Class H Amplifier
A Class H amplifier takes Class G's rail-switching idea to its limit by replacing discrete rail steps with a supply rail that continuously tracks the signal envelope, riding just above the output waveform at every instant rather than jumping between two or three fixed voltages, which squeezes out most of the dissipation that Class G still wastes between its rail steps and approaches a theoretical efficiency much closer to Class D's — while keeping Class AB's linear output stage and avoiding switching-frequency output filtering altogether.
The difference from Class G is continuous versus discrete, and that distinction changes the whole supply design. Class G switches abruptly between two or three fixed rail voltages, so a signal sitting just below the low-rail threshold still wastes dissipation against that fixed low rail; Class H's rail is itself a tracking power supply — typically a fast switch-mode converter whose reference input is the signal envelope — so the rail height follows the waveform smoothly and the voltage drop across the output devices stays close to a small, nearly constant margin at every signal level, not just at the handful of levels Class G's fixed rails happen to land on.
The tracking supply's bandwidth, not the output stage, becomes the hard design problem. Because the rail has to rise and fall fast enough to stay just ahead of the audio envelope without ever dipping below it (which would clip the output) or lagging behind it (which would waste the efficiency gain), the switch-mode tracking regulator needs a control bandwidth well beyond what a conventional point-of-load converter needs — this is the same envelope-tracking control problem that RF power amplifier designers solve in cellular base stations and handsets, where envelope tracking lets a linear RF PA run near its peak efficiency across the entire signal envelope instead of only at full power.
It still needs real headroom margin, which is the one thing it can't track away. A tracking supply that followed the signal exactly with zero margin would clip the instant the real signal got even slightly ahead of the rail's response time, so practical Class H designs keep a deliberate margin between the rail and the instantaneous output — tens to low hundreds of millivolts beyond what Class AB alone would need — which means Class H's efficiency gain comes from shrinking that margin as much as the tracking loop's speed allows, not from eliminating it.
Class H ended up a transitional technology rather than the long-term winner, and understanding why ties the whole amplifier-class story together. Despite closing much of the efficiency gap to Class D, Class H never fully displaced it in most high-volume audio products because the tracking supply adds real complexity, cost, and its own switching noise that has to be managed almost as carefully as a straight Class D design's output filter — so once Class D's measured audio performance improved enough through better feedback and control loops, most designs chose outright switching over the more delicate task of building an envelope-tracking linear amplifier.
Read a Class H amplifier through a *rail-as-a-moving-target* lens rather than a *rail-switching* lens: where Class G accepts a small number of discrete efficiency levels, Class H commits to continuously re-aiming the supply at the signal itself, trading a simple multi-rail switch for a fast tracking power converter — the efficiency ceiling moves closer to Class D, but so does the engineering difficulty, which is exactly why it ended up as a bridge technology between linear and switching amplification rather than the final answer.