Gain Staging
& Headroom
The foundation every other mix decision sits on. When levels are right, compressors, saturators, and limiters all behave as designed. When they run too hot or too quiet, you end up chasing problems that have nothing to do with the plugin settings. This covers the −18 dBFS reference, headroom, unity gain, and how to set levels at every stage.
What gain staging actually is
The short version: set every track to average −18 dBFS before you touch any plugins, and keep levels roughly there through every processor in the chain.
Good gain staging is invisible. You notice its absence, not its presence. A compressor hammering because it was fed 6 dB too hot, a limiter pumping on the master bus, a mix that sounds congested no matter what you do to the EQ: these are usually gain staging problems, not plugin problems.
Processors have an operating range
Every plugin is calibrated for a specific input level. Too quiet and it's noisy. Too loud and it clips internally before you hear it.
Gain compounds through a chain
Ten plugins each adding 2 dB and you've arrived 20 dB hotter than you started. That gain creep is invisible until the master bus is clipping.
Headroom is a resource
The gap between your peaks and 0 dBFS gets spent through the chain. Compressors and limiters need it to work. Use it up early and there's nothing left.
Plugins depend on it
Saturators, compressors, and distortion plugins are calibrated to a specific operating level. Wrong input level and their character disappears or turns destructive.
The measurement systems: dBFS, dBu, dBVU
The suffix after "dB" tells you what it's measured relative to. Three scales matter here:
| Scale | What it is |
|---|---|
| dBFS | Decibels Full Scale. 0 dBFS is the digital ceiling — the maximum the system can represent. Everything else is a negative number below it. This is the scale your DAW uses. |
| dBu | +4 dBu is the nominal operating level for pro audio hardware. When plugin docs mention this, they mean −18 dBFS in your DAW. |
| dBVU / VU | A slow-averaging meter (300 ms integration time) that tracks perceived loudness rather than instantaneous peaks. 0 VU = −18 dBFS. When a plugin says "calibrated to 0 VU," it wants −18 dBFS average input. |
The key conversion: −18 dBFS = 0 VU = +4 dBu
Digital (dBFS) Analog (VU) Analog (dBu) Meaning ───────────────────────────────────────────────────────────────────── 0 dBFS +18 VU +22 dBu Digital ceiling −6 dBFS +12 VU +16 dBu Peak headroom −12 dBFS +6 VU +10 dBu Hot signal −18 dBFS 0 VU +4 dBu ← Nominal operating level −24 dBFS −6 VU −2 dBu Conservative level −30 dBFS −12 VU −8 dBu Low signal −inf dBFS −∞ VU −∞ dBu Silence
Interactive signal chain
Drag the input gain slider and watch how level moves through a typical chain. The dashed line marks −18 dBFS / 0 VU.
Drag the input gain slider to simulate a recording level. Watch how signal propagates through the chain and affects each stage.
Why −18 dBFS is the target
Headroom
Headroom is the gap between your peak level and 0 dBFS. Boosts spend it, cuts earn it back. The goal is to spend it intentionally rather than letting it drain away through accumulated small boosts across 30 plugins.
A master limiter arriving at −1 dBFS is clamping every transient. The same limiter at −6 dBFS catches brief peaks and leaves the rest of the dynamics alone.
| Headroom condition | What happens at the limiter |
|---|---|
| −1 dBFS peak into limiter | Limiter applies 1 dB of ceiling, barely active. No audible pumping. Final output is correct. |
| −3 dBFS peak into limiter | Limiter applies 3 dB on peaks: short, clean gain reduction on transients. Slight perceived loudness gain. |
| −6 dBFS peak into limiter | Standard pre-mastering headroom. Limiter works in 4–6 dB range; can be driven to −14 LUFS without pumping. |
| −10 dBFS peak into limiter | Either you're leaving too much headroom (you'll miss loudness targets), or you need to add input gain before the limiter. |
| 0 dBFS peak into limiter | Disaster. Limiter is already at the ceiling with zero range. Any subsequent transient clips the limiter itself. |
Unity gain and the null test
Unity gain: output equals input, net 0 dB change. It's the default you deviate from on purpose. Compressor makeup gain accidentally adding 4 dB you didn't notice? That pushes the next stage 4 dB hotter than intended, silently, across every plugin in the chain.
The null test
Flip polarity on a duplicate of the processed signal and sum it with the original. Perfect unity gain produces complete silence. Any signal remaining tells you the processor is changing the level.
How to run a null test for unity gain: 1. Duplicate your channel in the DAW 2. On the duplicate, add the processor (EQ, compressor, etc.) you want to test with all controls at "flat" / bypass positions 3. On the duplicate, invert polarity (180° phase flip) 4. Solo both channels and play back Result interpretation: → Complete silence = unity gain confirmed (processor is flat/bypassed) → Signal remains = processor is adding or removing level → Amount of signal = magnitude of gain difference To measure the exact gain delta: → Turn off polarity invert → Match levels between original and processed until null test reaches silence → The amount you had to adjust = the gain delta the processor was adding
| Processor | Unity gain behavior |
|---|---|
| EQ (flat settings) | Should be unity gain. If it's not, the plugin has a color mode or input trim active. |
| Compressor (bypassed) | Should be unity gain. Most compressors have a gain parameter that should sit at 0 dB when not compressing. |
| Compressor (active) | Output is lower than input by the compression amount minus makeup gain. Set makeup gain to compensate exactly, then evaluate subjectively. |
| Saturator | Saturation adds harmonics and usually slight gain. At 0 drive it should be close to unity. |
| Limiter | Output ceiling is fixed; input sets how hard it engages. Unity gain through a limiter means ceiling equals input peak, leaving 0 dB of range. |
Gain staging through the full chain
| Stage | Target level |
|---|---|
| Clip gain | −18 dBFS average, peaks to −6 dBFS. Apply this before any plugins. Use clip gain, not the fader. |
| Track fader | Near unity (0 dB). Don't push it up to compensate for a quiet recording — fix it at clip gain. |
| EQ | Input at −18 dBFS. Trim the output to match input after any broad boosts. |
| Compressor | Input at −18 dBFS. Makeup gain back to unity — check with bypass or null test. |
| Saturation | Input at −18 dBFS. Trim the output to stop any gain bleeding into the rest of the chain. |
| Group bus | −14 to −6 dBFS average. Some rise from summing is normal. Set bus compressor threshold based on this range. |
| Master bus pre-limiter | −6 to −3 dBFS peak. If you're hitting −1 dBFS here, pull group faders down — not the master fader. |
| Limiter output | −1 dBTP ceiling. True peak limiting prevents inter-sample peaks from surviving the codec. |
Gain creep
Ten plugins on a channel, each adding a small boost. The channel hits the bus 6.5 dB hotter than clip gain. You pull the fader down and move on, but every downstream plugin is still running too hot.
Gain creep example — a single channel with common plugins:
Clip gain (normalized to): −18 dBFS average
+ EQ boost (2.5 dB shelf): −15.5 dBFS average
+ Comp makeup (1.5 dB): −14.0 dBFS average
+ Tape sat output (1.0 dB): −13.0 dBFS average
+ Exciter trim (0.5 dB): −12.5 dBFS average
+ Bus send return (+1 dB): −11.5 dBFS average
──────────────────
Net gain added: +6.5 dB
Result: Channel is 6.5 dB hotter than intended
Every downstream plugin sees wrong level
Headroom shrinks by 6.5 dB on the master busFix: after dialing in each plugin, trim its output back to neutral. Treat every plugin as unity-gain unless you're deliberately changing the level.
Digital vs soft clipping
Digital clipping
A sample exceeds 0 dBFS and the DAW truncates it: peaks cut flat, rounded curve replaced by a square top.
Digital clipping: Normal sine wave: ∿∿∿∿∿∿ (clean, rounded peaks) Clipped sine wave: ⌐¬⌐¬⌐¬ (flat-topped — hard clipping) Harmonic content added by digital hard clipping: Even harmonics (2nd, 4th): Minimal Odd harmonics (3rd, 5th, 7th, 9th…): Strong — these are the harsh ones Character: Brittle, buzzy, aggressive — "broken" distortion
The odd harmonics digital clipping adds (3rd, 5th, 7th) are dissonant. One clipped sample is audible as a sharp click. Sustained clipping is brittle, buzzy, and unfixable, because the original waveform is gone.
Soft clipping
Saturation plugins use an S-curve transfer function instead of a hard ceiling, so the signal bends rather than breaks. They generate even-order harmonics (2nd, 4th, 6th) which blend into the fundamental rather than fighting it.
Soft clipping transfer function:
Hard digital: _____ ← abrupt flat-topping
__| |__
Soft clip: ~~~~~ ← gradual rounding (soft knee)
_/ _
Harmonics added by soft clipping:
Even harmonics (2nd, 4th, 6th): Strong — blend with the fundamental
Odd harmonics: Much weaker than digital clipping
Character: Warm, full — the sound you're chasing with saturation plugins| Property | Digital hard clip vs Soft saturation |
|---|---|
| Onset | Abrupt — full clipping at the ceiling | Gradual — progressive saturation above threshold |
| Transfer curve | Linear until ceiling, then flat-topped | S-curve: soft knee, increasing saturation with level |
| Dominant harmonics | Odd-order: 3rd, 5th, 7th — dissonant | Even-order: 2nd, 4th, 6th — consonant |
| Sound character | Harsh, brittle, buzzy | Warm, full, glued |
| Information loss | Yes — waveform peaks are destroyed | No — waveform is reshaped, not destroyed |
| Recoverable? | No | Mostly — saturation can be dialed back |
VU meters vs peak meters
Your DAW channel meters show sample peak, which is useful for catching clipping but useless for judging perceived loudness or whether your plugins are being fed correctly. That's what VU metering is for.
Peak metering
Instantaneous maximum values, one sample at a time. Use it to verify you're not clipping and to check headroom before the limiter. Not useful for setting plugin levels.
VU metering
RMS averaging with ~300 ms integration time. The reading tracks perceived loudness. Aim for −18 dBFS average. In VU terms:
Professional VU metering standard (digital equivalents):
VU reading dBFS (approx) Meaning
─────────────────────────────────────────────────────────────────
0 VU −18 dBFS Nominal operating level — sweet spot
+3 VU −15 dBFS Peaks — comfortable for most material
+6 VU −12 dBFS Hot peaks — acceptable on transients
+12 VU −6 dBFS Danger — only very brief transients
+18 VU 0 dBFS Digital ceiling — do not sustain here
Safe target on hardware VU meters:
Average: 0 VU (−18 dBFS)
Peaks: +3 to +6 VU (−15 to −12 dBFS)
Absolute: never above +12 VU (−6 dBFS) except brief transients| Meter type | Use case |
|---|---|
| Peak meter | Clipping detection, headroom check, true peak verification on limiter output. |
| VU / RMS meter | Setting plugin input levels, calibrating to −18 dBFS, loudness-matched A/B comparisons. |
| LUFS meter | Streaming platform compliance, final mastering loudness target. |
| Correlation meter | Mono compatibility check. Essential on the master bus. |
Gain staging cheat sheet
Target levels by stage — electronic music
Stage Average Peak Status ──────────────────────────────────────────────────────────────────────── Individual tracks −20 to −12 dBFS −6 dBFS Plugins in operating range Channel strip out −20 to −10 dBFS −6 dBFS Maintain post-plugin Group bus −14 to −6 dBFS −3 dBFS Bus compression range Master bus pre-lim −8 to −3 dBFS −3 dBFS Limiter headroom zone Limiter output −12 to −8 dBFS −1 dBTP Streaming delivery safe Reference point: −18 dBFS average = 0 VU = +4 dBu = nominal operating level
dB scale quick reference
| 0 dBFS | Digital ceiling — hard limit |
| −1 dBTP | True peak delivery ceiling |
| −3 dBFS | Hot — brief peaks OK here |
| −6 dBFS | Ideal master bus peak |
| −12 dBFS | +6 VU — drum transient peaks |
| −18 dBFS | 0 VU = +4 dBu — nominal |
| −24 dBFS | −6 VU — conservative |
| −30 dBFS | Low — noise becomes an issue |
The most common mistakes
| Recording at −6 dBFS avg | Saturation and compression plugins are always clipping |
| Not compensating EQ gain | Louder → sounds better → masking the EQ choice |
| Comp makeup too high | +2–4 dB silent gain per track × 20 tracks = disaster |
| Pushing master fader up | Bypasses master bus gain staging intent |
| No headroom before limiter | Limiter pumps, dynamics collapse |
| Mixing on peak meters only | VU/RMS shows where plugins actually operate |
| Ignoring inter-sample peaks | Sample peak ≠ true peak — use dBTP |
Clipping types: the distinction
| Digital hard clip | 0 dBFS overflow — flat-top, odd harmonics, harsh |
| Soft saturation | Gradual S-curve, even harmonics, musical warmth |
| Intentional clipper | Plugin-controlled soft clip — can be musical |
| Brickwall limiter | Lookahead gain reduction — no flat-topping |
| Inter-sample peak | Reconstructed peak above sample peak — use dBTP |
Meter types and their purpose
| Peak meter | Clipping detection, headroom check |
| VU / RMS meter | Perceived loudness, plugin calibration |
| LUFS integrated | Streaming delivery targets |
| LUFS short-term | Section-by-section loudness balance |
| True peak (dBTP) | ISP-safe delivery ceiling verification |
| Correlation meter | Mono compatibility, phase check |
Numbers to internalize
| −18 dBFS | Nominal operating level. Every plugin calibrated to this input behaves predictably. 0 VU. +4 dBu. |
| −6 dBFS | Master bus peak before limiting. Standard pre-mastering headroom for electronic music. Enough range for a limiter to hit −14 LUFS without pumping. |
| −1 dBTP | Universal true peak delivery ceiling. All streaming platforms require this. Not the same as −1 dBFS. |
| 18 dB | The gap between nominal operating level (−18 dBFS) and the digital ceiling (0 dBFS). This is not wasted space. It is headroom for transients and dynamics. |
| 300 ms | VU meter integration time. The ballistic that makes VU readings correspond to perceived loudness rather than instantaneous peaks. |
| 0 dB GR | Ideal compressor and EQ unity-gain state after makeup compensation. Deviations should be intentional. |