From "Class Act: The Pros and Cons of Amplifier Designs"
For convenience, engineers usually characterize amplifiers by circuit topology, the type of active components they use, load type, and even operating voltage: CLASS A, B, C, D, etc. Circuit topology describes how current is "steered," or controlled, within a power amplifier before it is delivered to a speaker load.
Class A is the simplest, most basic topology. Reproduction of music requires speaker motion both in and out. To do this, amplifiers must "source and sink" current. In a Class A amp only one direction of current control is used. To generate both directions of output flow, a constant current stage is subtracted from a variable current stage.
Pros: Since neither output stage ever turns off, device non-linearity and turn-on/turn-off time can be minimized or ignored, resulting in very low distortion designs.
Cons: As the maximum output is limited to the constant current stage, at idle this stage must put out full power and the variable stage must absorb this full power. Transformer, heat sink, and output stage must be sized for continuous duty at maximum power. Because of cost and the amount of waste heat generated by this approach, Class A only appeals to esoteric hi-fi designers, where lack of efficiency or price is no object.
Class B topology uses two variable output stages, one to source current and the other to sink current.
Pros: This topology overcomes the poor efficiency of pure Class A designs, only delivering power as needed. Transformer and heatsinks can be sized to match typical demands of music being reproduced.
Cons: Both output stages turn completely off, then on again, during each cycle of a waveform. Time delay and low-level non-linearities cause severe distortion, called "crossover distortion," during transition from source to sink output stages. This type of distortion is worst at low output levels. Pure Class B is only used in the lowest-cost, lowest-fidelity designs.
Class A/B topology, as you may have guessed, is a combination of Class A and Class B. Using two variable output stages like Class B but keeping them from ever completely turning off, you get near Class B efficiency with near Class A's low-distortion performance.
Class C topology combines active devices with resonant magnetic components for high efficiency at radio frequencies. This topology is not used in audio-frequency designs.
Class D topology uses source and sink output stages that consist of full-on or full-off switches. These output stages toggle from full sink to full source at a rate significantly higher than the highest audio frequency to be reproduced. The ratio of time sinking to time sourcing controls the audio output, with a 50% ratio delivering zero output.
Pros: Class D offers significantly higher efficiency than even Class B, which at 1/3 power is wasting more power inside the amplifier than it delivers to the load. Losses in Class D designs are limited to turn-on time of the switching devices and resistive losses in these devices and output filtering.
Cons: Class D amps require more complex circuit designs with extensive shielding and filtering.
Class G & Class H topologies are variations on Class B that use multiple source and sink output stages. Low-level signals are handled by one pair of output stages, while higher-level signals are handled by other pairs. Each pair is optimized for the power range it delivers.
Pros: More efficient amplifiers can deliver the same output power with smaller transformers and less heat sink.
Cons: Circuit complexity increases, which adds cost. Switching distortion similar to Class B's crossover distortion occurs at each output level transition.