Setting the right level at every stage of the chain, from preamp to master bus, is one of the most rewarding and most misunderstood skills in the digital studio. Here’s how to think in dBFS, protect your headroom and build a clean gain structure, with no myths and no magic numbers.
Gain staging is the management of signal level at every step of the audio chain: tracking, conversion, the input to your DAW, each plugin, each bus, all the way to the master. Done well, you don’t hear it — you hear a clear, dynamic mix that’s easy to process. Done badly, it produces fatiguing mixes, plugins that don’t behave as expected and, at worst, distortion. The topic looks dry; it accounts for half the quality of a mix.
Two level worlds: analog dBu vs digital dBFS
The first source of confusion is that we juggle two scales. In the analog world we think in dBu (or dBV), an open-ended scale: a console’s 0 VU typically sits at +4 dBu, and above it there’s still headroom before the electronics really clip — often 18 to 24 dB higher. Saturation there is gradual and musical.
Digital thinks in dBFS (decibels Full Scale). Here the scale is closed: 0 dBFS is the absolute ceiling, the point past which the converter has no codes left. Every level is expressed as a negative value (-6, -18, -60 dBFS…). Going past 0 dBFS doesn’t give you soft saturation but digital clipping: a hard, ugly flat-topping of the waveform. That’s the fundamental difference: in analog, the red is a zone; in digital, it’s a wall.
The alignment point: why aim for -18 dBFS
How do you tie the two scales together? Through calibration. Studios traditionally align their analog 0 VU (+4 dBu) to a digital reference level, most often -18 dBFS (European EBU practice) or -20 dBFS (US SMPTE). In other words, when your average signal hovers around -18 dBFS you’re at the digital equivalent of analog nominal level, with roughly 18 dB of headroom above before clipping.
That number isn’t arbitrary. Most plugins that model analog gear — preamps, tube EQs, compressors, saturators — were calibrated to sound their best when hit at around -18 dBFS average. Feed them a signal riding at -3 dBFS and you’re constantly pushing them into a saturation zone they weren’t set up for, then wondering why they sound harsh. Aiming for an average (RMS) around -18 dBFS per track, with peaks up to -10/-6 dBFS, gives the whole processing chain the margin it was designed for.
Headroom: the margin that saves you
Headroom is simply the gap, in decibels, between your working level and the 0 dBFS ceiling. It’s a safety cushion. A mix where every track flirts with -1 dBFS has none left: one added bass move, one EQ resonance, one in-phase sum of two tracks, and the bus overflows. A mix built around -18 dBFS keeps a dozen decibels of room. You can breathe, add processing without fear of the red, and reach mastering with a clean signal rather than an already-crushed file.
Time to kill a stubborn myth: “record as hot as possible to maximize resolution.” That was vaguely relevant in 16-bit; it no longer is. At 24-bit the noise floor is so low (theoretically around -144 dBFS) that recording a vocal peaking at -12 dBFS costs you no audible quality. What you gain is precious margin and protection against any unexpected peak. Record comfortably low and sleep easy.
Where clipping actually happens
A crucial point that’s often misunderstood: inside a modern DAW, the mix engine runs in 32-bit floating point. In practice an internal bus reading +6 dBFS is not clipped: as long as you bring the level back down before the output (or before a fixed-point plugin), the information is fully preserved. Internal float forgives temporary overshoots.
The two places where clipping becomes real and destructive are: the converter (both at recording and at the card’s physical output, in fixed point) and the input of any fixed-point plugin or any stage that deliberately clips. That’s why the discipline is less about “never seeing red on a bus” and more about controlling the level entering your converters and your color plugins. The master output meter, for its part, should never touch 0 dBFS.
The method, stage by stage
- At tracking: set preamp gain for peaks around -12 to -6 dBFS. Neither timid (needless audible noise on quiet instruments) nor greedy (risk of a peak clipping the converter).
- At the mix input: before processing, bring each track to an average around -18 dBFS with an input trim, not the fader. Your analog-modeled plugins will behave as intended.
- On each plugin: watch the level going in and coming out. A compressor adding 6 dB of make-up should be compensated downstream so it doesn’t artificially inflate the next stage and skew your A/B comparison.
- On buses and master: keep margin. A saturating subgroup bus colors everything through it; a master touching 0 dBFS before the limiter even kicks in leaves you no mastering options.
The inter-sample peak (true peak) trap
One last pitfall, invisible on a classic meter: inter-sample peaks. Between two digital samples, the converter’s analog reconstruction can reach a level higher than the highest displayed sample value. A file capping at -0.1 dBFS on a sample meter can produce a true peak of +0.3 dB on playback, clipping the converter or a downstream MP3/AAC encoder. That’s why you leave true-peak margin (typically -1 dBTP) and use a true-peak meter at the master — a topic tied directly to loudness and true peak in broadcast and streaming.
The takeaway
Gain staging is neither superstition nor an artistic constraint: it’s signal hygiene. Think in dBFS, set your tracks around -18 dBFS average, protect your headroom, and know exactly where clipping becomes irreversible (converters, fixed-point plugins) and where it doesn’t (internal float buses). Everything else — color, dynamics, character — is then worked in good conditions. I’ve heard countless mixes that “sound bad for no obvious reason”: nine times out of ten it’s a sloppy gain structure upstream, not a bad plugin downstream.
To go further: our guide to audio compression (threshold, ratio, attack, release), whose settings depend directly on input level; the Universal Audio Volt 876 interface, where the gain chain begins; and the harmonic saturation plugins that behave best with proper input staging.
