---
Title: Levels – AudioLab
---

# Levels: dBFS and the generator

What the levels in dBFS mean, what an interface can measure, and which level the generator plays in each task. In short: **volts for the device, dBFS for the interface** – [see below](#volts-for-the-device-dbfs-for-the-interface).

- **dBFS** is the level relative to the converter's full scale (0 dBFS = maximum). Without calibration AudioLab does not know the voltage.
- Input overload is reported in the status line (*Input clipping detected*); reduce the level.
- Do not move the interface's gain knobs after calibration. With the outputs and the reference input calibrated, AudioLab notices: the reference reads more than 1 dB away from what the calibration expects, and the status line says *Reference … reads +3.2 dB against its calibration · was an input gain or the monitor volume knob moved? Calibrate again*. The calibration is taken at 1 kHz, so this is checked only for tones from 100 Hz to 10 kHz: an M4's front inputs fall 2.2 dB by 20 Hz of their own. The output inputs (after the device) cannot be watched this way – their level depends on the device.

## What dBFS means, and what can be measured

**0 dBFS is the strongest sine the converter records without clipping**; every −20 dB is ten times less voltage. How many volts 0 dBFS is depends on the interface's analogue stage and is found by calibration; a typical USB interface has about 5–8 V on its line inputs.

| dBFS | Voltage at 0 dBFS = 7.5 V | Meaning |
| --- | --- | --- |
| 0 | 7.5 V | full scale; above it the input clips |
| −0.1 | 7.4 V | AudioLab reports *CLIPPING* |
| −6 | 3.8 V | highest generator level in AudioLab |
| −10 to −30 | 2.4 V – 240 mV | best range for distortion measurements |
| −60 | 7.5 mV | an MM cartridge (about 5 mV) |
| −80 | 750 µV | an MC cartridge; reference after a −60 dB phono attenuator |
| −100 | 75 µV | below this a tone is not evaluated (*No test signal*) |
| about −110 to −115 | 10–25 µV | noise of a good interface over 20 Hz – 20 kHz |
| −140 to −155 | 1 µV and less | noise in one bin of a long FFT – the weakest tone visible in the spectrum |

**Two different floors.** The noise over the whole band (about −110 to −115 dBFS) sets **THD+N and S/N**, which add up everything in the band. The noise in one FFT bin is tens of dB lower, because a bin of a long FFT is about 1 Hz wide: that is why the spectrum's floor sits so low, and why a single tone (a harmonic, crosstalk, an IMD product) can be measured far below the total noise. A longer FFT lowers the floor per bin, not the total noise: THD+N stays the same, harmonics get cleaner.

- **THD+N** limited by noise is about *noise − tone level*: with −114 dBFS of noise, a tone at −20 dBFS gives at best −94 dB, at −40 dBFS −74 dB, at −60 dBFS −54 dB.
- **THD** of a good interface is around −100 to −110 dB; for a weak tone (below about −25 dBFS) the THD figure shows only noise.
- **S/N** is limited by the interface's dynamic range (a good USB interface: 110–120 dB A-weighted); a quieter device gets *interface limit*.
- **Single tones** (crosstalk, IMD, hum) are read in their bins, down to about −140 dBFS.

The noise comes from resistor and amplifier noise and from the power supply (a laptop charger can add several dB – measure on battery, a desktop computer through a [USB isolator](phono.html#a-desktop-computer-a-usb-isolator)); bits matter mainly for files (16 bits ≈ −96 dBFS, 24 bits ≈ −144 dBFS, below any real interface's noise). In practice keep the signal between **−6 and −30 dBFS**.

## Volts for the device, dBFS for the interface

A device does not know dBFS: it takes volts. dBFS is the scale of your interface, and how many volts it means depends on the interface – −20 dBFS is 245 mV on an interface with 2.45 V at full scale, but 100 mV on one with 1 V. So think of the level in **volts at the device's input**, and use dBFS to keep the interface in its good range.

### Typical levels of audio devices

Typical figures, for a first setting; the device's data sheet wins where it gives one. "Nominal" is the level a device works at, "maximum" what it gives at full scale or full volume.

| Device | Its input takes | Its output gives | Measure with |
| --- | --- | --- | --- |
| **CD player** | – (a test disc) | **2 V** at 0 dBFS on the disc; balanced outputs often 4 V | the disc's −20 dBFS tracks: 0.2 V |
| **Streamer, media player, DAC** | digital: the level is in **its own dBFS** | **2 V** at 0 dBFS, some 1–1.5 V; balanced often 4 V | a file at −20 dBFS: 0.2 V; −1 dBFS for its maximum |
| **Tuner** (FM, DAB, internet radio) | – (the broadcast or a stream) | about **0.3–1 V** at full modulation | a test tone station or file; read its level first |
| **Cassette deck** | line input about **0.1–0.5 V** brings its meters to 0 dB (record level at its usual setting) | about **0.3–0.6 V** when its meters show 0 dB – the deck's reference level (Dolby level, 200 nWb/m on the tape), not its maximum | a tape at Dolby level as the reference; THD and response about 20 dB below it; the maximum above 0 dB |
| **Reel-to-reel**, home | line input about **0.1–1 V** brings the meters to the reference level | about **0.5–1 V** at the reference level – the level of the machine's test tape (often 250–320 nWb/m), not its maximum | the test tape's reference tones, then lower |
| **Reel-to-reel**, studio | **+4 dBu = 1.23 V** brings it to the reference level (some +6 dBu) | **1.23 V** (+4 dBu) at the reference level, with headroom above it | +4 dBu, then lower |
| **MM cartridge** | – | **2.5–7 mV**, typically **5 mV** at 1 kHz, 5 cm/s | the test record's reference tracks |
| **MC cartridge** | – | **0.2–0.5 mV** (low output); high-output MC 1.5–2.5 mV | as MM |
| **Phono stage, MM** | **5 mV** at 1 kHz; overload often 50–100 mV | about **0.3–0.5 V** for 5 mV in (35–40 dB of gain) | **5 mV** through a −40 dB attenuator |
| **Phono stage, MC** | **0.5 mV** at 1 kHz | about 0.3–0.5 V (55–65 dB of gain) | **0.5 mV** through a −60 dB attenuator |
| **Preamplifier, line stage** | **0.3–2 V** (consumer −10 dBV = 316 mV) | about **1 V** nominal, several volts maximum | 316 mV – 1 V |
| **Integrated amplifier**, line input | **0.15–0.5 V** for its rated power | its power: see below | low, rising: a few tens of mV first |
| **Power amplifier** | its sensitivity, often **0.775–1.5 V** for full power | **2.83 V** = 1 W into 8 Ω; **28 V** = 100 W | through an attenuator to the interface; start low |
| **Headphone amplifier** | line, **0.3–2 V** | a few hundred mV to several volts | about 0.25 V at gain 1 ([step by step](step-headphone-amp.html)) |
| **Studio gear** | **+4 dBu = 1.23 V** nominal, up to +20–24 dBu (7.75–12.3 V) | +4 dBu nominal | 1.23 V |
| **USB card's input** (External ADC) | line about **1–2 V** at its full scale; a phono input as above | digital: **its own dBFS** | 245 mV to find its full scale ([step by step](step-usb-card.html)) |

**Two limits on your side.** The interface's input takes what its full scale is (often 5–8 V on line inputs): a power amplifier's output needs an attenuator or a load box with a divider. The interface's output gives at most its own full scale (on many 2–3 V): enough for every line input, and for a phono stage only through an attenuator.

1. **Find what the device takes** in the table above, or in its data sheet.
2. **Set it in volts.** With the outputs calibrated, the keys under *Level* play volts – **100 mV, 316 mV, 0 dBu, 1 V, 2 V** – and the line under the slider says what any level is (*≈ 245 mV at the output*). With an attenuator chosen in CONNECTIONS → *Attenuator before the device*, the level and its keys are given **at the device's input**: 5 mV for a phono stage is typed as 5 mV.
3. **Then check dBFS** – they guard the interface:
   - the generator stays under *Settings → Generator output limit*;
   - the inputs that read the device peak **below −6 dBFS** (no red row in INPUTS); a device with gain lifts its output above the level it got;
   - the interface itself is cleanest between about −10 and −30 dBFS.

The recommendations in dBFS below are for an interface with a few volts at full scale; turn them into volts for your device, and keep the device's rated level where the table says so.

## Which level the generator plays

SIGNAL has three kinds of level, and **Settings → Generator output limit** (−40 to −6 dBFS, up to 0 dBFS with a warning) stands above them all: nothing plays louder, whatever the task, the slider or the sweep asks for.

- **Level** – one level for the whole task.
- **From / To / Step [dBFS]** – the steps of a level sweep (THD vs level, IMD vs level); *Level* is hidden then.
- **Distortion limit [%]** (THD vs level) – not a level the generator plays, but a threshold: the bar says at what level the sweep crossed it. With a load, the power where THD+N reaches it (an amplifier's rating); without, the output level where THD – the harmonics alone – reaches it (a tape's MOL or full modulation: a tape's THD+N holds the skirts of wow, flutter and modulation noise around the tone, often over 10 % at every level, and would never cross 2 or 3 %).

**Level is one value for all tasks**, remembered after a restart: what you set for Multitone (−6 dBFS) stays when you switch to a power amplifier. Check it before every RUN when you change the device. With calibrated outputs the line under the slider says what it means in volts (*≈ 245 mV at the output*).

**Sine or peak.** dBFS are sine-referenced, so for a sine the level is its RMS and its peak at once. A chirp, two IMD tones and a multitone are set by their **peak**: at the same *Level* they reach the same peak as the sine – the output does not clip any sooner – but their RMS, and each of their tones, are lower.

**The device's gain.** *Level* is what goes **into** the device; its output, and so the input that reads it, gets the level plus the device's gain. The recommended levels below assume about 0 dB between the interface output and input – a power amplifier through its attenuator, a phono stage through its −40 dB box. A line preamplifier with +10 dB at −6 dBFS reaches +4 dBFS and clips the input: lower *Level* by the gain (here to −16 dBFS). A phono stage has up to +19 dB more at 20 Hz than at 1 kHz, so a frequency sweep needs that much room. Watch INPUTS: a row turns red when its peak rises above −1 dBFS, a warning just before full scale; at full scale the input clips, and a clipped input gives wrong readings. The live tasks say *Input clipping detected*; a sweep (THD vs level, THD vs frequency, Stepped sine, IMD vs level, Crosstalk) names the steps where an input clipped – *Input clipping at 3 steps from -10 dBFS · those readings are wrong* – in the status line while it runs and when it ends; Noise & S/N says *Input clipping during the tone · the S/N is wrong*. A sweep over frequency plays one level at every step, so it names the frequencies and that level – *Input clipping at 31 steps at 20 – 20000 Hz with -6 dBFS*: at that level the input clipped there; from which level it starts, a frequency sweep cannot tell.

**The highest level without clipping.** Find the frequency where the device gives the most (a phono stage: 20 Hz; a flat device: 1 kHz), then either:

- **THD vs level** at that frequency: the status line says *Input clipping at … steps from -12 dBFS*, so the clipping starts between the last good step and −12 dBFS; or
- **Level meter** with *Play test signal* at that frequency: lower *Level* until the input's bar is no longer red.

Then stay 3–6 dB below it: a device's gain can change a little as it warms up, and a tape recorder's with the tape and the bias.

| Task | What *Level* is | You set it? | Recommended | Why |
| --- | --- | --- | --- | --- |
| Level meter, Spectrum | the sine of the generator (when it plays) | yes | as needed; −20 dBFS | only to have a tone to look at; often the generator is off and an external signal is measured |
| THD+N meter | the test tone | yes | the device's **rated level** in volts (line: 316 mV – 1 V; phono: 5 mV); −20 dBFS for a first look | the interface itself is cleanest at about −10 to −30 dBFS: lower, its noise rises against the tone; higher, its own distortion |
| THD vs frequency | the tone of every step | yes | the rated level in volts, or −20 dBFS | one level across the band; at the top of the band harmonics fall outside the band and THD+N shows mostly noise |
| THD vs level | – (*From / To / Step*) | yes | **−60 → −6 dBFS, step 2 dB**; a power amplifier: a low *To* first, raised until the curve turns up at clipping | from noise-limited at the bottom to clipping at the top; 2 dB steps find the distortion limit closely enough; an amplifier driven far into clipping heats its load and speakers |
| Noise & S/N with a tone | the tone that is the signal of S/N | yes | the **rated level** (often 1 kHz at 0.775 V or 1 V; a tape: the level at its distortion limit) | data sheets quote S/N against the rated level; a lower tone gives a worse S/N by the same number of dB |
| Noise & S/N without a tone | – | no | – | S/N is relative to full scale; AudioLab first plays a tone at the output limit to find where full scale lies |
| IMD | the **peak** of the two tones together | yes | −10 to −20 dBFS, below clipping | the peak of the sum is what clips; each tone plays lower (SMPTE 4 : 1: the high tone about 14 dB below the peak) |
| IMD vs level | – (*From / To / Step*, peak of the sum) | yes | as THD vs level | the same range, from noise to clipping |
| Multitone | the **peak** of all tones | yes | **−6 dBFS** | each tone plays about 25 dB lower; a lower level drowns the gaps between the tones in noise |
| Gain & phase, Tape response | the **peak** of the chirp | yes | **−20 dBFS** | the device stays in its linear range and peaks of the response have room; a tape stays below saturation, where the treble compresses first |
| Stepped sine | the tone of every step | yes | −20 dBFS | as Gain & phase; one tone at a time, so it can also go lower for a device with a lot of gain |
| Phono with *inverse RIAA* | the level at 1 kHz | yes | −20 dBFS or lower | the pre-emphasis plays the treble up to 19.6 dB louder (20 kHz) and the bass quieter; above the output limit the treble is held back and AudioLab warns |
| Crosstalk | the tone in the driven channel | yes | the **rated level** in volts | the higher the tone, the further the leak stands above the noise, and the more points are filled – but the device must not clip |
| Output Z | the tone | yes | **low**, −20 dBFS and below | the device must not clip or limit into the load; only the level ratio counts, so a low level costs nothing |
| Speed & flutter (record → playback) | the tone recorded on the tape | yes | −10 to −20 dBFS | well above the tape hiss, below saturation |
| Speed & flutter, Test tape (from tape) | – | no | – | the tone is on the tape |
| Player | – | no | – | the levels are in the test file (−1 to −90 dBFS) |

**Levels AudioLab chooses by itself:**

- **The device's delay** (before a sweep, Crosstalk, Noise): a short 1 kHz probe at most −20 dBFS, never louder than the loudest step.
- **S/N against full scale**: a 1 kHz tone at the output limit, then silence.
- **Calibration**: −10 dBFS, or the output limit when that is lower.

**Rated levels to aim for:** see [Volts for the device, dBFS for the interface](#volts-for-the-device-dbfs-for-the-interface); a CD player gives about 2 V at full scale.
