---
Title: Reading the charts – AudioLab
---

# Reading the charts

What the chart of each task shows, which line is which, and the other views you can switch to. The legend always names every curve; this page says what they mean.

## Line styles

The **solid** line is always the result of the measurement. Everything that only helps to read it is broken:

| Style | Meaning |
| --- | --- |
| **solid** | the measured result: the device output, or the main figure (THD+N, IMD, gain, crosstalk) |
| **a colour of its own** | the **reference** – what the interface itself shows at the device input (spectra, Multitone, the sweeps) |
| **dashed** | a second quantity on the **right axis** (phase, weighting curve); the **reference curve** in the sweeps; a **kept trace**; **limits**; the **cursor** |
| **dotted** | a **derived** curve: the estimate of the device's own THD, the d2 and d3 parts of IMD; a **pinned trace** (the interface floor) |
| **open circle** | a value that is only an **upper bound** – the true value hides in the noise (Crosstalk) |
| **dot with a label** | a marked line of the spectrum: a harmonic, an IMD product, a peak, a flutter source |

Every curve has its colour in the same order in every chart (left and right channel, their second curve, their reference), so L and R never share one; the **L**, **R** and **REF** keys beside the RUN button show or hide them. *L → R* or *R → L* in a name says which channel plays and which one is read.

## Views common to many charts

- **Enlarge** (above the chart) moves the chart to its own window, for example to a second monitor.
- **Keep** saves the curves as a dashed **trace** over the next measurements with the same kind of chart (the same axes); **Traces (n)** shows, renames, pins, saves and removes them. A pinned trace (the interface floor) is dotted and grey and appears by itself whenever the task runs on that interface.
- **Limits** draw an upper and a lower limit, dashed, and judge the result against them.
- **The cursor**: move the mouse over the chart; a dashed vertical line, a dot on every curve and their values in the top-left corner. On a spectrum it snaps to the highest point nearby, so a harmonic or a hum line is easy to hit.
- **Settings → Chart**: *Show levels in volts (dBV)* (after calibrating the inputs), *Fixed level axis* (the same level range every time, as on an analyser), *Noise spectra as density (per √Hz)*.
- Only the chart changes with these choices: nothing is measured again.

## Level meter

**Bars**, one for each input: the bar is the **RMS level**, the diamond above it the **peak** (0 dBFS = clipping). The number above the bar is the reading. Reference inputs and device outputs have the colours of their channel. The dashed line is **0 dBFS**, full scale. With calibrated inputs the bars read in dBV, on a second axis when only some inputs are calibrated.

![Level meter: four bars, In 1 calibrated and read on the right axis in dBV, the others in dBFS with the dashed full-scale line](figures/level_meter_chart.jpg)

## Spectrum

The level of every frequency of the signal: the device output, and in its own colour the reference (the device input), shown with **REF**. The noise floor is the "grass" at the bottom, tones are the needles.

![Spectrum with Peaks and Log f on: the tone, hum lines and the noise floor, each peak labelled](figures/spectrum_chart.jpg)

| Key or setting | What it shows |
| --- | --- |
| **Log f** | logarithmic frequency axis: every octave the same width, so hum and the low end get room. Off: the analyser's linear axis, where harmonics stand evenly spaced |
| **Peaks** | the ten strongest lines of each output marked with their frequency and level – the tone, its harmonics, hum, interference |
| **Spectrogram** | the spectrum over the last minute: time runs downward, level is colour – hum or interference that comes and goes |
| *Noise spectra as density* (Settings) | the axis in dBFS/√Hz (or dBV/√Hz): the noise then reads the same at any FFT length; tones read lower |

Details: [Spectrum](spectrum.html).

## THD+N meter

The **spectrum** of the device output (and of the reference with **REF**), with the **harmonics** of the tone marked by dots and labelled H2, H3 …, and the strongest other line (hum, interference) labelled *spur*.

![THD+N meter: the spectrum of the output with the harmonics H2 … H10 marked and the hum labelled spur](figures/thdn_meter_chart.jpg)

| Key | What it shows |
| --- | --- |
| **dBc** | levels relative to the fundamental: the tone sits at 0 dB and each harmonic reads directly in dBc |
| **Log f** | logarithmic frequency axis (as in Spectrum) |
| **Scope** | the signal in **time**: a few periods of the output (left axis) and the **residual** – what is left without the fundamental, magnified on the right axis in % of the peak. Spikes at the peaks = clipping, steps at the zero crossings = crossover distortion, a slow wave = hum, a fast one = oscillation |

![Scope: four periods of the output on the left axis and the residual, magnified, on the right axis](figures/scope_loopback.jpg)

Details: [Distortion](distortion.html).

## THD vs level, THD vs frequency

One point per step. **Solid**: the device's THD+N and THD (or SINAD when Settings → Analysis → *Distortion figure* is SINAD). **Dashed**: the THD+N of the **reference** – the interface's own distortion; where the two meet, the device is better than the setup can show. **Dotted**: *Subtract reference THD* – an estimate of the device's own THD, drawn only where the output is at least 6 dB above the reference ([why](distortion.html)).

![THD vs level: THD+N solid falling with level as the noise loses weight, THD solid lower with its minimum near −16 dBFS, the reference's THD+N dashed, the estimate dotted on top of THD](figures/thd_vs_level_chart.jpg)

How to read it: the device's **THD+N** (the upper solid line) falls 1 dB for every dB of level – at low levels it is the device's noise, not its distortion. Its **THD** (the lower solid lines) has its minimum where the harmonics rise out of the noise (here near −16 dBFS) and then grows as the device starts to distort. The **dashed** THD+N of the reference lies far below the device's THD+N over the whole sweep: the setup is not what limits the result. The **dotted** estimate compares the device's THD with the reference's THD (not drawn); here the reference is much cleaner, so the estimate lies on the THD.

| Choice above the chart | Options |
| --- | --- |
| **Y** | distortion in **dB**; distortion in **%** on a logarithmic axis; the **harmonics** H2–H5 each in dBc, so you see which order grows; the **output level**, where clipping or compression bends the line |
| **X** (THD vs level) | the **generator level** (dBFS); the **output level** of the device (V with calibrated inputs); the **output power** (W, needs a load in the wiring and calibrated inputs) |

With a *Distortion limit* the bar says where the curve crossed it. A sweep can be re-read with another analysis (band, harmonics, weighting) without measuring again.

## Noise & S/N

The **spectrum** of the device output with the generator silent: the noise and hum that the S/N figures sum up. **REF** adds the noise of the interface (its floor). The keys of Spectrum (**Log f**, density) apply. The figures themselves (unweighted, A, ITU-R 468) are in the bar.

Details: [Noise & S/N](noise.html).

## IMD

The **spectrum** around the two test tones, with the **intermodulation products** marked by dots. **dBc** shows them relative to the tones the IMD method refers to; **Log f** as in Spectrum.

![IMD, SMPTE with dBc and Log f: the 7 kHz tone at 0 dBc with its products marked by dots beside it, the 60 Hz tone 12 dB above it (SMPTE reads the products against the high tone)](figures/imd_chart.jpg)

## IMD vs level

One point per step. **Solid**: total IMD; **dotted**: its **d2** (second-order) and **d3** (third-order) parts; **dashed**: the IMD of the **reference**. **Y**: IMD in dB or in % on a logarithmic axis.

![IMD vs level: solid IMD of L and R, dotted d2 and d3, dashed the reference, over the generator level](figures/imd_vs_level_chart.jpg)

## Multitone

The spectrum of the many tones on a logarithmic axis. The **thin solid** line is the spectrum (the tones are its needles), the **dashed line with points** the distortion and noise **between the tones** (TD+N per band) – the curve that counts. The reference is drawn paler. The dots on the tones give the response.

![Multitone of a line preamplifier at −16 dBFS: 31 tones, each near −30 dBFS, the TD+N per band dashed with points, the reference paler below](figures/multitone_chart.jpg)

How to read it: the dashed line is on the same axis as the tones (dBFS, or dBV after calibration) – it is the distortion and noise of each band as a sine of the same power. Relative to its tone it is the **distance** between the tone's dot and the dashed point below it: a tone at −30 dBFS and its band at −119 dBFS give −89 dB (the 16 kHz band of the chart above, its worst). Details list these distances for seven bands, and the bar gives TD+N over all of them. The dashed line rises with frequency even for a clean device: the bands are a third of an octave wide, so a higher band is wider and holds more noise.

Details: [IMD and Multitone](imd.html).

## Gain & phase, Tape response

**Solid**: the **magnitude** (gain in dB, left axis); **dashed** in the same colour: the **phase** (in degrees, right axis). Frequencies where the measurement is not reliable (low coherence: noise, a stop band) are left out, so the curve can have gaps.

![Gain & phase of a line preamplifier: the magnitude solid at +10 dB, the phase dashed on the right axis](figures/gain_phase_chart.jpg)

| Choice (ANALYSIS) | What changes |
| --- | --- |
| *Compare with phono curve* | the magnitude shows the **deviation from RIAA**: a perfect phono preamplifier is a flat line at its gain. The phase stays the device's own (for RIAA about −50° at 1 kHz) |
| *Magnitude from power ratio* | the magnitude from the power ratio – survives the speed variations of a tape, but noise adds to it (Tape response uses it always) |
| *Loop correction* | removes the response of the interface's own inputs, measured once without the device |

Details: [Frequency response](response.html).

## Stepped sine, Test tape

One point per frequency step. **Solid**: the **gain**; **dashed** on the right axis: the **phase** (Stepped sine with a reference; points where the tone is too weak for a phase are left out). A Test tape shows the steps found on the tape for L, R and **L−R**, relative to the step nearest 1 kHz. *Compare with phono curve* works here too.

![Stepped sine of a phono preamplifier compared with RIAA: the deviation solid near 0 dB, +0.26 dB at 20 kHz; the phase dashed on the right axis](figures/stepped_riaa_chart.jpg)

Details: [Stepped sine](response.html), [Test tape](tape.html#test-tape).

## Output Z

**Bars** per channel: how much the level drops with the load connected against without it (dB); the label gives the drop and the impedance. Until both readings are there, the label says which one is missing.

![Output Z: a bar per channel, −0.578 dB with a load of 31.9 and 31.4 Ω, Z 2.19 Ω (L) and 2.16 Ω (R)](figures/output_z_chart.jpg)

Details: [Output Z](output-z.html).

## Crosstalk

**L → R** and **R → L**: how much of the tone of the playing channel appears in the other one, in dB (lower is better). **Filled points** joined by a line are the crosstalk itself; **open circles** are upper bounds where the leak does not stand 10 dB above the hum and noise.

![Crosstalk over frequency: filled points for the measured values, open circles for upper bounds](figures/crosstalk_chart.jpg)

Details: [Crosstalk](crosstalk.html).

## Speed & flutter

Over **time**: the **speed deviation** in % (0.5–200 Hz, unweighted), the same **weighted** with the IEC 60386 curve – what the weighted figures are made of – and the **level** of the tone on the right axis (dropouts show as dips).

![Speed & flutter over time: the speed deviation, unweighted and weighted, and the level of the tone](figures/flutter_time_chart.jpg)

**Spectrum** (key above the chart): the speed deviation **by frequency**, % RMS on a logarithmic axis. Shaded bands mark **drift**, **wow**, **flutter** and the **scrape** band above 200 Hz (shown, not counted). The **dashed grey** curve on the right axis is the IEC 60386 weighting. The labelled dots are the strongest lines – a platter, a capstan, a pinch roller.

![Speed & flutter, Spectrum: lines at 1 Hz (wow) and 8 Hz (flutter) over the shaded bands, the IEC weighting dashed](figures/flutter_spectrum_chart.jpg)

Details: [Tape recorder](tape.html).

## Player

The **spectrum** of the part of the test file being played, for L and R; the status line says which part it is. The results fill the bar and Details as the parts pass.

Details: [Player](player.html).
