Testing an Audio Amplifier with a Signal Generator
A signal generator, an oscilloscope, a multimeter and a set of power resistors are all you need to measure an audio amplifier properly: its gain, frequency response, maximum output power, distortion, noise and stability. These measurements tell you whether a new design meets its specification, whether a repair has restored an amplifier to health, and what is really wrong when an amplifier sounds bad. The same techniques apply to preamplifiers, headphone amplifiers, active filters and any other analog signal chain.
An audio amplifier test bench: generator into the input, a power load resistor in place of the speaker, and a scope, meter and sound-card FFT on the output. Bridge-tied outputs must be measured differentially.
View LargerThe Test Setup
Connect the generator to the amplifier input, and replace the loudspeaker with a load resistor of the speaker's nominal impedance (4 or 8 Ω) rated for the amplifier's full power — a 100 W amplifier needs a 100 W (or larger) resistor on a heatsink. A resistor gives repeatable results and spares your ears and your speakers. Connect the oscilloscope across the load and, if you have one, a true-RMS multimeter as well.
Start every test at low volume and increase it gradually while watching the scope. Allow the amplifier and the load resistor to cool between full-power tests.
Gain and Frequency Response
Set the generator to a 1 kHz sine wave at a level well below clipping — for a power amplifier, enough for about one-tenth of full output — and measure the input and output voltages. The voltage gain is Vout/Vin, or 20 log₁₀(Vout/Vin) dB.
Then, keeping the input level constant, change the frequency in steps (20, 50, 100, 200, 500 Hz, 1, 2, 5, 10, 20, 50 kHz) and record the output at each. The frequencies where the output falls to 0.707 of its 1 kHz value (−3 dB) are the lower and upper limits of the frequency response. A generator's sweep mode with the scope displaying the output makes this a single picture.
Input 200 mV RMS at 1 kHz; output 5.0 V RMS across 8 Ω. Gain = 5.0 / 0.2 = 25 = 28 dB. Output power at this level = 5.0² / 8 = 3.1 W.
Output falls to 3.54 V (−3 dB) at 12 Hz and 65 kHz: the amplifier's −3 dB bandwidth is 12 Hz to 65 kHz — comfortably wider than the 20 Hz–20 kHz audio band.
Maximum Output Power
Increase the input at 1 kHz until the tops of the sine wave just begin to flatten on the scope — the onset of clipping. Read the RMS voltage across the load (or the peak-to-peak voltage, divided by 2√2). The maximum continuous output power is P = VRMS² / R.
Clipping begins at 56 Vpp across 8 Ω. VRMS = 56 / 2.828 = 19.8 V. P = 19.8² / 8 = 49 W — matching a 50 W rating. Repeat with a 4 Ω load if the amplifier is rated for it, and with both channels driven together: a weak power supply sags when both channels work at once, reducing the clipping point.
Harmonic Distortion
Total harmonic distortion (THD) is too small to see on an oscilloscope until it reaches about 1–3%. To measure lower levels, look at the output spectrum: most digital oscilloscopes have an FFT function, and a computer sound card with free audio-analysis software makes an excellent low-cost distortion analyzer, provided the signal is attenuated to the sound card's input range. The generator's own distortion must be well below what you are trying to measure — a typical function generator manages about 0.1%, while a good sound card output can be ten to a hundred times cleaner.
In decibels: a harmonic 60 dB below the fundamental is 0.1%; 80 dB below is 0.01%.
Measure THD at several power levels. It is usually lowest at moderate power, rises sharply at clipping, and in Class AB amplifiers with too little bias it rises again at very low levels because of crossover distortion.
Noise and Signal-to-Noise Ratio
Short the amplifier's input (or terminate it with a resistor equal to the source impedance it is designed for), set the volume control to maximum, and measure the RMS output noise across the load with a true-RMS meter or the scope's RMS measurement, ideally through a 20 Hz–20 kHz filter. The signal-to-noise ratio is the maximum output voltage divided by the noise voltage, in decibels.
Noise output = 0.4 mV RMS. Full output = 19.8 V RMS. S/N = 20 log₁₀(19.8 / 0.0004) = 94 dB — a good result for a power amplifier. A hum component at 50/60 Hz or its harmonics, visible on the scope or in the FFT, points to grounding or supply ripple problems rather than to the amplifier's own noise.
Input and Output Impedance
Input impedance: put a resistor Rs in series with the generator and measure the amplifier's output with and without it. If the output falls from V₁ to V₂, the input impedance is Rs × V₂ / (V₁ − V₂). Choosing Rs so that the output halves gives Rin = Rs directly.
Output impedance and damping factor: measure the output voltage with no load (Vopen) and with the rated load RL (Vload). The output impedance is Rout = RL × (Vopen − Vload) / Vload, and the damping factor is RL / Rout.
Vopen = 5.02 V, Vload = 5.00 V with 8 Ω. Rout = 8 × 0.02 / 5.00 = 0.032 Ω. Damping factor = 8 / 0.032 = 250. A typical solid-state amplifier has a damping factor of 100 or more; a valve amplifier with an output transformer may be below 20.
The Square-Wave Test
Feed the amplifier a 1 kHz and then a 10 kHz square wave at moderate level. The output square wave reveals the amplifier's character at a glance:
- Sloping tops at 1 kHz (or lower frequencies) show limited low-frequency response.
- Rounded leading edges show limited high-frequency response; the rise time relates to the upper −3 dB frequency by f ≈ 0.35 / trise.
- Overshoot and ringing after each edge show a poorly damped response — a sign of marginal stability. Repeat with a capacitor (for example 0.47–2.2 µF) across the load resistor to simulate a difficult speaker load: an amplifier that rings heavily or breaks into oscillation needs attention to its Zobel network, output inductor or compensation, as described in the Parasitic Oscillation and Stability lesson.
Frequently Asked Questions
Can I use a real loudspeaker instead of a load resistor?
For quick checks at low level, yes, but not for proper measurements. A loudspeaker's impedance varies widely with frequency, so the readings depend on the speaker; a sine wave at full power is far more punishing than music and can destroy a tweeter or a woofer; and full-power testing is painfully loud. Use a resistive load for measurements and a speaker only for listening tests.
My amplifier measures 50 W but the box says 200 W. Is it faulty?
Probably not. Many products quote “peak” or “PMPO” power, total power for all channels, or power at 10% distortion, any of which can be several times the continuous (RMS) power per channel at low distortion that you have measured. Compare like with like: continuous power per channel, into a stated load, at the onset of clipping or a stated distortion such as 1%.
The load resistor gets very hot. Is that normal?
Yes — it is dissipating the amplifier's entire output power. A 50 W sine wave into a 50 W resistor will make it too hot to touch within seconds unless it is mounted on a heatsink or immersed in oil. Use a resistor rated well above the test power, keep full-power tests short, and keep the resistor away from anything that can melt or burn.
Why does my class D amplifier's output look noisy on the scope?
Some of the switching frequency (typically 300 kHz to 1 MHz) remains after the output filter, especially in filterless designs that rely on the speaker's inductance. On the scope it appears as a fuzzy band on the audio waveform. It is inaudible, but it confuses voltage and distortion measurements. Use the scope's 20 MHz bandwidth limit, or add a simple RC low-pass filter (for example 1 kΩ and 10 nF, about 16 kHz – or a commercial audio measurement filter) between the load and the measuring instrument.
Test Your Knowledge
Answer the questions below to check your understanding of this lesson. Every answer can be found in the lesson above.