The hook
Two churches, same console, same speakers, same size room. In one, the system is dead silent between songs and effortlessly clean when the band plays. In the other, there's a faint hiss under everything, the vocal breaks up on the loud choruses, and the whole mix feels like it's fighting itself.
The difference isn't the gear and it isn't talent. It's about fifteen small decisions made before the service started: one knob per channel, set with intention instead of hope. This is gain staging. It's the least glamorous skill in audio and the one that quietly separates every clean system from every ugly one.
The lesson
Start with the picture. Every audio channel lives between a floor and a ceiling.
The floor is noise. Every electrical circuit generates a faint hiss all by itself. That's physics, not a defect. The floor is always there, always whispering.
The ceiling is clipping. From lesson 1.4: a digital system has an absolute maximum, 0 dBFS, and signal that hits it gets flattened into harsh garbage.
Your signal has to live between the two. Gain staging is the craft of placing it there: comfortably above the floor, safely below the ceiling, at every stage of the chain.
Both failure modes come from bad placement:
Too low, and hiss wins. Set the preamp gain timidly and the signal comes through weak. Somebody downstream (the fader, the amp, the powered speaker's knob) has to make up the loss, and whatever boosts the signal boosts the floor right along with it. That underlying hiss in church number two? Almost always weak gain early, big boosts late. The hiss was invited in one knob at a time.
Too high, and the ceiling wins. Set the gain hot and the loud moments (the chorus, the shout, the snare hit) slam into 0 dBFS and shatter. Worse, clipping at the preamp is permanent: every EQ and fader after it just reshapes the distortion. You cannot un-clip downstream.
So where exactly should the signal sit? Here's the number that runs the industry, and it deliberately looks conservative:
Average around −18 dBFS. Peaks below about −10 dBFS.
That feels low the first time you see it. The meter's barely halfway up! That's the point. The gap between −10 and 0 is headroom: reserved space for the moment the singer steps up or the drummer digs in on the congregation's favorite chorus. Live sources are wild; they will jump 10 dB when the room fills with energy. Headroom is how you absorb that jump gracefully instead of shattering on it. And at 24-bit (lesson 1.4), running "low" costs you nothing: the floor is still some 120 dB below your signal. The corridor is enormous. Use the middle of it.
Now the procedure. This is the same on every console ever made:
- Have them perform, for real. Not "check, check." The actual singing voice, at the actual intensity they'll use in the service. Musicians always play harder live. Ask for "loudest song, loudest section."
- Start with the fader down and the gain low.
- Bring the gain up slowly, watching the channel meter, until the average sits around −18 dBFS and the loud peaks touch −10 or so. Many digital boards mark this healthy zone in green/yellow — stay green, kiss yellow, never red.
- Then set the fader to place the source in the mix. From lesson 0.1: gain sets a clean signal once, the fader balances it all service.
- Leave the gain alone. If a channel needs to be louder in the mix later, that's a fader move. You touch gain again only if the source itself changed — new mic, different singer, guitarist swapped amps.
Do that fifteen times before rehearsal, and you've built church number one: silent floor, effortless peaks.
One more layer, because the word "staging" is plural on purpose: the preamp is the most important gain decision, but not the only one. The chain from lesson 2.1 has several places where level gets set: channel gain, fader, bus levels, the main output, the amp or powered-speaker knobs. The principle at every one of them is the same: each stage should hand the next stage a healthy level, so nobody downstream is rescuing or taming a mistake made upstream. The classic broken pattern is a chain wearing a disguise: main fader pulled way down because the amps are cranked, or amps whispering while every channel runs hot. It "works" until it hisses, distorts, or leaves you no room to move. A well-staged system has every knob doing its own small job near the middle of its range.
Go deeperoptional
The −18 dBFS convention is an inheritance, and a smart one. Analog consoles were designed around a nominal operating level of 0 VU = +4 dBu, with transformers and op-amp stages engineered to sound clean there and saturate progressively 15–20 dB above it. When digital arrived, alignment standards mapped that analog sweet spot to the digital scale: the EBU convention sets 0 VU ≡ −18 dBFS (SMPTE uses −20). So "average −18" literally means "run the digital console where the analog world ran its consoles for fifty years." Plugin and DSP algorithms modeled on analog gear are typically calibrated to behave best fed at that same internal level.
Why the floor exists at all: thermal (Johnson–Nyquist) noise. Any resistor at room temperature generates noise voltage from the random motion of charge carriers: about −128 dBu of noise EMF from a 150 Ω source impedance in a 20 kHz bandwidth. That's the physical limit; a preamp's EIN (equivalent input noise) spec tells you how close it gets, with good modern designs around −127 to −129 dBu. Practical consequence: the preamp's noise contribution is essentially fixed relative to its input, so the more of your total system gain you take at the preamp (up to the healthy target), the better the final signal-to-noise ratio. Later stages then need less gain, and their noise contributions land further below the signal. This is the same cascade logic (Friis) as in lesson 2.1's deep end.
Digital summing gives you one modern mercy: mix buses on current consoles run at 32-bit float or wide fixed-point, so an internal bus "over" usually isn't destroying anything until it reaches an output converter. The hard ceilings live at the A/D (your preamp gain decision) and the D/A (your output level decision). That's why preamp clipping is fatal and unfixable while hot internal buses are merely bad practice.
On meters: know what yours shows. Sample-peak meters can under-read true intersample peaks by a couple of dB; RMS/average meters respond slowly to transients, so a snare can clip a converter while an averaging meter reads −20. Digital console channel meters are typically peak-reading post-preamp, which is what you want for gain setting. And the analog world's VU ballistics (300 ms integration) explain why an old-school VU can sit at 0 while peaks fly 15 dB higher. The VU was always a loudness-ish meter, not a peak meter, and the headroom convention was baked in around exactly that behavior.
Remember
- Every channel lives in a corridor: noise floor below, 0 dBFS ceiling above. Gain staging places the signal comfortably in the middle.
- Target: average around −18 dBFS, peaks below −10. The gap to zero is headroom: the shock absorber for the loudest moment, which always comes.
- Hiss = gain too low early, boosted late. Distortion = gain too hot early. Both are placement errors, not gear failures.
- The procedure: real performance level → gain up to the target on the meter → then fader → then leave gain alone.
- Every stage hands the next a healthy level. No knob rescues another knob's mistake.
Your turn
- Re-gain your three most important channels with the full procedure, at real performance energy.
- Find the healthy zone on your console's meters — which colors correspond to −18 and −10, and where red begins.
- Hunt one disguise in your system: a main fader way down compensating for hot amps, or a powered speaker cranked to rescue a weak feed. Fix the pair together.
- Explain to another volunteer why the meter "only" reading halfway is correct, using the word headroom.