Class D DIY Amplifier Theory: PWM Switching, IC Specs, and Heat Math

A Class D amplifier is a switching amplifier that converts an analog audio signal into a high-frequency pulse-width modulated (PWM) square wave to drive speakers with theoretical efficiencies exceeding 90%. In a real circuit or installation, this topology completely changes the physical footprint: it eliminates the massive aluminum heat sinks and heavy linear transformers required by Class AB designs, replacing them with compact switching power supplies (SMPS) and surface-mount MOSFETs that run cool enough to touch. Builders commonly confuse 'Class D' with 'digital audio'—but the signal path remains entirely analog; the 'D' simply denotes the next letter after C in historical amplifier classification, not digital signal processing.

The PWM Engine: How Class D Switching Actually Works

Unlike Class A or AB amplifiers that act as variable resistors to drop voltage across output transistors, a Class D amp operates its output MOSFETs strictly as switches—they are either fully ON or fully OFF. The core of this operation is a high-speed comparator. Your analog audio input is fed into one side of the comparator, while a high-frequency triangle wave (the carrier, typically switching between 300 kHz and 600 kHz) is fed into the other.

When the audio voltage is higher than the triangle wave, the output switches high; when it drops below, the output switches low. This generates a PWM square wave where the pulse width is directly proportional to the instantaneous audio voltage. This square wave drives an H-bridge of MOSFETs, which then feeds into an LC low-pass filter (an inductor and a capacitor). The filter strips away the high-frequency carrier, leaving only the amplified analog audio signal to drive the speaker voice coil.

To visualize this, imagine controlling the flow of a garden hose not by squeezing the nozzle (which wastes energy as heat and friction, like Class AB), but by rapidly turning the tap fully on and fully off thousands of times a second. The bucket (the LC filter and speaker) averages out the rapid pulses into a smooth, continuous stream of water.

Worked Numeric Example: Heat and Power Supply Sizing

Let's look at what this efficiency actually means on the bench when designing a 50W RMS into 4 ohms desktop amplifier.

• Class AB Baseline (approx. 50% efficiency): To deliver 50W to the speaker, the power supply must provide 100W. The remaining 50W is dissipated as heat. You need a massive heat sink (often >2°C/W thermal resistance) and a 100W+ linear transformer.
• Class D Reality (approx. 92% efficiency): To deliver 50W to the speaker, the power supply provides roughly 54.3W. Only 4.3W is wasted as heat across the MOSFETs and inductor. This 4.3W is easily handled by the small copper ground plane on the PCB, requiring zero external heat sink.

However, because Class D amps rely on switching power supplies, you must account for peak current draw. A 50W Class D amp playing dynamic music with a 12dB crest factor will demand transient current spikes that a cheap, under-specced SMPS cannot handle, resulting in clipped bass. Always size your 24V SMPS for at least 1.5x the continuous RMS rating (e.g., a 75W to 100W supply for a 50W amp).

Benchmarking the Silicon: DIY Class D IC Comparison

Choosing the right silicon is the most critical decision in a Class D DIY amplifier build. The market is dominated by a few key integrated circuits, each with distinct trade-offs regarding voltage tolerance, THD+N (Total Harmonic Distortion plus Noise), and quiescent current. Below is a spec-sheet comparison of the most common ICs you will encounter in 2026.

The TI TPA3116D2 remains the undisputed king of budget DIY desktop amps due to its low voltage requirement and forgiving layout. However, if you are building high-power active subwoofers or floor-standing speakers, the IRS2092 (which requires external MOSFETs, allowing for massive power scaling) or the newer MA5332BMS (which features an inductor-less filter topology for certain loads) are vastly superior choices.

Where You Meet This In Practice (And What Goes Wrong)

You will encounter Class D topology in almost every modern active studio monitor, Bluetooth portable speaker, car audio subwoofer plate amp, and soundbar. In the DIY space, it is the default choice for battery-powered audio projects and ultra-compact desktop builds. However, the transition from linear to switching amplifiers introduces specific failure modes that trip up builders accustomed to traditional LM3886 or Class AB designs.

Power Supply Rejection Ratio (PSRR) and Ground Loops

The most common complaint on DIY audio forums regarding Class D amps is a high-pitched whine or hiss. This is rarely the amplifier IC's fault; it is almost always a Power Supply Rejection Ratio (PSRR) issue. Class D amps inherently have poor PSRR at high frequencies. If you power your TPA3116D2 board with a cheap, unfiltered 24V SMPS that has 50mV of high-frequency switching ripple, that ripple will modulate directly into the audio band. According to Analog Devices' technical literature on Class D design, maintaining a clean DC bus with low-ESR bulk capacitors (like 3300µF Nichicon or Panasonic snap-ins) right at the board's power terminals is mandatory to starve the amp of high-frequency noise.

Dead-Time and Crossover Distortion

In the H-bridge, there is a microsecond delay called 'dead-time' inserted between turning off the high-side MOSFET and turning on the low-side MOSFET to prevent a direct short (shoot-through). If this dead-time is too long, you get crossover distortion (a harsh, grainy sound at low volumes). If it is too short, you get shoot-through (spikes in current that melt the silicon). Modern ICs like the MA5332BMS handle this internally with adaptive dead-time control, but if you are rolling your own discrete IRS2092 build, tuning the dead-time resistors on the gate driver is a mandatory oscilloscope task.

Builder's FAQ