Ok, let's look at the Class D topology.

The basic circuit layout is the same as Class B for all intensive purposes. What is different is the signal that is fed into the output devices.

Instead of feeding these devices an analog musical waveform, the audio is fed into a circuit called a Pulse Width Modulator.

The whole theory behind this topology relies on how slow a speaker reacts to a signal. You see, if you send a very brief voltage spike to a sub, the weigh of the cone will not allow it to move very far, if at all. If you send a series of these pulses together, the speaker will move just a bit. If the pulses are a bit wider, we get more motion from the cone...
 


This simple graph shows two waveforms. If we take the center line as 0 volts, the blue line represents the signal sent to the top transistor, and the yellow to the bottom transistor. Now, these small spikes get amplified, and fed to the speaker. They move very fast. Each spike lasts less that 1/100,000th of a second. The spike is at a high level for 1/4 of the duration of the cycle. If we average out all these spikes, we get a voltage that is at 1/4 of the maximum level. Are you catching on yet?
 


In this graph, the frequency is the same, but the signal is 'high' for half the cycle. This results in an average voltage of half of the maximum.
 


And lastly, the same frequency again, only a 75% duty cycle, that is, the percentage of the cycle that the signal is high. We get 3/4 of the maximum available voltage.

You can see that if we combine different pulse widths, we can recreate a waveform of almost any shape. Bingo!

If we switch the output devices all the way and back off again very quickly, and do so at different rates, we can get an audio waveform to be produced by the speaker.
 


In reality, the pulses are fed to the output devices at a frequency between 30 and 50 kHz, that is, 30 to 50 thousand pulses per second. This is required to produce a smooth waveform to the speaker.

The incredible advantage to this topology is that the output devices are either all the way on, or all the way off. They are never half way on, and acting like resistors. When in the resistive region (anywhere between completely off or on), they heat up. Anything that heat up wastes energy in the form of heat.

In application, these amplifiers have the highest overall efficient. Varying between 70% and 80% in 1/3 power tests. Class B amps come in around 30% to 40% efficient. Class A amps are the worst of the bunch, coming in at 10-15% efficient...
 

TopologyTopology efficiencyAmp EfficiencyBenefitDrawback
Class A25%15%Low DistortionInefficient
Class B75%35%EfficientCrossover Distortion
Class D95+%75%Most EfficientNoisy, for subs only.

Hopefully that helps to demystify what the different topologies mean.