Audio Amplifier Circuit Design

Introduction

An audio amplifier raises the power of a small audio signal to a level that can drive a loudspeaker or a pair of headphones. The source signal from a microphone, a digital-to-analog converter, or a line output may carry only a few millivolts and a vanishingly small current. The amplifier must reproduce the shape of that signal faithfully while delivering watts of power into a low-impedance load. Achieving fidelity and efficiency at the same time is the central challenge of audio amplifier design.

This article surveys the principles that govern audio amplifier circuits: the operating classes that trade efficiency against linearity, the topologies that arrange transistors into voltage and current gain stages, the biasing schemes that hold the output stage at a stable operating point, the role of negative feedback in suppressing distortion, and the supporting disciplines of power supply design, thermal management, and load handling. The treatment emphasizes solid-state amplifiers, which dominate modern audio, while noting where vacuum-tube practice differs.

Amplifier Classes

The class of an amplifier describes how its active devices conduct over the cycle of the input waveform. Conduction angle, the fraction of the cycle during which a device carries current, determines both the theoretical efficiency and the dominant distortion mechanism. Audio designers choose a class to balance heat, cost, and sound quality.

Class A

In a Class A stage, the output device conducts for the entire signal cycle, with a quiescent bias current at least as large as the peak load current. Because the device never cuts off, Class A avoids the crossover distortion that afflicts stages handing off between two devices, and it offers the most linear transfer characteristic. The penalty is efficiency. A single-ended Class A stage delivering into a resistive load through a coupling network reaches a theoretical maximum efficiency of only twenty-five percent, and a transformer-coupled or inductively loaded stage reaches at most fifty percent. The device dissipates its full quiescent power even when no signal is present, so Class A amplifiers run hot and demand large heat sinks. They survive in high-end and low-power audio, where their linearity justifies the waste.

Class B

A Class B stage uses two complementary devices, each conducting for one half of the cycle, so the conduction angle of each device is one hundred eighty degrees. With no quiescent current, the stage dissipates almost nothing at idle, and the theoretical maximum efficiency rises to about seventy-eight and a half percent for a sinusoidal signal. The drawback is crossover distortion: near the zero crossing, both devices are nearly off, and the transfer characteristic develops a flat, nonlinear region as conduction passes from one device to the other. Pure Class B is rarely used for high-fidelity audio because this distortion is most audible at low signal levels.

Class AB

Class AB is the workhorse of audio power amplification. A small quiescent bias keeps both output devices conducting through the region around the zero crossing, so each device conducts for slightly more than half the cycle. This bias fills in the dead zone of Class B and greatly reduces crossover distortion, while the idle dissipation remains far below that of Class A. Efficiency falls between the two parent classes and depends on signal level and the chosen bias point. Almost every conventional linear power amplifier for loudspeakers operates in Class AB.

Class D

A Class D amplifier does not operate its devices in their linear region at all. Instead, it switches them fully on and fully off at a high frequency, encoding the audio signal in the duty cycle of the switching waveform through pulse-width modulation or a related scheme. Because a switch in saturation or cutoff dissipates little power, Class D amplifiers reach efficiencies above ninety percent and run cool enough to fit in compact, battery-powered, and high-power installations alike. A passive low-pass filter at the output recovers the audio and removes the switching carrier. The design challenges shift from linearity to switching speed, electromagnetic interference, dead-time control, and the fidelity of the modulator and output filter. Modern Class D designs achieve distortion low enough for high-fidelity use, and they now dominate portable, automotive, and subwoofer applications.

Other Classes

Several further classes refine these basics. Class G and Class H amplifiers improve the efficiency of a linear output stage by switching or modulating the supply rails so that the devices see only as much voltage headroom as the instantaneous signal requires, reducing dissipation on large peaks. Class C, which conducts for less than half a cycle, is highly efficient but too nonlinear for audio and belongs to tuned radio-frequency work. Class AB and Class D, sometimes combined in hybrid designs, account for the great majority of audio output stages built today.

Amplifier Topologies

Topology describes the overall arrangement of stages and the path that the signal takes from input to load. A complete power amplifier is a cascade of stages, each contributing a particular kind of gain, wrapped in a feedback loop that sets the closed-loop behavior.

The Three-Stage Architecture

The classic solid-state power amplifier comprises three stages in cascade. A differential input stage compares the input signal with a feedback sample and produces an error current; it sets the input impedance, the noise floor, and the common-mode rejection. A voltage-amplifier stage, often called the second stage or the trans-impedance stage, converts that error current into the large voltage swing required at the output, and it is here that the dominant-pole compensation capacitor usually connects. An output stage then provides current gain at near-unity voltage gain to drive the low-impedance load. This topology, refined over decades, underlies most discrete and integrated audio power amplifiers.

Single-Ended and Push-Pull

A single-ended stage uses one active device referenced to a supply rail or to ground through a load, a current source, or a transformer. Single-ended Class A stages are prized in some audiophile and instrument amplifiers for their simplicity and even-order harmonic character. A push-pull stage uses two devices working in opposition, one sourcing current into the load and the other sinking it. Push-pull operation cancels even-order distortion and the quiescent components of supply current, and it is essential to the efficiency of Class B and Class AB.

Bridged and Balanced Configurations

A bridge-tied load configuration drives a loudspeaker from two amplifier channels operating in antiphase, doubling the voltage swing across the load and quadrupling the available power into a given impedance for the same supply rails. Bridging is common in automotive and portable equipment, where the supply voltage is limited. Fully balanced and differential topologies carry the signal as a pair of opposite-polarity paths throughout, improving immunity to supply noise and ground disturbances at the cost of duplicated circuitry.

Integrated and Discrete Realizations

Designers may build an amplifier from discrete transistors or select a monolithic power amplifier or operational amplifier. Integrated audio amplifiers package the input, voltage, and output stages, along with protection circuitry, on a single die, simplifying design and improving consistency. Discrete designs offer freedom in device selection, supply voltage, and output power, and they remain common in high-power and high-end equipment.

Gain Stages

Voltage gain in an audio amplifier is developed in the early stages, where signal currents are small and linearity is easiest to achieve. The output stage that follows generally provides little or no voltage gain.

The Input and Differential Stage

A differential pair forms the input of most power amplifiers. It accepts the signal at one input and the feedback at the other, amplifying their difference while rejecting noise common to both. Degeneration resistors in the emitters or sources linearize the pair and set its transconductance. A current mirror converts the differential output to a single-ended current, doubling the effective gain. The quiescent current, device matching, and choice of bipolar or field-effect input devices govern the input noise, offset, and bias current of the whole amplifier.

The Voltage-Amplifier Stage

The voltage-amplifier stage supplies most of the open-loop voltage gain, often several thousand. It typically consists of a common-emitter or common-source transistor loaded by a current source to obtain high gain, sometimes arranged as a cascode to improve linearity and bandwidth. Because this node swings nearly the full output voltage, its linearity strongly influences the distortion of the finished amplifier. The compensation capacitor that sets the dominant pole usually bridges this stage, exploiting the Miller effect to place a controlled, stable roll-off.

Gain Distribution and Stability

The total open-loop gain, the product of the gains of the stages, must be large enough that negative feedback can reduce distortion and stabilize the closed-loop gain, yet shaped so that the loop remains stable. Frequency compensation tailors how the open-loop gain falls with frequency so that the gain crosses unity with adequate phase margin. The distribution of gain among the stages, and the placement of poles and zeros, are therefore inseparable from the feedback design discussed below.

Output Stages and Biasing

The output stage delivers current to the load. It is the source of crossover distortion, the largest contributor to heat, and the part of the amplifier most exposed to the hazards of reactive and low-impedance loads.

Emitter-Follower and Source-Follower Outputs

Most output stages use complementary devices in a follower configuration, which presents a low output impedance and a voltage gain near unity. An emitter follower built from bipolar transistors offers high transconductance and predictable forward voltage, while a source follower built from power field-effect transistors offers freedom from secondary breakdown and a simpler drive at high frequency. Compound arrangements such as the Darlington pair and the complementary-feedback, or Sziklai, pair increase current gain so that the driver stage need supply only a modest current.

Setting the Quiescent Bias

The quiescent current of a Class AB output stage must be set carefully. Too little bias leaves residual crossover distortion; too much wastes power and risks thermal runaway. A bias spreader, commonly a transistor connected as an adjustable voltage source and known as a V_BE multiplier, develops the small voltage that holds both output devices slightly conducting at idle. Mounting the spreader transistor on the output heat sink lets it track the temperature of the output devices and compensate for the fall of their base-emitter voltage as they warm, stabilizing the bias against temperature.

Thermal Stability and Runaway

Bipolar output transistors carry a positive thermal feedback risk. As junction temperature rises, the base-emitter voltage needed for a given current falls, so a fixed bias drives more current, which raises temperature further. Without compensation this loop can lead to thermal runaway and device destruction. Emitter ballast resistors of a fraction of an ohm provide local negative feedback that equalizes current sharing among paralleled devices and damps the runaway tendency. Power field-effect transistors with a negative temperature coefficient of drain current at high currents are inherently more tolerant, although their behavior near the bias point still requires attention.

Output Protection

An output stage must survive short circuits, reactive loads, and excessive drive. Safe-operating-area protection limits the simultaneous voltage and current applied to the output devices, often through circuits that sense output current and reduce drive when the load impedance falls too low. A series Zobel network and an output inductor stabilize the amplifier against capacitive loads and the reactance of loudspeaker cables. These measures preserve reliability, though aggressive protection can itself introduce distortion if it intrudes during normal operation.

Feedback and Total Harmonic Distortion

Negative feedback is the principal tool for reducing distortion and stabilizing gain in linear audio amplifiers. A fraction of the output is subtracted from the input, so the amplifier corrects its own errors. Understanding both its power and its limits is central to good design.

How Negative Feedback Reduces Distortion

Applying negative feedback reduces the closed-loop gain and, in the same proportion, reduces nonlinear distortion, output impedance, and the sensitivity of gain to device parameters and supply variation. The amount of correction equals the loop gain, the open-loop gain multiplied by the feedback fraction. A large loop gain across the audio band can suppress harmonic and intermodulation distortion by orders of magnitude. The feedback network, usually a resistive divider from the output to the inverting input, sets the closed-loop gain precisely from stable passive components.

Measuring Distortion

Total harmonic distortion expresses the fidelity of an amplifier as the ratio of the combined power of the harmonics generated from a pure sine wave to the power of the fundamental, usually as a percentage or in decibels. When measured together with the residual noise, it is reported as total harmonic distortion plus noise. Intermodulation distortion, produced when two or more tones mix through a nonlinearity to create sum and difference frequencies, often correlates better with audible degradation because its products are not harmonically related to the signal. Competent audio power amplifiers achieve total harmonic distortion well below one tenth of one percent across the audio band.

Stability and Compensation

Feedback can cause oscillation if the loop gain still exceeds unity where the accumulated phase shift approaches one hundred eighty degrees. Frequency compensation, most often a dominant-pole capacitor across the voltage-amplifier stage, shapes the open-loop response so that the gain crosses unity with sufficient phase margin. The trade-off is bandwidth and slew rate: heavy compensation that guarantees stability also limits how fast the output can change. Slew-rate limiting produces its own form of distortion on fast transients, so the designer balances the compensation against the need for speed.

The Debate Over Feedback Depth

Audio engineering holds a long-running discussion about how much global feedback is desirable. Large amounts of feedback lower steady-state distortion figures dramatically but can convert benign low-order harmonics into a spray of higher-order products if the open-loop amplifier is poorly behaved, and they demand careful compensation. Some designers favor highly linear open-loop stages with modest feedback, while others apply deep feedback to well-compensated, wide-bandwidth circuits. Both philosophies can yield excellent results when executed with care.

Power Supply and Power-Supply Rejection

An audio amplifier reproduces, at its output, the difference between its input and a sample of its output; anything that corrupts the supply rails can leak into that output. The power supply is therefore an integral part of the amplifier rather than a mere accessory.

Supply Requirements

A linear power amplifier draws current from its rails in step with the audio signal, so the supply must deliver large, rapidly varying currents while holding its voltage reasonably constant. Bulk reservoir capacitance supplies the transient demand of musical peaks, while the rectifier and transformer, or a switching pre-regulator, replenish the average current. Insufficient capacitance or excessive supply impedance causes the rails to sag on loud passages, clipping the output prematurely and modulating the sound. Dual rails referenced to a central ground allow the output to swing symmetrically about zero and permit direct coupling to the load.

Power-Supply Rejection Ratio

The power-supply rejection ratio quantifies how well an amplifier ignores variations on its supply rails, expressed as the ratio of a change in supply voltage to the resulting change referred to the input or output. Ripple from the rectifier at twice the line frequency, and noise from a switching supply, will appear as hum or whine at the output if rejection is inadequate. Differential input stages reject supply disturbances common to both rails, but the voltage-amplifier stage and the output stage are more exposed. Current-source loading, cascoding, careful grounding, and local regulation of the early stages all raise the rejection ratio. Because feedback reduces the effect of disturbances inside the loop, supply rejection generally worsens at high frequencies where the loop gain has fallen.

Grounding and Layout

The large return currents of the output stage share conductors with the sensitive input ground unless the layout prevents it. A star-grounding scheme, which returns each stage to a single reference point, keeps high-current paths from injecting voltage into low-level paths. Separating the input ground, the feedback ground, and the high-current supply ground, and joining them at one point, is among the most consequential decisions in achieving low hum and a clean noise floor.

Thermal Design

Every linear amplifier converts part of its supply power into heat, and that heat must be removed to keep junction temperatures within safe limits. Thermal design determines both the reliability and the achievable output power of an amplifier.

Power Dissipation in the Output Stage

The output devices dissipate the product of the voltage across them and the current through them. In Class AB, the worst-case dissipation does not occur at full output but at an intermediate level, often near one third to one half of the maximum output power into a resistive load, where both the device voltage and the device current are substantial. Reactive loudspeaker loads shift the voltage and current out of phase, increasing the instantaneous dissipation and stressing the output devices beyond what a resistive load of the same magnitude would suggest. Thermal design must account for these worst cases, not merely the average.

Heat Sinks and Thermal Resistance

Heat flows from the transistor junction to the case, through the insulating interface to the heat sink, and from the heat sink to the surrounding air, each step characterized by a thermal resistance in degrees per watt. The designer sizes the heat sink so that the sum of these resistances, multiplied by the worst-case dissipation, keeps the junction below its rated maximum at the highest expected ambient temperature, with margin for reliability. Thermal interface material reduces the resistance between case and sink, while an electrically insulating but thermally conducting washer isolates the device collector or drain from the grounded sink where required.

Thermal Tracking and Class D Considerations

As noted in the discussion of biasing, a temperature-sensing element on the heat sink lets the bias voltage track the output devices and prevent thermal runaway. Class D amplifiers dissipate far less heat than linear designs, but their switching devices still produce conduction and switching losses that require modest heat sinking, and the output filter inductor must be sized to avoid saturation and excessive core loss. Forced-air cooling and integrated thermal-shutdown circuits extend the safe operating envelope of high-power amplifiers.

Speaker Loads

The loudspeaker is a reactive electromechanical load whose behavior departs sharply from the resistor often assumed in textbook calculations. Designing an amplifier to drive real loudspeakers requires understanding that load.

Impedance Is Not Resistance

A loudspeaker is rated by a nominal impedance, commonly four, six, or eight ohms, but its actual impedance varies strongly with frequency. It rises to a peak at the driver resonance, dips between resonances, and climbs again at high frequency because of the voice-coil inductance. The phase angle between voltage and current swings positive and negative across the band. An amplifier must remain stable and deliver adequate current into this varying, partly inductive and partly capacitive load, including the regions where the impedance dips well below the nominal value and demands extra current.

Damping Factor and Output Impedance

The damping factor is the ratio of the load impedance to the amplifier output impedance. A high damping factor means a low output impedance, which lets the amplifier control the motion of the loudspeaker cone by absorbing the back electromotive force the moving coil generates, tightening transient response in the bass. Negative feedback lowers the output impedance and so raises the damping factor. Beyond a moderate value, however, further increases yield diminishing audible benefit because the resistance of the voice coil and the loudspeaker cable dominate the total series resistance.

Power Delivery and Clipping

The power an amplifier can deliver depends on both its voltage swing and its current capability into the actual load impedance. Halving the load impedance doubles the current required for a given voltage, so an amplifier rated for a certain power into eight ohms may not double that power into four ohms unless its supply and output devices can sustain the extra current. When the demanded output exceeds the available rail voltage, the amplifier clips, flattening the peaks of the waveform and generating high-order harmonics. Sustained clipping not only sounds harsh but can damage high-frequency drivers by directing excessive average power into them.

Output Networks and Cable Effects

A small series inductor isolates the amplifier from the capacitance of long loudspeaker cables and from electrostatic loudspeakers, preserving stability, while a parallel Zobel network presents a controlled resistive load at high frequency. The resistance and inductance of the cable add to the source impedance seen by the loudspeaker and slightly reduce the effective damping factor, which is why heavy-gauge, short cable runs are preferred for low-impedance loads.

Summary

Audio amplifier design is the art of delivering power to a loudspeaker without disturbing the shape of the signal. The operating class sets the fundamental trade between efficiency and linearity: Class A is the most linear and least efficient, Class B the most efficient linear option but burdened by crossover distortion, Class AB the practical compromise for loudspeaker amplifiers, and Class D the efficient switching alternative now widespread in portable and high-power equipment.

Within a chosen class, the three-stage architecture of differential input, voltage-amplifier, and output stages organizes the gain so that linearity is achieved where signals are small and current is delivered where the load demands it. Careful biasing of the output stage suppresses crossover distortion and guards against thermal runaway, negative feedback reduces distortion and output impedance at the price of careful compensation, and a robust power supply with high rejection keeps rail disturbances out of the signal. Sound thermal design and a clear understanding of the reactive loudspeaker load complete the picture. Mastery of these interlocking disciplines, rather than any single circuit trick, distinguishes an amplifier that measures and sounds well from one that does not.