A class-D amplifier or switching amplifier is an electronic amplifier in which the amplifying devices (transistors, usually MOSFETs) operate as electronic switches, instead of as linear gain devices as in other amplifiers. The signal to be amplified is a train of constant amplitude pulses, so the active devices switch rapidly back and forth between a fully conductive and nonconductive state. The analog signal to be amplified is converted to a series of pulses by pulse width modulation, pulse density modulation or other method before being applied to the amplifier. After amplification, the output pulse train can be converted back to an analog signal by passing through a passive low pass filter consisting of inductors and capacitors. The major advantage of a class-D amplifier is that it can be more efficient than analog amplifiers, with less power dissipated as heat in the active devices.
Output stages such as those used in pulse generators are examples of class D amplifiers. However, the term mostly applies to power amplifiers intended to reproduce signals with a bandwidth well below the switching frequency.
Class D amplifiers work by generating a variable duty cycle square wave of which the low-frequency portion of the spectrum is essentially the wanted output signal, and of which the high-frequency portion serves no purpose other than to make the wave-form binary so it can be amplified by switching the power devices.
A passive low-pass filter removes the unwanted high-frequency components, i.e., smoothes the pulses out and recovers the desired low-frequency signal. To maintain high efficiency, the filter is made with purely reactive components (inductors and capacitors), which store the excess energy until it is needed instead of converting some of it into heat. The switching frequency is typically chosen to be ten or more times the highest frequency of interest in the input signal. This eases the requirements placed on the output filter. In cost sensitive applications the output filter is sometimes omitted. The circuit then relies on the inductance of the loudspeaker to keep the HF component from heating up the voice coil. It will also need to implement a form of three-level (class BD) modulation which reduces HF output, particularly when no signal is present.
The structure of a class D power stage is essentially identical to that of a synchronously rectified buck converter, a type of non-isolated switched-mode power supply (SMPS). Whereas buck converters usually function as voltage regulators, delivering a constant DC voltage into a variable load and can only source current (one-quadrant operation), a class D amplifier delivers a constantly changing voltage into a fixed load, where current and voltage can independently change sign (four-quadrant operation). A switching amplifier must not be confused with linear amplifiers that use an SMPS as their source of DC power. A switching amplifier may use any type of power supply (e.g., a car battery or an internal SMPS), but the defining characteristic is that the amplification process itself operates by switching.
Theoretical power efficiency of class D amplifiers is 100%. That is to say, all of the power supplied to it is delivered to the load, none is turned to heat. This is because an ideal switch in its on state will conduct all the current but has no voltage across it, hence no heat would be dissipated. And when it is off, it will have the full supply voltage standing across it, but no current flows through it. Again, no heat would be dissipated. Real-life power MOSFETs are not ideal switches, but practical efficiencies well over 90% are common. By contrast, linear AB-class amplifiers are always operated with both current flowing through and voltage standing across the power devices. An ideal class B amplifier has a theoretical maximum efficiency of 78%. Class A amplifiers (purely linear, with the devices always "on") have a theoretical maximum efficiency of 50% and some versions have efficiencies below 20%.
The term "class D" is sometimes misunderstood as meaning a "digital" amplifier. While some class D amps may indeed be controlled by digital circuits or include digital signal processing devices, the power stage deals with voltage and current as a function of non-quantized time. The smallest amount of noise, timing uncertainty, voltage ripple or any other non-ideality immediately results in an irreversible change of the output signal. The same errors in a digital system will only lead to incorrect results when they become so large that a signal representing a digit is distorted beyond recognition. Up to that point, non-idealities have no impact on the transmitted signal. Generally, digital signals are quantized in both amplitude and wavelength, while analog signals are quantized in one (e.g. PWM) or (usually) neither quantity.
The binary waveform is derived using pulse-width modulation (PWM), pulse density modulation (sometimes referred to as pulse frequency modulation), sliding mode control (more commonly called "self-oscillating modulation" in the trade.) or discrete-time forms of modulation such as delta-sigma modulation.
The most basic way of creating the PWM signal is to use a high speed comparator ("C" in the block-diagram above) that compares a high frequency triangular wave with the audio input. This generates a series of pulses of which the duty cycle is directly proportional with the instantaneous value of the audio signal. The comparator then drives a MOS gate driver which in turn drives a pair of high-power switches (usually MOSFETs). This produces an amplified replica of the comparator's PWM signal. The output filter removes the high-frequency switching components of the PWM signal and recovers the audio information that the speaker can use.
DSP-based amplifiers which generate a PWM signal directly from a digital audio signal (e. g. SPDIF) either use a counter to time the pulse length or implement a digital equivalent of a triangle-based modulator. In either case, the time resolution afforded by practical clock frequencies is only a few hundredths of a switching period, which is not enough to ensure low noise. In effect, the pulse length gets quantized, resulting in quantization distortion. In both cases, negative feedback is applied inside the digital domain, forming a noise shaper which has lower noise in the audible frequency range.
Distortion and switching speed
Two significant design challenges for MOSFET driver circuits in class D amplifiers are keeping dead times and linear mode operation as short as possible. "Dead time" is the period during a switching transition when both output MOSFETs are driven into Cut-Off Mode and both are "off". Dead times need to be as short as possible to maintain an accurate low-distortion output signal, but dead times that are too short cause the MOSFET that is switching on to start conducting before the MOSFET that is switching off has stopped conducting. The MOSFETs effectively short the output power supply through themselves, a condition known as "shoot-through". Meanwhile, the MOSFET drivers also need to drive the MOSFETs between switching states as fast as possible to minimize the amount of time a MOSFET is in Linear Mode, the state between Cut-Off Mode and Saturation Mode where the MOSFET is neither fully on nor fully off and conducts current with a significant resistance, creating significant heat. Driver failures that allow shoot-through and/or too much linear mode operation result in excessive losses and sometimes catastrophic failure of the MOSFETs.
The switching power stage generates both high dV/dt and dI/dt, which give rise to radiated emission whenever any part of the circuit is large enough to act as an antenna. In practice, this means the connecting wires and cables will be the most efficient radiators so most effort should go into preventing high-frequency signals reaching those:
- Avoid capacitive coupling from switching signals into the wiring.
- Avoid inductive coupling from various current loops in the power stage into the wiring.
- Use one unbroken ground plane and group all connectors together, in order to have a common RF reference for decoupling capacitors
- Include the equivalent series inductance of filter capacitors and the parasitic capacitance of filter inductors in the circuit model before selecting components.
- Wherever ringing is encountered, locate the inductive and capacitive parts of the resonant circuit that causes it, and use parallel RC or series RL snubbers to reduce the Q of the resonance.
- Do not make the MOSFETs switch any faster than needed to fulfil efficiency or distortion requirements. Distortion is more easily reduced using negative feedback than by speeding up switching.
Power supply design
Class D amplifiers place an additional requirement on their power supply, namely that it be able to sink energy returning from the load. Reactive (capacitive or inductive) loads store energy during part of a cycle and release some of this energy back later. Linear amplifiers will dissipate this energy away, class D amplifiers return it to the power supply which should somehow be able to store it. In addition, half-bridge class D amps transfer energy from one supply rail (e.g. the positive rail) to the other (e.g. the negative) depending on the sign of the output current. This happens regardless of whether the load is resistive or not. The supply should either have enough capacitive storage on both rails, or be able to transfer this energy back.
The actual output of the amplifier is not just dependent on the content of the modulated PWM signal. The power supply voltage directly amplitude-modulates the output voltage, dead time errors make the output impedance non-linear and the output filter has a strongly load-dependent frequency response. An effective way to combat errors, regardless of their source, is negative feedback. A feedback loop including the output stage can be made using a simple integrator. To include the output filter, a PID controller is used, sometimes with additional integrating terms. The need to feed the actual output signal back into the modulator makes the direct generation of PWM from a SPDIF source unattractive. Mitigating the same issues in an amplifier without feedback requires addressing each separately at the source. Power supply modulation can be partially canceled by measuring the supply voltage to adjust signal gain before calculating the PWM  and distortion can be reduced by switching faster. The output impedance cannot be controlled other than through feedback.
Despite the complexity involved, a properly designed class D amplifier offers the following benefits:
- Reduced power waste as heat dissipation and hence:
- Reduction in cost, size and weight of the amplifier due to smaller (or no) heat sinks, and compact circuitry,
- Very high power conversion efficiency, usually better than 90% above one quarter of the amplifier's maximum power, and around 50% at low power levels.
- Home Theatre systems. In particular the economical "home theatre in a box" systems are almost universally equipped with class D amplifiers. On account of modest performance requirements and straightforward design, direct conversion from digital audio to PWM without feedback is most common.
- Mobile phones. The internal loudspeaker is driven by up to 1 W. Class D is used to preserve battery lifetime.
- Powered speakers
- High-end audio is generally conservative with regards to adopting new technologies but class D amplifiers have made an appearance
- Active subwoofers
- Sound Reinforcement and Live Sound. The Crest Audio CD3000, for example, is a class-D power amplifier that is rated at 1500 W per channel, yet it weighs only 21 kg (46 lb). Similarly, the Powersoft K10 is a class-D power amplifier that is rated at 6000 W per 2-ohm channel, yet it weighs only 12 kg (26.5 lb).
- Bass amplifiers Again, an area where portability is important. Example: Yamaha BBT500H bass amplifier which is rated at 500 W, and yet it weighs less than 5 kg (11 lb). The Promethean P500H by Ibanez is also capable of delivering 500 W into a 4 Ohm load, and weighs only 2.9 kg. Gallien Krueger MB500, also rated at 500W weighs no more than 1.36kg.
- Class A amplifier (a linear, non-PWM, amplifier class)
- Class B amplifier (a linear, non-PWM, amplifier class)
- Class C amplifier (a non-PWM amplifier class)
- Class T amplifier (a proprietary implementation of Class D)
- Delta-sigma modulation
- Sliding mode control
- The generic analysis of sliding mode control is quite math heavy. The specific case of 2-state self-oscillating class D amplifiers is much more intuitive and can be found in Globally Modulated Self-Oscillating Amplifier with Improved Linearity, 37th AES Conference
- The Analog Devices AD1990 Class-D audio power amplifier is an example.
- Sandler et al., Ultra-Low Distortion Digital Power Amplification, Presented at the 91st AES convention
- Analytical and numerical analysis of dead-time distortion in power inverters
- irf.com - IRAUDAMP7S, 25W-500W Scalable Output Power Class D Audio Power Amplifier Reference Design, Using the IRS2092S Protected Digital Audio Driver, page 26, 2008-08-29
- Putzeys et al. All Amplifiers etc., Presented at the AES 120th convention
- Boudreaux, Randy, Real-Time Power Supply Feedback Reduces Power Conversion Requirements For Digital Class D Amplifiers
- Group review of "high end" class D offerings and round-table discussion with amplifier designers
- "Home > Products > CD 3000(r)". Crest Audio. Retrieved 2013-07-16.
- Burning man Festival K10
- Yamaha BBT 500H Specifications
- List of PWM Amplifiers
- Sánchez Moreno, Sergio Class-D Audio Amplifiers - Theory and Design - Contains material on the theory and design of Class-D amplifiers.
- Haber, Eric Designing With Class D Amplifier ICs - Some IC-oriented Class D design considerations
- Harden, Paul Introduction to Class C,D,E and F, The Handiman's Guide to MOSFET "Switched Mode" Amplifiers, Part 1 - An article on basic digital RF amplifier design intended for ham radio operators but applicable to audio Class D amplifiers.