Why Class D Amplifier Efficiency Is Taking Over Hi-Fi
Class D is a switching topology where the output transistor acts like a light switch, not a dimmer: it is either fully on or fully off, with no in-between state. It sits between the amplifier’s controller and the speaker’s output terminals, replacing the half-open, heat-soaking transistors of Class A and Class AB stages.
Heat is the problem, not the sound. In a linear amp the output transistor rests at a halfway-open point and bleeds off power as waste heat. Class D snaps the transistor between fully on and fully off, so that resistive loss drops to near zero.
Pulse-width modulation does the translating. The amp chops the audio signal into pulses of varying width. A small inductor and capacitor at the speaker terminal average those pulses back into a smooth waveform the driver cone can follow.
The trade-off lives in the treble. Fast switching throws off high-frequency noise and electromagnetic interference. Well-built boards tame it with sharper crossover filters and careful layout; cheaper ones let a faint buzz leak into cymbals and violin overtones.
Newer switch materials are closing the last gap. Gallium-nitride and silicon-carbide transistors switch faster than older silicon parts, shrinking the output filter and pushing noise beyond the range of human hearing.
The Definition, In One Breath
Picture a water tap. In a traditional linear amplifier, the tap sits at half-open and stays there. Water (current) trickles through the valve the whole time, and most of that trickle turns into heat at the valve body. Now imagine a different tap: it snaps fully open or fully shut, with no middle position. When it is open, water floods through with almost no resistance. When it is shut, nothing flows. You control the total water reaching the bucket by how long the tap stays open during each cycle. That is pulse-width modulation (PWM): you encode the size of the desired output signal as the duty cycle (the fraction of time the switch stays on) within each tiny slice of the audio waveform.
Class D is the name for this switching approach. The output transistor (the electronic tap) rarely rests at a halfway point. It is either saturated (fully on, behaving like a short piece of wire with almost no voltage drop) or cut off (fully off, behaving like an open circuit with zero current). Because the transistor spends its time in one of those two extremes, the power it wastes as heat at the halfway point collapses to near zero. That thermal savings lets engineers shrink heat sinks, drop the amp into a portable speaker or a sub-woofer enclosure, and still deliver the same output power a brick-sized Class A unit would need. The entire power-delivery stage becomes small, light, and cool enough to fit where a linear design simply could not.
Where The Name Comes From
Audio engineers originally sorted amplifiers by how long the output transistor conducts during one full cycle of the input signal. Class A means the transistor is on for the full 360 degrees of the cycle. Class B means each of two transistors handles 180 degrees, one for the positive half, one for the negative. Class AB splits the difference, letting each transistor handle a little more than half to smooth the handoff. Class C conducts for less than half, a scheme used in radio transmitters where you recover the full waveform with a tuned tank circuit.
When switching topology arrived for audio in the 1990s, engineers needed a label that said, “This one does not conduct for a smooth arc at all. It snaps between two states.” The next letter in the sequence, D, filled that slot. It was not meant to imply a quality step beyond C. It simply meant a different conduction style: two states, no partial state. The word “switching” got stapled to the class letter because the whole point is that the transistor works as a relay, not a variable resistor.
The naming also answers a practical question a designer faces: “How much heat do I need to pull off this board?” In Class A, the transistor dissipates power even at idle, because it usually carries current through a partial voltage drop. In Class D, at idle the switch is off and dissipates nothing; at full output, the on-state resistance is tiny and the off-state carries no current. The heat you must manage collapses. That is why the letter D has become the stamp on every portable speaker, Bluetooth soundbar, and powered sub-woofer enclosure built in the last decade.
What It Is, And What It Is Not
What it is:
Class D is a topology, a wiring and switching pattern, not a specific chip or a specific sound. It describes how the final output stage moves power from the power supply to the speaker. The signal entering the amp is still a continuous analog audio voltage. A small controller inside the board converts that voltage into a pulse train (a string of square-wave pulses whose width tracks the audio level). A tiny output filter (one inductor and one capacitor) sits at the speaker terminal and averages the pulses back into a smooth voltage the driver cone can follow. The whole chain runs: analog in, digital-style chopping in the middle, analog out.
What it is not:
- It is not “digital audio.” The audio signal that enters the amp is a continuous voltage. The switching is a method of moving power, not a method of representing sound. Your CD player, streaming box, or turntable all deliver the same analog voltage to the input terminal.
- It is not a quality grade. A Class D amp can sound clean and transparent, or it can buzz and crackle in the upper register. The class letter says nothing about build quality.
- It is not the same thing as the amplifier’s input stage. The input, volume control, and any digital-to-analog converter in a streaming amp are separate blocks. Class D names only the final power-delivery stage.
- It is not inherently superior to Class A or Class AB for every use. A well-built Class A unit running at low power in a small studio monitor can sound as smooth as a well-built Class D at the same output. The class describes a mechanism, not a ranking.
Terms People Mix Up
| Look-alike term | What it actually means | One-line tell |
|---|---|---|
| Pulse-width modulation (PWM) | The chopping pattern that encodes audio level into pulse width | It is the “how” inside Class D, not the class itself |
| Switching amplifier | Any amp whose output stage snaps between two states, including Class D and some Class F designs | Class D is one species; “switching amp” is the wider family |
| Digital amplifier | An amp that accepts a digital input (USB, S/PDIF) and builds the analog signal on-board | It may use Class D, Class AB, or any output topology; the word describes the input, not the output stage |
| Class AB | A linear topology where each output transistor handles just over half the cycle | The transistor still rests at a partial conduction point; Class D has no partial point |
| EMI (electromagnetic interference) | Unwanted radio-frequency energy radiated by the fast switching edges | It is a by-product you must filter and route around, not a feature of the music |
The Audible Consequence
Low end (bass and sub-bass):
Because the output filter smooths the pulse train into a near-identical copy of the original waveform, the bass driver receives the same shape it would from a Class A amp. In practice, you will not hear a coloration difference in the 20 to 200 hertz region if the filter is well tuned. The real gain shows up in the enclosure: the thermal headroom means the amp can deliver its full rated wattage into the sub-woofer without overheating. Bass does not compress or thin out when you play loud, long passages. A portable sub that could not have housed a Class A design now fits in a shoebox-sized box because the amp runs cool.
Midrange and vocals:
The midrange sits in the zone where switching noise and the audio signal overlap the least, so vocals and lead instruments typically stay clear and uncolored. A well-laid-out board keeps the switching-frequency harmonics (often in the tens of kilohertz range) far above the voice fundamentals around 100 to 1000 hertz. You hear the singer’s breath, the attack of a plucked guitar string, and the body of a cello bow as if the amp were simply passing the signal through with nothing added.
Highs (treble and air):
This is where the trade-off becomes audible. The fast edges of each pulse carry energy well above the 20 kilohertz hearing limit. If the output filter and PCB routing are sloppy, that energy leaks into the 5 to 20 kilohertz band where cymbal shimmer and room “air” live. You hear it as a faint metallic buzz or a harsh edge on hi-hats and violin overtones. In a well-built design, the filter and shielding pull that energy down enough that the treble sounds open and natural. In a cheap unit, the treble gets a gritty, sizzling texture that makes you reach for the speaker.
Newer gallium-nitride (GaN) and silicon-carbide (SiC) switches raise the switching frequency further, often into the hundreds of kilohertz or beyond. That pushes the harmonic noise even farther away from the audible band, and the output filter components (the inductor and capacitor) can shrink because they work at a higher rate. The last audible gap that kept some listeners loyal to a big, warm Class A cabinet is closing quickly.
Quick Answers
Does Class D Make Your Speakers Sound “Digital”?
No. The signal at the speaker terminal is a smooth analog voltage. The switching happens inside the amp’s power stage, not in the signal path you feed it. Your turntable, CD player, or streaming box delivers a continuous waveform. The amp’s internal chopping is invisible to that input. What you hear at the driver cone is an analog reproduction of the waveform you sent in. The word “digital” applies to how the signal reached the amp, not to how the power stage moves it.
Why Does A Class D Amp Run Cooler Than A Class A Amp?
In a Class A stage, the output transistor usually carries current and usually drops a partial voltage, so it bleeds off power as heat even when the music is quiet. In Class D, the transistor is either a near-short (fully on, tiny voltage drop) or an open circuit (fully off, zero current). Either way, the power lost inside the transistor itself is near zero. The waste moves to the speaker, which is where you want it: the cone converts that power into sound, not into a hot metal plate.
What Is Pulse-Width Modulation Doing, In Plain Words?
Think of a bicycle brake. You control speed by how long you squeeze the lever during each rotation. PWM works the same way: the amp’s controller decides, for each tiny slice of the audio waveform, how long to keep the switch closed. A louder note means a longer on-time; a softer note means a shorter on-time. The inductor-capacitor filter at the output then averages all those on-times into a smooth voltage that matches the original note’s height. The speaker rarely sees the individual pulses.
Can You Hear The Difference Between A Good Class D And A Bad Class D?
Usually, yes. A well-designed board keeps switching-frequency harmonics well above 50 kilohertz and routes the high-current paths so they do not pick up the audio signal. The result is a clean, transparent treble. A poorly designed board lets those harmonics dip into the 10 to 20 kilohertz range, and you hear a faint buzz on cymbals, a hiss above the music, or a sizzle on fast transient hits. The class letter on the box does not tell you which one you have.
Why Are GaN And SiC Switches Changing The Game?
Older silicon transistors switch fast, but their on-resistance and switching loss grow at very high frequencies. Gallium-nitride and silicon-carbide materials turn on and off with far less energy, even at several hundred kilohertz. That lets the output filter use a smaller inductor and capacitor, shrinks the whole board, and pushes the audible-noise floor lower. The last reason some listeners reached for a big Class A unit for its “warmth” is disappearing as those newer switches go into mass production.
(This article first appeared on hifireport.com. Reproduction without permission is not allowed.)