An amplifier class defines the fraction of the input signal cycle during which the active output devices (transistors or MOSFETs) conduct current, directly dictating the circuit's trade-off between audio linearity and power efficiency. In a real circuit, the chosen class changes your power supply sizing, heat sink mass, and output filter complexity. While audiophiles often debate the subjective 'warmth' of these topologies, the physics of conduction angles and thermal dissipation are absolute. This guide cuts through the forum myths and gives you the hard numbers, bench realities, and a concrete decision path for selecting the right amplifier IC for your next build.
The Core Trade-Off: Linearity vs. Efficiency
Every amplifier topology forces a compromise between how accurately it reproduces the input waveform (linearity/low distortion) and how much of the drawn power actually reaches the load (efficiency). The remaining power is burned off as heat.
- Class A: The output devices conduct for the full 360° of the signal cycle. Think of it like a water valve left permanently wide open; the bias current flows even when there is no audio signal. This eliminates crossover distortion entirely but results in massive idle heat.
- Class B: Devices conduct for exactly 180° (one handles the positive half, the other the negative). It is highly efficient but suffers from severe crossover distortion at the zero-crossing point, making it unusable for high-fidelity audio without heavy feedback.
- Class AB: The practical compromise. Devices conduct for slightly more than 180° (typically 200° to 220°). A small bias current keeps both transistors just barely 'on' around the zero-crossing, eliminating the crossover glitch of Class B while maintaining reasonable efficiency.
- Class D: A switching topology. The audio signal is compared against a high-frequency triangle wave to generate a Pulse Width Modulated (PWM) square wave. The output MOSFETs are either fully on or fully off, minimizing the time spent in the high-resistance linear region, yielding massive efficiency gains.
Class A: 50% (practical ~25%) | Class AB: 78.5% (practical ~60%) | Class D: ~100% (practical 90-95%)
The Worked Math: Heat Dissipation Across Topologies
Abstract percentages do not tell you what size heat sink to buy. Let us run a concrete numeric example to see how the different classes of amplifiers behave thermally on the bench.
The Scenario: You are designing an amplifier to deliver 50W RMS into an 8Ω load using a continuous sine wave test tone. We need to calculate the required power supply capacity and the heat sink dissipation ($P_{heat}$) for Class A, Class AB, and Class D.
1. Class A (Assume 25% practical efficiency)
To get 50W out, the power supply must deliver: $P_{in} = 50W / 0.25 = 200W$.
Heat dissipated by the output transistors: $P_{heat} = 200W - 50W = 150W$.
Bench Reality: You need a massive 200W power supply and a heat sink with a thermal resistance ($\theta_{SA}$) of less than 0.2°C/W to keep the junction temperature safe. This requires forced air or a multi-pound extruded aluminum block.
2. Class AB (Assume 65% practical efficiency at full output)
Power supply delivery: $P_{in} = 50W / 0.65 = 76.9W$.
Heat dissipated: $P_{heat} = 76.9W - 50W = 26.9W$.
Bench Reality: A standard 3°C/W passive finned heat sink will keep the chip around 80°C above ambient. Easily manageable in a well-ventilated chassis.
3. Class D (Assume 92% practical efficiency)
Power supply delivery: $P_{in} = 50W / 0.92 = 54.3W$.
Heat dissipated: $P_{heat} = 54.3W - 50W = 4.3W$.
Bench Reality: The IC package itself (like a TSSOP with an exposed thermal pad) can often dissipate 4W directly to the PCB copper pour without any external heat sink.
Where You Meet Amplifier Classes in Practice
Understanding where these topologies live in the wild helps you reverse-engineer commercial gear and set realistic expectations for your DIY builds.
- Class A: Found almost exclusively in boutique, high-end audiophile power amps (e.g., Pass Labs) and low-power headphone amplifiers. Because heat scales with power, Class A is rarely used above 30W per channel in consumer gear. If you are building a desktop headphone amp driving 32Ω IEMs, Class A is highly practical and yields excellent measurement specs.
- Class AB: The workhorse of the late 20th and early 21st centuries. You will find this inside classic AV receivers, studio monitor speakers, and professional power amps. It is the standard for applications requiring high slew rates and wide bandwidth without the complexity of high-frequency switching filters.
- Class D: Dominates modern portable Bluetooth speakers, car audio, active subwoofers, and powered PA systems. Because it runs cool and draws minimal current, it is the only viable choice for battery-powered devices or ultra-compact enclosures where airflow is restricted.
Decision Tree: Picking the Right Amp IC for Your Build
Do not get paralyzed by topology debates. Use this decision matrix to select the right silicon for your specific application constraints.
| Application Scenario | Required Topology | Recommended IC / Module | Key Design Constraint |
|---|---|---|---|
| Battery-powered portable speaker (ESP32/I2S source) | Class D | MAX98357A (I2S) or PAM8403 (Analog) | Must operate on single 3.7V-5V LiPo rail; zero heat sink. |
| High-fidelity desktop DAC/Amp (Low noise floor critical) | Class AB | LM3886 (Power) or OPA1612 (Headphone) | Requires dual-rail (±15V to ±35V) linear power supply. |
| High-power active subwoofer (200W+ into 4Ω) | Class D | TPA3255 or pre-built ICEpower module | Requires heavy-gauge power traces and strict LC filter layout. |
| General-purpose DIY bookshelf speakers (2x50W) | Class D | TPA3116D2 (Module) | Best balance of cost ($5-$10), power, and ease of use. |
Common Confusions and Bench Mistakes
When discussing the different classes of amplifiers on forums, a few persistent myths and bench errors trip up hobbyists. Here is what you need to know to avoid them.
Does 'Class D' mean the amplifier processes digital audio?
No. This is the most common confusion. Class D refers to the switching topology of the output stage, not the input signal format. The input to a standard Class D amp chip is an analog voltage. The chip internally generates a high-frequency PWM square wave to drive the MOSFETs. If you want to amplify a digital I2S stream directly, you need a specific 'Digital Input Class D' IC (like the MAX98357A), which includes an onboard DAC.
Does Class AB always sound better than Class D?
Historically, early Class D amps suffered from high Total Harmonic Distortion plus Noise (THD+N) and poor high-frequency response due to the output LC filter interacting with the speaker's impedance. Modern Class D ICs (like the TPA3255) use advanced feedback loops and high switching frequencies (400kHz+), achieving THD+N figures below 0.01%—often measuring cleaner than equivalently priced Class AB designs. The 'Class D sounds harsh' argument is largely outdated for post-2015 silicon.
Why does my oscilloscope show a square wave on the Class D output?
A classic bench mistake. If you probe the speaker terminals of a Class D amplifier directly with an oscilloscope, you will see a 400kHz PWM square wave, not your audio sine wave. The inductor-capacitor (LC) low-pass filter on the output is required to strip away the switching frequency and leave only the audio band. If you are measuring before the inductor, you are looking at the raw switching node. Always probe after the LC filter, and ensure your scope probe is rated for the high dV/dt of the switching node to avoid damaging your scope's input stage.
When designing your next audio circuit, let the thermal math and power supply constraints drive your topology choice, not audiophile folklore. For high-power, compact, or battery-driven builds, Class D is the undisputed king. For ultra-low noise, dual-rail bench instruments or headphone amps where heat is negligible, Class AB remains a reliable, predictable workhorse. Pick the IC that matches your physical constraints, layout the PCB with attention to ground returns, and let the silicon do the heavy lifting.
