination to make the level of the source signal coming into the console compatible with the input level of the channel preamp.
Mixer-channel pads generally reduce the input signal strength by a fixed amount, usually around 20 dB. The pad is placed ahead of any transformers or other electronics in the circuit and should be engaged only when the input signal is too hot to be comfortably handled by the channel preamp. The trim control is a continuously variable potentiometer that adjusts the channel preamp gain level. Channel preamps typically offer around 30 dB of gain boost, far more than most other gain stages in the signal chain, so be particularly careful when adjusting them. If the input level is set too high, the preamp will be driven into clipping, causing distortion; if it is set too low, excessive noise will result.
Most consoles with a solo function on the channel strip show the input level of a single channel on a meter when the channel is placed in Solo mode. To adjust the trim, zero all the controls on the channel strip and lower the fader completely. Put the channel into Prefader Solo mode (sometimes called PFL for prefader listen) and monitor it with headphones so you can evaluate the sound source for distortion or hum. Have the vocalist or instrumentalist sing, talk, or play his or her instrument while you watch the solo meter level; bring up the trim level until the meter approaches 0 VU on the loudest transients. If you hear distortion on the headphones or the trim control can't be turned down low enough to get the solo meter down to 0 VU, engage the pad.
In practice, you'll probably want to set the channel level to somewhere between -6 and -10 VU during sound check. Things tend to get louder during the actual show, and it's preferable to incur a little noise rather than clipping the input stages when the sounds get heavier onstage. Only after you have set the input level properly should you bring up the fader and add the signal source to the house mix.
Levels can also change during the course of the show — guitarists are notorious for hedging their bets by playing softly during sound check and then cranking up the sound when the crowd arrives. Have no fear, though: if the signal level starts creeping up into the hot zone, you'll probably notice the peak-overload LED on the channel strip blinking at you. Similarly, you may find that you need to pull the fader down really low to make the signal fit in the mix. In either case, adjust the trim back on the offending channel and readjust the level of the channel fader in the final mix. However, be aware that adjusting the trim during a live show will also affect the monitor sends from that channel, which can make the musicians onstage very unhappy. You may have to trim the input while turning up the monitor sends to counteract the reduction in signal strength. Still, that beats having a clipping channel sound bad the whole night.
INSERTS AND LOOPS
Once the source sound has passed the channel preamp stage, it is routed to the mixer's internal mixing buses, but it can make several stops along the way. Many mixers feature channel inserts that let you patch an outboard processor, usually a dynamics processor, into the signal path just after the preamp stage so that the entire audio signal must pass through it before reaching the EQ section and other internal circuitry.
Most channel inserts do not have send and receive level controls, so you will have to rely on the processor's input and output level controls (assuming there are any) to set the gain at that point in the signal chain. When using a compressor, the idea is to adjust the compressor's output (or makeup) gain to compensate for any gain reduction. For example, if you read 10 dB of gain reduction on the compressor's gain reduction meter, crank the output level up 10 dB.
Similarly, if you want to patch in an external processor without sending the entire signal through it, or if you want to be able to route signals to it from more than one input channel, the usual method is to connect the processor to an effects or auxiliary bus. The signal from the effects or aux bus send is routed to the external processor's input, and the signal from the external processor's output is routed to the effects or aux return (or another mixer channel input). Unlike channel inserts, effects and aux sends and returns nearly always have level controls. Try setting the effects or aux send and receive levels in the mixer's master section to their halfway points, and then slowly turn up the processor's input level control until you get a consistently robust level. If the processor has a mix control, set it for 100 percent effect.
Something else to consider when using external audio processors is their operating level. The insert points and effects buses on large professional mixers generally operate at a +4 dB level, whereas those on semipro and consumer-grade mixers generally operate at -10 dB. Fortunately, many outboard processors can be switched between the levels.
If the processor's input gain control must be set very low to prevent clipping the meter, you're probably asking its -10 input to handle a +4 signal from the console, which is not nice to do. In that case, trim back the input of the console strip itself until you can get the processor's input control somewhere up around 50 percent. Conversely, setting the processor's input to +4 dB for a console with a -10 dB level will result in extra noise. It's important to think it all out and listen during sound check.
Finally, outboard processors can behave in dramatically different ways, so you need to understand each one. For example, the overload LED on one processor might flash when the input signal reaches -6 dB, whereas on another it might not flash until the unit has been driven into distortion. Or you might get a perfectly clean signal when cranking the output level of one processor to maximum and find that another one gets increasingly noisy past the halfway point.
ON THE BUS
The internal bus structure of a mixing console is subject to headroom and S/N considerations as well. Whereas some consoles like to have their mixing buses driven hard, others' buses can be clipped quite easily. A good example of an inexpensive live console that needed the buses driven hard was my old Peavey console from 20 years ago. There was a lot of noise in the mixing buses, but by running the output faders down around 2 or 3 (on a scale of 1 to 10) and driving the input stages a little hotter, it was possible to get a decent S/N at the outputs.
On the other hand, I had a more recent vintage Alesis 16-channel live console that didn't have extra headroom in the mix bus but was very quiet. In that case, I ran the output faders around 8 or 9 and then trimmed back the inputs until the output was at the right level.
The easiest way to determine the correct gain-staging approach is to plug in a dynamically consistent signal source, such as a drum machine, and listen for any crunching or distortion at the console output with headphones. If there's a lot of noise on the outputs with the faders up and no input signal present, bring the fader down until it's manageable.
If you hear distortion on the console outputs even when the meters read below 0 VU and the output faders are below halfway on the console, the internal mixing buses are clipping. In that case, bring the input faders down and the output faders back up. High-end consoles have extremely quiet buses and a lot of headroom, so you won't run into that sort of problem with them. But many inexpensive consoles can be tweaked in the way I described to sound better than you might imagine. If you want to go further, you can use an oscilloscope and a signal generator to actually see clipping in the various stages and adjust the levels accordingly.
HIT 'EM HARD
You can also tweak the gain structure between the equalizer and the amplifier to improve the S/N. For example, if you have sufficient gain from the equalizer output, you can raise its level by 10 dB and trim the input on the amp by the same amount to attenuate any hum or ground-loop problems between the console and amplifiers. That can really help in a quiet mixing situation.
Proper grounding, balanced inputs, and shielded cables should, in theory, allow for an ultraquiet connection between the console equalizer and the amplifiers. However, that's rarely the case in the real world. I'm always tweaking things one way or another to get the outputs as hot as possible without clipping and then turning down the inputs on the next stages.
Nearly any engineer can properly operate a really expensive console with plenty of headroom and low noise, but it takes an engineer with chops to make a cheap, unforgiving board sound great. I have observed many guest engineers working with the same equipment get results ranging from fabulous to mediocre or worse, depending on how they ran the levels. So don't feel put down as an engineer when you are given some inexpensive gear and asked to make it sound great. To me, making an inexpensive system sound like a million bucks is like the ultimate challenge to spin straw into gold.
Getting a P.A. to sound its best takes more than a great set of mixing ears for a particular music style. It requires understanding how each piece of gear in the signal chain works and exploiting its potential to the max while working around any weak points. Once you reach that level, you are truly one with the P.A. system; you can make it do anything you want.
Mike Sokol is a live-sound and recording engineer, musician, and computer integrator with 30 years' experience on both sides of the microphone. He lives in western Maryland with his wife, four boys, and three cats.
An article by Glenn While at the Florida State University School of Music defines technical audio terms related to gain structuring.
The Study Hall section of ProSoundWeb.com contains technical articles on clipping, audio math, gain structuring, and related topics.
A primer on signal-to-noise ratios has automated calculators.
Digital gear's operating characteristics are different from analog gear's, and those differences need to be taken into account when structuring gain, particularly when digital and analog devices are interfaced in a signal chain. For example, the inputs on analog gear can typically be driven to +2 or +3 dB before any detectable distortion is introduced, whereas with digital gear, as soon as you pass 0 dB, you're likely to get nasty digital distortion of the most unforgiving kind.
At the same time, if you are using, say, a 16-bit digital reverb unit, one bit of its word length will be used for each 6 dB of audio. So if you keep the signal to the input so low that you never get as high as -12 decibels full scale (dBFS) on the input meter, you're working with only 14 bits. If you run the input in the -24 dBFS range, you're asking the processor in the reverb to work with 12-bit audio, which can sound quite grungy. So make sure that the input level is in the -10 dBFS range most of the time and that it never quite peaks out.
Another difference between digital and analog gear is that with digital gear you can usually crank the output level all the way up without getting any distortion or unwanted artifacts (though some engineers go only to 90 percent, just to be safe). With a robust input level and the output level set to full, you probably won't need to turn the mixer's effects return level up very high to add an appropriate amount of reverb into the mix. If you were to go the other way around and just barely drive the reverb's input while boosting the return to get enough reverb level, you'd most likely hear lots of quantization artifacts and other noise as a result.
These considerations apply when interfacing any digital and analog gear, so think everything through as you analyze your audio signal chain.