A Class AB amplifier is a solid-state audio circuit that combines the low-distortion linearity of Class A with the higher efficiency of Class B by biasing its output transistors slightly into conduction even when no audio signal is present. In a real circuit, this topology changes the game by eliminating the harsh "crossover distortion" dead zone inherent in pure Class B designs, dropping Total Harmonic Distortion (THD) from unusable >1% levels down to <0.05% while keeping heat dissipation manageable. Hobbyists and junior engineers commonly confuse Class AB with Class D (which uses high-frequency PWM switching and LC output filters rather than linear transistor operation) or Class H (which is simply a Class AB core with dynamic power rail switching to improve efficiency at high volumes).

Think of a Class B amp as a two-lane highway merge where both lanes close completely when no cars are present; the first car in each direction has to wait for the gate to open, causing a jarring stop-and-go traffic jam (crossover distortion). Class AB leaves the gates cracked open just enough for a trickle of traffic to flow continuously, ensuring a seamless handoff when real volume arrives.

Bench Note: While modern Class D modules (like Hypex Ncore or Purifi Eigentakt) have largely conquered the subwoofer and portable PA markets, Class AB remains the undisputed baseline for low-noise, high-fidelity analog signal paths in studio monitors and audiophile integrated amplifiers.

The Core Mechanism: Biasing Past the Crossover Dead Zone

In a standard complementary push-pull output stage, an NPN transistor handles the positive half of the audio waveform, while a PNP transistor handles the negative half. Silicon bipolar junction transistors (BJTs) require a base-emitter voltage (VBE ≈ 0.6V) to begin conducting. If you drive the bases directly from a voltage source, there is a ±0.6V dead band where neither transistor is on. The output sits at zero while the input crosses the middle, creating a notch of severe distortion.

Class AB solves this by inserting a bias spreader—typically a VBE multiplier circuit or a string of forward-biased diodes—between the bases of the output devices. This spreader holds the bases apart by roughly 1.2V to 1.4V, forcing both output transistors into a slight state of conduction (quiescent bias) even when the input signal is exactly 0V. As the audio signal swings positive, the NPN device takes over fully while the PNP device gracefully turns off, and vice versa for the negative swing. For a deeper look at the mathematical derivation of this bias network, Elliott Sound Products provides an exhaustive breakdown of VBE multiplier design.

Worked Numeric Example: Heat and Efficiency in a 75W Stage

Let’s calculate the real-world thermal and power numbers for a typical discrete Class AB output stage driving an 8Ω loudspeaker. We will use dual ±35V DC power rails.

  • Peak Output Voltage: ~35V (minus ~1V for transistor saturation and emitter resistor drops, so ~34V peak).
  • RMS Output Voltage: 34V / √2 = 24.04V.
  • Maximum Continuous Sine Wave Power: (24.04V)2 / 8Ω = 72.2 Watts.

Now, let's look at the quiescent (idle) heat. We set our bias trimpot to yield a quiescent current (Iq) of 50mA. Because the total voltage across the output stage from rail to rail is 70V, the idle power dissipation is:

Pidle = 70V × 0.05A = 3.5 Watts

This 3.5W is dissipated as heat on the heatsink before a single watt of audio is played. Compare this to a pure Class A amplifier delivering 72W, which would dissipate over 140W of heat continuously. The table below illustrates the efficiency trade-offs across amplifier classes at maximum rated power.

Amplifier Class Theoretical Max Efficiency Real-World Efficiency (72W into 8Ω) Idle Heat Dissipation
Class A 25% (inductive load) / 50% (transformer) ~20% > 150W
Class AB 78.5% ~55% - 65% 3W - 10W
Class B 78.5% ~65% ~0W
Class D > 95% ~85% - 92% < 2W

For a comprehensive review of the underlying semiconductor physics governing these efficiency limits, refer to the Electronics Tutorials guide on Class AB amplifiers.

Where You Meet Class AB Amps in Practice

You will rarely see pure Class B in audio; it is almost exclusively a theoretical stepping stone. Class AB, however, is everywhere in analog signal chains:

  • Powered Studio Monitors: The Yamaha HS8 and KRK Rokit G4 series use Class AB output stages to drive their tweeters and woofers, prioritizing phase coherence and low high-frequency distortion over the absolute efficiency of Class D.
  • AV Receivers: Mid-tier home theater receivers from Denon and Marantz rely on discrete Class AB push-pull stages using Sanken or Toshiba TO-3P output transistors to deliver clean multi-channel power.
  • Op-Amp Output Stages: Almost every classic linear operational amplifier you use on a breadboard—from the NE5532 to the OPA1612—features a Class AB output stage internally to drive low-impedance loads without crossover notch distortion.
  • Guitar Pedals: Line drivers and buffer pedals utilize Class AB biased emitter followers to maintain signal integrity over long cable runs.

Bench Scenario Walkthrough: The Thermal Runaway Trap

Setting the bias on a discrete Class AB amplifier is a rite of passage for DIY audio builders. It is also where most beginners destroy their output stages. Here is a real-world walkthrough of a biasing procedure that went wrong.

The Setup

I was commissioning a 100W discrete Class AB amplifier board (similar to the classic ESP P68 design). The output stage used two pairs of complementary TO-3P BJTs mounted on a large extruded aluminum heatsink. Each output transistor had a 0.22Ω 5W ceramic emitter resistor to force current sharing.

The Numbers and Procedure

  1. Power the amp via a current-limited bench supply set to ±20V (half voltage) to prevent catastrophic failure during initial testing.
  2. Connect a multimeter in DC millivolt mode across the 0.22Ω emitter resistor of the first NPN output transistor.
  3. Target voltage: 11mV. (Using Ohm's Law: V = I × R → 0.011V = I × 0.22Ω → I = 50mA quiescent current).
  4. Slowly adjust the blue 10kΩ multi-turn bias trimpot on the VBE multiplier circuit until the meter reads exactly 11mV.
  5. Repeat for the PNP transistor to ensure symmetry.

The Outcome

I adjusted the trimpot, locked in 11mV across both sides, and connected a test speaker. The amp played music beautifully with zero audible crossover distortion. Satisfied, I powered it down, mounted the PCB in the chassis, and powered it up at full ±40V rail voltage to run a 1kHz sine wave test into a dummy load.

What Went Wrong: Thermal Runaway

Three minutes into the sine wave test, a sharp smell of burning phenolic resin filled the room. I killed the power, but it was too late. The 0.22Ω emitter resistors were scorched black, and one of the output transistors had shorted internally.

The Root Cause: The VBE multiplier transistor (the component setting the bias voltage) was mounted flat on the PCB, not thermally coupled to the main heatsink. Silicon BJTs have a negative temperature coefficient of roughly -2mV / °C. As the output transistors dissipated power, the heatsink temperature rose by 40°C. The output transistors' VBE dropped by 80mV, meaning they required less base voltage to conduct. However, the bias spreader on the cool PCB kept pushing the exact same static voltage. The output transistors began conducting massive amounts of idle current, which generated more heat, which dropped their VBE further, in a positive feedback loop known as thermal runaway. The quiescent current spiked from 50mA to over 4 Amps in seconds, melting the emitter resistors.

The Fix: The VBE multiplier transistor must be physically clamped to the same heatsink as the output devices. As the heatsink warms up, the multiplier transistor warms up equally, dropping its bias voltage output at the exact same -2mV/°C rate, maintaining a stable 50mA idle current regardless of temperature.

Frequently Asked Questions

Why do Class AB amps still use massive heatsinks if they are efficient?

While Class AB is more efficient than Class A, its maximum theoretical efficiency (78.5%) is only achieved at absolute maximum clipping. At typical listening volumes (1W to 5W), a Class AB amp operates much closer to Class A efficiency, dissipating the majority of the drawn rail power as heat. A 100W rated amp playing at 1W might still dissipate 20W to 30W of heat per channel, requiring substantial aluminum mass to keep the junction temperatures safe.

Can I use MOSFETs instead of BJTs in a Class AB output stage?

Yes, lateral MOSFETs (like the Exicon ECX10N20/10P20) are highly prized in Class AB designs. Unlike BJTs, MOSFETs do not suffer from secondary breakdown and have a positive temperature coefficient at high currents, which inherently prevents thermal runaway at high power levels. However, they require a higher bias voltage (typically 3V to 5V total across the gates instead of 1.2V for BJTs) and a more complex VGS multiplier circuit.

Is crossover distortion completely eliminated in Class AB?

It is reduced to negligible levels, but not mathematically eliminated. Because the transconductance (gain) of a transistor changes slightly as it transitions from the Class A biased region into the high-current Class B region, there is a tiny "gain glitch" at the crossover point. High-end designs use massive amounts of global negative feedback and high bias currents (sliding bias) to push this distortion below the noise floor of the measurement equipment.