---
Title: Spectrum and Level meter – AudioLab
---

# Spectrum and Level meter

The signal itself: what is in it, and how loud each input is – the first look before any other measurement.

## At a glance

| Task | What it tells you | Wiring | Level | A good result |
| --- | --- | --- | --- | --- |
| **Spectrum** | hum, interference, harmonics, anything that should not be there | any; *Play test signal* only when needed | −20 dBFS, or the external signal | hum below −100 dBFS at 50/100 Hz |
| **Level meter** | the level of all four inputs; calibration; overload | any | as needed | no input near 0 dBFS (no red row, no *CLIPPING*) |

## Spectrum

Checks the signal: hum, interference, harmonics; a quick look before another measurement. Choose *Play test signal* as needed and press **RUN**.

![Spectrum of the right output with Peaks and Log f on: a 10 kHz tone at −26.8 dBFS, hum at 50 Hz (−88 dBFS) and 150 Hz (−95 dBFS) above a noise floor around −105 dBFS; the reference input below, its floor near −160 dBFS](figures/spectrum_chart.jpg)

**Settings → Analysis:**

- *FFT window*: Hann for general use, FlatTop for exact tone levels, BlackmanHarris for weak components next to a strong tone.
- *FFT length*: longer gives finer resolution (Δf above the chart) but slower updates.
- *Averages*: 4–8 calm the noise floor.

The device outputs and, each in its own colour, the references (the **REF** key hides them). The noise floor in the chart is not the total noise – it depends on the FFT length and window (see ENBW above the chart, and [What dBFS means](levels.html#what-dbfs-means-and-what-can-be-measured)).

### Keys above the chart

| Key | What it does |
| --- | --- |
| **Log f** | A logarithmic frequency axis from 10 Hz, as in the response charts – every octave the same width, so hum, its harmonics and the low end get room. Off: the analyser's linear axis, where a tone's harmonics, IMD products and sidebands stand evenly spaced and the top octave takes half the width. Also in Noise, THD+N meter and IMD. Only the chart changes; the choice is remembered. Kept traces and limits show only on the axis they were made on. Multitone is always logarithmic, the spectrogram linear. |
| **Peaks** | Marks the ten strongest lines of each shown output with a triangle, their frequency and level (in the axis' unit) – the tone, its harmonics, hum, interference – as the peak markers of a spectrum analyser. A line must stand clear of the noise beside it: about 15 dB in a single capture, less with averaging, so averaging brings out weaker lines; the harmonics of the strongest line need about 10 dB. The sidelobes of a stronger line do not count, nor do sidebands within ±1 % of it (wow and flutter). **Copy** lists every marked line. Frequency and level are refined between bins. |
| **Spectrogram** | The spectrum over the last minute – time down from now, frequency on the spectrum's own linear axis from 0 Hz (with the same margins, so a frequency stays in place when the key flips the view), the level as colour. It shows what comes and goes: hum that switches with a device nearby, interference, the modulation noise of a tape around its tone. Every capture counts, not the average. It shows the first shown channel (switch the other off with L / R); with Settings → Chart → *Fixed level axis* the colours span Top to Bottom, else the floor to the loudest point. |

![Spectrogram of the last minute on an M4 loopback, no test signal: a steady line at 20 kHz, and bursts at 15, 20 and 22 kHz and in the bass where a device nearby was switched on and off](figures/spectrogram_chart.jpg)

The spectrogram above: a minute on a loopback with no test signal. The steady line at 20 kHz belongs to the interface; the short bursts at 15, 20 and 22 kHz and the low end came when a device nearby was switched on and off – in the spectrum they would flash by unnoticed.

**Settings → Chart → Noise spectra as density** draws the spectrum per √Hz (dBFS/√Hz, dBV/√Hz after calibration), so noise reads the same at any FFT length; tones then read low by the window's noise bandwidth.

## Level meter

For calibration, setting levels and spotting overload. All four inputs are shown at once as bars, labelled by the wiring (for example *In 1 · L ref*). For each input:

| Value | Meaning |
| --- | --- |
| **RMS** | true RMS without DC in dBFS (a full-scale sine reads 0 dBFS); after calibration also in **V**, **dBV** and **dBu** (dBu = dBV + 2.22); with a load chosen in *On the device output*, the power into it (pW to W) |
| **tone** | frequency and level of the strongest component – what the calibration uses; while the generator plays, its tone (within ±5 % of its frequency), so mains hum on a channel whose tone is weak is not read instead |
| **peak** | largest sample in dBFS, after calibration also in **Vpk** – the **diamond** in the chart; for a sine it sits on top of the bar in dBFS and 3 dB above it in dBV, higher for music, noise or clicks (the crest factor); at 0 dBFS the reading says **CLIPPING** |
| **gain** | on a device output with a reference input: its tone against the reference's tone, averaged over a second, in hundredths of a dB – in dBV when both inputs are calibrated, else in dBFS (then the gain only if both inputs are set alike) |

Each bar has its RMS value written above it – in **volts** when the input is calibrated, otherwise in dBFS – always at the same height. The bar under the chart has fixed columns: **In 1 and In 2 on the first row, In 3 and In 4 on the second** (L and R side by side in stereo tasks), each with a label in its bar's colour, so an entry does not move when its numbers change. **Bold** is the main value (RMS in volts on a calibrated input, otherwise in dBFS); the tone is teal, the peak purple, and every number says what it is under the mouse.

A calibrated input reads in dBV and its diamond is the real peak voltage, so a sine's peak sits 3 dB above its bar (Vpk = √2 · Vrms); an uncalibrated input reads in dBFS with a dashed full-scale line, and as dBFS is sine-referenced, a sine's peak sits on top of its bar. With only some inputs calibrated the dBFS bars use the left axis and the dBV bars the right one, shifted by the calibration, so a bar stands at the same height for the same level against full scale.

![The Level meter with In 1 calibrated and the other inputs not](figures/level_meter_chart.jpg)

Here only **In 1** is calibrated (3.162 V at 0 dBFS): its bar is labelled **316 mV** and reads −10 dBV on the right axis, which stands level with −20 dBFS on the left. Its diamond sits 3 dB higher, at the peak of 447 mVpk. In 2 – In 4 are not calibrated: dBFS on the left axis, the diamond on top of the bar, and the red dashed line at full scale (0 dBFS). The tone is 1 kHz, −20 dBFS at the references and −10 dBFS at the outputs: a device with a gain of +10 dB.

*Play test signal* plays a tone for calibration or for checking the wiring; without it the meter reads whatever reaches the inputs – an external signal, or with the device switched off the true floor of each input.

### REL and Max hold: turning a trimmer

**Large** above the chart writes each used input's level large over its bar – with **REL** on also its rel figure, with **Max hold** how far below the maximum – to be read from across the room while turning a trimmer (remembered). Two keys above the chart add a third line to each input, in hundredths of a dB. Both read the **tone** level averaged over the last second, so a tape's level wavering on the head holds still.

- **REL** stores each input's tone level now as its 0 dB; *rel* is the level against it. The stored level stays when the tone changes: REL on a test tape's 1 kHz, then its 10 kHz tone reads the treble against 1 kHz while you turn the playback EQ trimmer. An input with no tone when the key is pressed takes its own first settled level once its tone comes. Switch the key off and on to store a new level.
- **Max hold** keeps each input's highest level since the key went on; *max* is how far the level is below it (0 or less). For a tape recorder's bias set past its maximum (overbias): record the tone the service manual names, raise the bias until the playback passes its peak, then on until *max* reads the drop the manual gives – or switch REL off and on at the peak and read *rel*, which a change of the tone or level does not upset (see [Tape recorder](tape.html#recording-bias-record-level-and-record-eq)). For a head's azimuth by level: turn the screw past the peak, then back until *max* reads 0.00 dB.
- The maximum is **marked across each bar** (orange), as the tick a peak-hold meter leaves, with the held level and its tone over it (*max −73.52 dBV at 10.0 kHz*); how far below it the level is now is *max* in the readout. It stays until the key goes off and on – not another tone, a new run or the tone ending reset it, so the peak is there to return to.
- A reading counts only when its **tone carries most of the input's power** (within 1 dB of the RMS). Hum and lines in the noise, left when nothing plays, waver by tenths of a dB and are not taken: Zoom says *no tone*, the hold stays empty until a tone plays, and Balance and *gain* need a tone on both inputs.
- When the tone ends – a test tape moves on or stops – and only hum or hiss is left, *rel* and *max* keep the tone's last figures and the status line says *no tone on In …*.
- The figures are averaged over **Settings → Analysis → Level meter → Averaging** (1 s, up to 10 s). A tape's level wavers with its contact on the head; in Zoom 2–3 s make it clear whether a turn of the screw raises or lowers the level, at the cost of a slower reaction – turn, then wait that long.
- **Zoom** (switches Max hold on) draws each bar against its own maximum, from −6 to 0 dB: the bar climbs to the orange line as the trimmer or the head reaches the peak, a tenth of a dB visible from across the room, with the figure written large on it and the held level over the line.
- **Balance** writes large over each pair how far **In 2 is from In 1** and **In 4 from In 3** (the tone levels averaged over a second; in dBV when both inputs of a pair are calibrated), readable from the trimmer while you balance two channels. Within the tolerance of **Settings → Analysis → Level meter** (0.1 dB unless changed) the figure turns green. The pairs are fixed by the inputs, not by the wiring: put the two channels to compare on In 1 and In 2, or on In 3 and In 4.

| Adjustment | Signal | Read |
| --- | --- | --- |
| Playback level | test tape's reference tone | the output in V or dBV against the nominal level; L and R alike |
| Playback treble (EQ) | test tape: REL on 1 kHz, then the 10 kHz tone | *rel* to what the tape's standard asks (0 dB when flat) |
| Record level | *Play test signal* 1 kHz, record, monitor from tape | *gain* 0.00 dB: tape as loud as the source |
| Bias (overbias) | *Play test signal* at the manual's tone well below saturation, record, monitor from tape; REL on the peak | *rel* at the manual's drop, e.g. −4 dB for a common tape at 19 cm/s (*max* shows the same) |
