class e amplifier
A Class E amplifier uses a single switching device driving a carefully shaped reactive network so the voltage across that device is forced to zero, and its slope also near zero, at the exact instant it turns back on — eliminating switching loss almost entirely rather than just minimizing it, which pushes theoretical efficiency toward 100% and makes Class E the design of choice anywhere RF power has to be generated at megahertz-to-gigahertz switching rates where Class D's hard-switching losses would otherwise dominate.
Zero-voltage switching is the entire point, and it is a waveform-shaping problem, not a device problem. Hard-switched topologies like Class D turn a device on while voltage is still present across it, dumping the energy stored in that voltage into heat every single cycle; Class E's shunt capacitor and series-tuned LC network are sized specifically so the voltage waveform across the switch rings down to zero — and arrives there with zero slope — right as the drive signal commands it back on, so there is effectively nothing left to dissipate at the transition, which is a fundamentally different efficiency mechanism than simply switching faster or using a lower-resistance device.
One switch is enough because the reactive network does the work a second device would otherwise do. Class D needs two complementary devices to pull the output both high and low; Class E gets away with a single switch plus a passive network because the network's own ringing supplies the return swing, which simplifies the power stage and removes the shoot-through risk that complementary switching topologies have to guard against, at the cost of a design that is far more sensitive to its exact component values and load impedance than a push-pull stage.
It only behaves correctly at one frequency and one load, which is the tradeoff nobody escapes. Because the zero-voltage condition depends on a precise relationship between the switching frequency, the shunt capacitance, and the series-tuned network's reactance, a Class E stage detuned by a mismatched antenna or a shifted operating frequency stops achieving zero-voltage switching and its efficiency collapses toward Class D or worse — this sensitivity is why Class E shows up in fixed-frequency applications like RFID readers, wireless power transfer coils (the same zero-voltage-switching principle underlies Qi charging and resonant wireless-power designs from companies like WiTricity), and dedicated ISM-band RF transmitters rather than general-purpose amplifiers.
The device still needs real voltage-rating headroom despite the lossless-sounding switching. Because the shunt capacitor rings the switch voltage up well above the supply rail before bringing it back to zero — commonly 3.5 times the DC supply voltage at the peak — Class E designs require power MOSFETs or GaN devices rated for that peak swing, not just the nominal supply, which is a specification trap that catches designers who only check average or RMS voltage.
Read a Class E amplifier through a *timing-as-efficiency* lens rather than a *device-as-efficiency* lens: nothing about the transistor itself changed, but the reactive network around it was tuned so the moment of switching coincides exactly with the moment voltage and its slope both reach zero, which converts a loss that every other switching class treats as unavoidable overhead into one that, at the right frequency and load, barely exists at all.