A Class AB amplifier is a push-pull audio circuit that combines the low distortion of Class A with the high efficiency of Class B by biasing both output transistors slightly 'on' to eliminate crossover distortion. If you are designing an audio output stage, repairing a vintage receiver, or sizing a power supply for a DIY speaker build, understanding this topology dictates your heatsink mass, linear power supply sizing, and overall thermal management strategy. Unlike switching amplifiers, a Class AB stage operates in the linear region, meaning it changes how you must handle continuous heat dissipation and low-frequency power supply ripple.
The Core Mechanism: Eliminating Crossover Distortion
In a pure Class B amplifier, two transistors (one NPN, one PNP) handle opposite halves of the AC audio waveform. The NPN pushes current during the positive half-cycle, and the PNP pulls current during the negative half-cycle. The problem? Silicon bipolar junction transistors (BJTs) require about 0.6V to 0.7V at the base-emitter junction before they begin to conduct. When the audio signal crosses the zero-volt centerline, neither transistor is conducting. This creates a dead zone known as crossover distortion.
Think of crossover distortion like a relay handoff where both runners drop the baton for a split second; Class AB keeps both runners holding the baton simultaneously during the exchange. By injecting a small DC bias voltage (usually around 1.2V to 1.4V total, or roughly two diode drops) between the bases of the output pair, both transistors remain slightly conducting even when the audio signal is at absolute zero. This 'idling' state is the defining characteristic of Class AB.
The Math: Efficiency, Heat, and a 50W Numeric Example
Let's run the numbers on a discrete 50W Class AB output stage driving an 8Ω loudspeaker using a dual ±35VDC linear power supply. This is a standard topology for high-fidelity desktop amplifiers.
1. Peak Voltage and Current
To deliver 50W into 8Ω, the peak voltage required at the speaker terminals is calculated via the power formula:
V_peak = √(2 × P × R) = √(2 × 50 × 8) = √800 ≈ 28.28V
With a ±35V supply, we have roughly 6.7V of headroom to account for transistor saturation voltages (Vce_sat) and the voltage drop across the emitter resistors.
2. Power Drawn from the Supply
The DC power drawn from the dual rails to produce a continuous 50W sine wave is:
P_DC = (2 × V_CC × V_peak) / (π × R_L) = (2 × 35 × 28.28) / (π × 8) ≈ 78.8W
3. Heat Dissipation (The Critical Metric)
The power that must be dissipated as heat by the output transistors is the difference between what the supply delivers and what the speaker receives:
This 28.8W must be split between the NPN and PNP output pairs. If you build this using ON Semiconductor MJL21193/MJL21194 complementary power BJTs, you cannot just bolt them to a tiny piece of metal. Assuming a maximum junction temperature (Tj) of 150°C, a 25°C ambient room, and a conservative safety margin targeting 100°C junction temp, your heatsink thermal resistance must be no higher than 2.5°C/W per device. In practice, you use a single massive extruded aluminum finned heatsink rated for roughly 1.0°C/W to handle the combined thermal load safely.
Where You Meet Class AB in Practice (And What It Changes)
You will find Class AB topology in heavy, high-end AV receivers, powered studio monitors (like the analog sections of classic KRK or Yamaha nearfields), and the power sections of most solid-state guitar amplifiers. Choosing Class AB fundamentally changes two things in your installation or build:
- Power Supply Architecture: Class AB amplifiers have relatively poor Power Supply Rejection Ratio (PSRR) at low frequencies compared to modern Class D amps. This means you cannot use a cheap, noisy Switched-Mode Power Supply (SMPS). You must use a heavy, unregulated linear power supply featuring a massive toroidal transformer and large filter capacitors (e.g., 10,000µF per rail) to prevent 60Hz/120Hz mains hum from bleeding into the audio.
- Physical Footprint and Weight: Because of the linear power supply and the large aluminum heatsinks required to dissipate that 28.8W of heat, a 50W Class AB amplifier will easily weigh 15 to 25 lbs. A 50W Class D amplifier using a modern ICEpower module weighs less than 2 lbs.
Common Confusions: Class AB vs. Class D and Class H
When reading datasheets or forum debates, topology names get thrown around loosely. Here is what people commonly confuse with Class AB:
| Topology | How It Works | Max Theoretical Efficiency | Primary Use Case |
|---|---|---|---|
| Class AB | Linear push-pull with slight forward bias to eliminate dead zones. | 78.5% | Audiophile Hi-Fi, studio monitors, guitar amps. |
| Class D | Switching topology using Pulse Width Modulation (PWM) at 300kHz+. | >90% | Subwoofers, portable Bluetooth speakers, PA systems. |
| Class H | A Class AB output stage paired with a dynamic, multi-tiered power supply that switches to higher voltage rails only during loud audio peaks. | Variable (higher than AB) | Pro-audio PA amplifiers requiring high power with lower heat than standard AB. |
Class H is the most frequent point of confusion. Class H is not a different output stage; it is literally a Class AB output stage with a smarter power supply. If you see a 'Class H' pro-audio amp, the output transistors are still operating in the linear Class AB region.
Decision Tree: Which Amplifier Topology Should You Build?
Do not default to Class AB just because it is 'classic.' Use this decision path to select the right topology for your specific project constraints.
| If your project requires... | Then choose... | Concrete Part / Module Pick |
|---|---|---|
| >200W for a subwoofer or PA speaker where weight and heat are critical. | Class D | Hypex NC400 or ICEpower 1200AS module. |
| Battery-powered portable audio where every milliamp of efficiency counts. | Class D | Texas Instruments TPA3116D2 IC. |
| <100W desktop Hi-Fi where absolute lowest THD+N, zero EMI switching noise, and pristine midrange are paramount. | Class AB | Texas Instruments LM3886TF (see below). |
FAQ: Biasing, Thermal Runaway, and Practical Gotchas
How do I measure and set the quiescent bias current on a discrete Class AB amp?
Never measure current by breaking the circuit and putting your multimeter in series with the high-voltage rail—one slip will short the supply. Instead, measure the DC voltage drop across the 0.22Ω (or 0.33Ω) emitter resistors on your output transistors. Using Ohm's Law (I = V/R), if you want 50mA of idle bias current, you adjust the bias trimpot until your multimeter reads exactly 11mV across a 0.22Ω resistor (0.011V / 0.22Ω = 0.05A). Always let the amp warm up for 15 minutes before taking the final reading, as the current will drift as the silicon heats up.
What is thermal runaway and how do I prevent it?
Thermal runaway is the classic Class AB failure mode. As silicon heats up, its base-emitter voltage (Vbe) drops by roughly 2mV/°C. If your bias voltage remains fixed, the transistor turns 'on' harder, drawing more quiescent current, which creates more heat, which drops the Vbe further in a runaway loop until the silicon melts. To prevent this, the transistor that generates the bias voltage (the Vbe multiplier) must be physically bolted directly to the same heatsink as the output devices. This allows the bias circuit to track the thermal state of the output stage and automatically reduce the bias voltage as the heatsink gets hot.
Can I use a switching power supply (SMPS) with a Class AB amp?
You can, but you will likely regret it unless the SMPS is specifically designed for audio with massive output capacitance and high switching frequencies (>100kHz). Standard industrial SMPS units lack the instantaneous current delivery and low-frequency ripple rejection that a heavy copper linear transformer provides. If you force an SMPS onto a high-end Class AB stage, expect a noticeable 60Hz hum and compressed bass dynamics during transient peaks.
