Measuring a power amplifier

Step by step, from building the load to the last number. Each step says what to set, what you should see, and what to check when you do not. The overview of all devices is on What to measure, the controls are explained in the Manual.

An integrated amplifier on a wooden shelf: volume and input selector knobs, tone controls and a display showing a 44.1 kHz input

At a glance

What is worth measuring on a power amplifier, with which task and what a good result is; the steps below show how.

WhatTaskSetGood value
THD+N at 1 W, 1 kHzTHD+N meter1000 Hz; raise Level until the bar shows 1 W0.01 % good, 0.001 % excellent
Power at 1 % THD+NTHD vs level1 kHz, From −40 To −6 dBFS, Step 1 dB; Load setas rated; the bar reads “1 % THD+N at … W”
Distortion over the bandTHD vs frequency20 Hz – 20 kHz, 31 points, level for a few Wa mild rise towards 20 kHz is normal
Kind of distortionTHD+N meter → Scope1 kHz at a few W, then near clippinga smooth residual; steps at the zero crossings = crossover distortion (bias too low), spikes at the peaks = clipping
Adjusting the quiescent current liveMultitone (or THD+N meter) with RUNlevel for a few W; turn the trimmer slowlythe lowest TD+N, and no change when warm
S/NNoise & S/NS/N against a test tone on, level = rated power; read the A value100–120 dB re rated power
Frequency responseGain & phase−20 dBFS20 Hz – 20 kHz ±0.5 dB
CrosstalkCrosstalkstereo wiring, level for a few W−60 to −90 dB at 1 kHz
Damping factorOutput ZLoad 8 Ω; read without, then with the load resistor (step 7.7)above 100 (Z below 0.08 Ω)
HumSpectrumgenerator off, input connectedbelow −100 dBFS at 50/100 Hz

1. Safety first

2. What you need

ItemFor
Audio interface with four inputs and its manufacturer's driver (e.g. MOTU M4)stereo with a reference per channel; two inputs work one channel at a time
Dummy load per channel: two 4 Ω resistors rated for the power (e.g. 100 W, 1 %, in an aluminium case) on a heatsinkreplaces the speaker: 8 Ω or 4 Ω
Balanced attenuator per channel, switched −22 / −12 / −10 dB (step 3)brings the speaker voltage down to line level
True-RMS multimetercalibration, measuring the attenuator, checking the power
Cables: RCA to the amplifier input with a Y-splitter for the reference; TRS from the attenuator to the interfacewiring

3. Building the load and the attenuator

Schematic in two parts. Dummy load: two 4 ohm 100 W resistors R1 and R2 in series; J1 (8 ohm) at the top, J2 (4 ohm) at their junction, J3 (COM) at the bottom, each with a second post (J9, J8, J6) for the attenuator. Attenuator: from J7 (IN AMP) through 9.1 k (R6, bridged by SW2A) and 1 k (R3), from J5 (COM) through 1.1 k (R7, bridged by SW3A) and 1 k (R4); the output is taken balanced across R5 (1 k) to the tip and ring of J4, whose sleeve goes to the interface ground

Two blocks per channel, joined by short cables. Dummy load: the amplifier's plus to J1 (8 Ω) or J2 (4 Ω), its minus to J3 (COM); each has a second post for the attenuator (J9, J8, J6). Attenuator: J7 (IN AMP) and J5 (COM) take the voltage, the output is read balanced across R5. The jack's sleeve goes to the interface ground only, never to COM, so a bridged (BTL) amplifier is safe too.

SW2ASW3AAttenuation3 V at the interface at
openopen1 k / 13.2 k = −22.4 dB40 V: 196 W into 8 Ω – above the load's rating
closedopen1 k / 4.1 k = −12.3 dB12.3 V: 19 W into 8 Ω, 38 W into 4 Ω
closedclosed1 k / 3 k = −9.5 dB9 V: 10 W into 8 Ω, 20 W into 4 Ω

Measure the attenuation once for each setting you use: play a 1 kHz tone into the attenuator, read the voltage before and after it with the multimeter, and note 20·log10(Vafter / Vbefore) – you enter this value in AudioLab as the Divider of the output inputs. Use wire of at least 1.5 mm² for the power connections.

4. Wiring

Stereo amplifier wiring, references first: interface OUT L through a Y-splitter to the amplifier input L and to IN 1 (reference L); OUT R through a Y-splitter to the amplifier input R and to IN 2 (reference R); each speaker output (+ and −) by speaker cable to its dummy load and attenuator; the attenuators go by balanced TRS cables to IN 3 (output L) and IN 4 (output R)

5. Setting up AudioLab

  1. CONNECTIONS: Audio interface = your interface, Wiring = Stereo, references first, On the device output = 8 Ω (or 4 Ω, as wired; it starts at Nothing (open) every time AudioLab opens).
  2. Calibration (once per interface): CALIBRATION → Calibrate…, an interface output straight to each input, multimeter across the same connection – see Calibration.
  3. Divider: in the calibration summary enter the measured attenuation (e.g. 20.3) in Divider [dB] for IN 3 and IN 4. AudioLab then shows the voltage on the load, and with the load set, the power in W.
  4. Settings → Chart → Show levels in volts (dBV) on.
  5. Settings → Generator output limit: a level the amplifier cannot take beyond its rating.
  6. Amplifier volume (if it has one) at maximum or a fixed position; note it.

6. Know the floor first

Before the amplifier, connect an interface output straight to inputs 3 and 4 (no amplifier) and run THD+N meter and Noise & S/N – or simply watch the reference rows during every measurement. That is the best the interface can show. Optionally Keep it and pin it in TRACES, so it shows under later measurements of the same task.

What the attenuator does to the floor: it brings the signal down by 22 dB, the interface noise stays. With an interface noise of about −114 dBFS and 1 W into 8 Ω (2.83 V on the load, 0.21 V after −22.4 dB, about −31 dBFS on a 7.5 V input), THD+N cannot read better than about −83 dB (0.007 %). At 10 W it is about −93 dB (0.002 %) – or −103 dB with the attenuator on −12 dB, which still keeps the interface input below 3 V. A better amplifier reads at this limit – read its THD (harmonics only, which sit far below the noise in their own bins) or measure at higher power.

7. Measurements

7.1 THD+N at 1 W, 1 kHz

  1. MEASUREMENT → THD+N meter; SIGNAL: 1000 Hz, Level −40 dBFS; ANALYSIS: Unweighted; Settings → Analysis: Band up to 20000, Up to harmonic 10, FFT 65536, Blackman-Harris, Averages 4.
  2. RUN, raise Level step by step until the bar shows 1 W (2.83 V on 8 Ω).
  3. Read THD+N and THD of both channels; tick Details for the harmonics.

Good: THD+N 0.01 %, excellent 0.001 % (see the floor above – at 1 W read THD). Check: multimeter on the load posts shows 2.83 V (= 1 W into 8 Ω) within a few percent.

7.2 Power at 1 % THD+N

  1. MEASUREMENT → THD vs level; SIGNAL: 1000 Hz, From −40, To about 6 dB above the 1 W level, Step 1 dB.
  2. SINGLE. With the load set and the outputs calibrated, the axis runs in W and the bar reads 1 % THD+N at … W.
  3. If it reads not reached, raise To (within the generator limit) and repeat.

Good: close to the rated power (often a little below with a weak mains or 1 % resistors). The curve falls with level, then rises sharply at clipping – that knee is the usable power.

7.3 Distortion over the band

THD vs frequency, 20 Hz – 20 kHz, 31 points, level for a few W, SINGLE. Good: flat, a mild rise towards 20 kHz is normal (less feedback at high frequencies). Above 10 kHz THD reads “—” (harmonics above 20 kHz), THD+N stays valid.

7.4 Frequency response

Wiring = Stereo (optional, see step 4). Gain & phase, −20 dBFS, RUN. Good: 20 Hz – 20 kHz ±0.5 dB; with volts on the gain is the amplifier's voltage gain (typically +26 to +32 dB). With both channels the bar adds L−R.

7.5 Noise and S/N

Noise & S/N, S/N against a test tone on, tone at the rated power level (or 1 W for a comparison), SINGLE. Read the A-weighted value. Good: 100–120 dB re rated power. interface limit = the amplifier is at least that quiet.

7.6 Crosstalk

Crosstalk, level for a few W, 20 Hz – 20 kHz, SINGLE. Good: −60 to −90 dB at 1 kHz, rising towards 20 kHz.

7.7 Damping factor (output impedance)

Output Z compares the output voltage without and with the load. Without the load, the attenuator is connected straight to the amplifier's output (J7, J5) and the dummy load is off; with it, the amplifier is back on J1 and the attenuator on J9/J6 (step 3).

  1. Output Z, 1 kHz, low level (a few hundred mV on the output). No load (amplifier to the attenuator only), On the device output = Nothing (open), RUN.
  2. Amplifier on J1 (8 Ω), attenuator on J9/J6, On the device output = 8 Ω.

Good: damping factor above 100 (Z below 0.08 Ω); a modern amplifier often reads Z below … – too stiff to resolve with an 8 Ω load. Some amplifiers do not like running without a load – keep the level low and the open reading short.

7.8 Hum

Spectrum, Play test signal off, amplifier input connected. Good: 50/100 Hz below −100 dBFS.

8. Is the result valid?

CheckWhat it tells
Output well above the reference (THD+N, IMD, noise)the number is the amplifier's. Output close to the reference = you measure the interface; the amplifier is at least that good.
interface limit in Noise & S/Nthe amplifier is quieter than the interface can show
Watts in AudioLab = V²/R from the multimeter on the loadcalibration and divider are right
No CLIPPING, no Input clipping detectedthe interface input is not overloaded (otherwise switch the attenuator to −22 dB, both switches open)
Fundamental not “No test signal”the signal reaches the inputs the wiring expects (Inputs line above the chart)
Both channels similara large difference points to a fault in one channel (or one cable)

9. When it does not fit

SymptomLikely cause
THD+N far worse than expected, spectrum full of lines at 50/100/150 Hzground loop: laptop charger (measure on battery; a desktop computer through a USB 2.0 High Speed isolator), mains of the amplifier and the PC on different sockets
Higher noise than beforelaptop charger connected, multimeter leads left on, loose cables
Strange spikes high in the spectrum, unstable readingsthe amplifier oscillates (no load, long speaker wires) or a class D carrier without the AES17 filter
Output reads “No test signal”cable not fully inserted, wrong Wiring or Generator outputs
Power in W clearly offdivider value not entered or not measured, Load not matching the post used
Level drops when more inputs are connecteda TS jack in a balanced quarter-inch output of the interface – use TRS or the RCA outputs