Designing a Discrete Class-AB Sound Amplifier Circuit: Topology, Values, and Testing

For a 5W to 15W high-fidelity bench audio amplifier, the definitive choice is an op-amp-driven Class-AB push-pull output stage. Specifically, pairing a low-noise Texas Instruments NE5532 dual op-amp with a complementary BD139 (NPN) and BD140 (PNP) transistor pair delivers excellent thermal stability, low crossover distortion, and drives 8-ohm speakers cleanly. This guide provides the exact component values, node mappings, and failure-mode analysis required to design, build, and verify this sound amplifier circuit on a breadboard.

The Core Decision: Why Class-AB Push-Pull Over Alternatives?

When designing a sound amplifier circuit, you must choose between Class-A, Class-B, Class-AB, and Class-D topologies. Class-A offers the lowest distortion but wastes massive amounts of power as heat (efficiency < 30%). Class-B eliminates the heat but introduces severe crossover distortion at the zero-crossing point. Class-D is highly efficient but requires complex LC output filters and introduces high-frequency switching noise that complicates PCB layout.

Class-AB is the engineering compromise that dominates linear audio. By biasing the output transistors just above cutoff, we eliminate the crossover dead-zone of Class-B while maintaining reasonable efficiency (~60%). We use an op-amp in the voltage gain stage to leverage its high open-loop gain, which forces the push-pull output stage into a tight negative feedback loop, linearizing the transistor transfer curves and dropping Total Harmonic Distortion (THD) below 0.1%.

Topology Breakdown: Node Labels and Signal Flow

To understand how the circuit corrects its own errors, we must map the signal flow through four critical nodes. This topology places the output transistors inside the op-amp's feedback loop.

• Node A (Signal Input): The non-inverting input of the NE5532. AC-coupled via a 10µF film capacitor to block DC offset from the source. A 10kΩ pull-down resistor to ground sets the input impedance.
• Node B (Op-Amp Output): The raw voltage-amplified signal exits the op-amp here. This node drives the bias network. It does not drive the speaker directly.
• Node C (Bias Network Midpoint): Located between two series 1N4148 diodes (and a 100Ω trimpot). This node splits the signal to the bases of the NPN and PNP transistors. The diodes maintain a fixed ~1.2V gap between the bases, keeping both transistors slightly conducting to prevent crossover distortion.
• Node D (Emitter Junction / Audio Out): The emitters of the BD139 and BD140 meet here. This is the low-impedance, high-current output that drives the 8-ohm speaker. Crucially, Node D also feeds back to the inverting input of the op-amp via a 10kΩ resistor to close the global negative feedback loop.

Because Node D is fed back to the op-amp's inverting input, any non-linearity introduced by the transistors between Node C and Node D is instantly corrected by the op-amp adjusting its output at Node B. For a deeper theoretical background on this feedback mechanism, refer to the All About Circuits textbook chapter on Class-B and Class-AB amplifiers.

Component Behavior and Failure Modes at the Extremes

Designing a robust sound amplifier circuit requires understanding what happens when components drift, fail, or are incorrectly sized. The following behavior table details the impact of altering key elements, followed by the catastrophic failure modes at the extremes.

What Breaks at the Extremes (Open/Short Scenarios)

Design Walkthrough: Sizing the BD139/BD140 Stage

Let's lock in the exact component values for a +/- 12V DC dual-rail power supply driving an 8-ohm speaker. This configuration yields approximately 7.5W RMS of clean audio power.

1. Voltage Gain Stage: We set the non-inverting gain using $A_v = 1 + (R_f / R_{in})$. Choosing $R_f = 10k\Omega$ and $R_{in} = 1k\Omega$ gives a gain of 11. A 1V RMS input signal yields 11V RMS at the output. Into an 8-ohm load, $P = V^2 / R = 121 / 8 \approx 15W$ peak power, or roughly 7.5W continuous RMS before clipping.
2. Output Transistors: The BD139 (NPN) and BD140 (PNP) are rated for 1.5A continuous collector current and 12.5W dissipation. At 7.5W RMS into 8 ohms, the peak current is $I_{peak} = V_{peak} / R = 15.5V / 8\Omega \approx 1.9A$. Since audio is dynamic, the 1.5A continuous rating handles this easily, but peak current touches 2A. Design choice: Add small heatsinks to keep the junction temperature below 100°C.
3. Bias Network: The base-emitter voltage ($V_{be}$) of a silicon BJT is ~0.6V. Since we have an NPN and PNP in series, we need $0.6V + 0.6V = 1.2V$ of bias. Two 1N4148 signal diodes in series provide exactly this ~1.2V drop. We place a 100Ω multi-turn trimpot in series with the diodes to allow fine-tuning of the quiescent current.
4. Emitter Resistors: We use 0.33Ω, 5W wirewound resistors on the emitters. These drop a small amount of voltage but provide vital local negative feedback, stabilizing the bias point against thermal runaway as the transistors heat up. See Electronics Tutorials on Class AB Amplifiers for the thermal runaway math.

Breadboard Testing and Verification Sequence

Never plug a newly wired amplifier directly into an unprotected power supply and speaker. Follow this exact step-by-step verification sequence to prevent destroying your components.

1. Verify Power Rails: With the op-amp and transistors removed, power the breadboard. Measure the rails with your DMM. Confirm you have +12.0V and -12.0V relative to the ground rail. Power down and discharge the decoupling capacitors (100µF on each rail) by shorting them with a 1kΩ resistor.
2. Test the Op-Amp Stage: Insert the NE5532. Wire the feedback loop (10kΩ from output to inverting input, 1kΩ from inverting input to ground). Power up. Measure the DC voltage at the op-amp output (Node B). It should read within ±20mV of 0V. If it reads >100mV, check your ground connections and resistor values.
3. Install the Bias and Output Stage: Power down. Insert the 1N4148 diodes, the 100Ω trimpot (set it to its maximum resistance for safety), the BD139, BD140, and the 0.33Ω emitter resistors. Connect an 8-ohm dummy load resistor (rated for 10W) to the output. Do not connect a real speaker yet.
4. Set Quiescent Current: Power up. The current limit on your supply should engage if the bias is too high. Set your DMM to measure DC millivolts across one of the 0.33Ω emitter resistors. Slowly turn the 100Ω trimpot down. Stop when you read 16.5mV. By Ohm's law ($I = V/R$), $16.5mV / 0.33\Omega = 50mA$. This 50mA quiescent current is the sweet spot for keeping the transistors just on the edge of conduction, eliminating crossover distortion without generating excessive heat.
5. Signal Injection: Inject a 1kHz sine wave at 100mV RMS from your function generator. Probe the output (Node D) with your oscilloscope. You should see a clean 1.1V RMS sine wave. Look closely at the zero-crossing point on the scope; if there is a flat 'step' or glitch, increase the trimpot resistance slightly to raise the quiescent current to 60mA and re-measure.
6. Final Load Test: Once the waveform is clean, swap the dummy load for your 8-ohm speaker. Increase the input signal to 1V RMS and verify the audio is clear, undistorted, and that the transistor heatsinks remain warm but touchable (under 60°C) after 10 minutes of continuous operation.

By strictly following this node-mapped topology and verifying the quiescent current via voltage drop across the emitter resistors, you eliminate the guesswork from discrete audio design. The NE5532 and BD139/BD140 combination remains a benchmark for bench-top sound amplifier circuits because it forces you to understand the physics of push-pull biasing while delivering highly listenable, low-distortion audio.