Doherty Amplifier
A Doherty power amplifier splits an RF signal between a main amplifier biased in class AB and a peaking amplifier biased in class C, combining them through a quarter-wave transformer so efficiency stays high across a 6-10dB power back-off range instead of collapsing the way a single class-AB stage does. William Doherty patented the idea at Bell Labs in 1936 for AM broadcast transmitters, and it sat mostly dormant for seven decades until LTE and 5G handed RF engineers a modulation problem it happens to solve almost perfectly: orthogonal frequency-division signals carry most of their energy well below peak power, so a power amplifier that is only efficient at peak power wastes most of the battery or wall power it draws.
The quarter-wave line is not decoration — it is the mechanism that lets the peaking amplifier boost the main amplifier's output impedance instead of just adding current to it. At low power only the main amplifier conducts, and the quarter-wave transformer presents it with a high impedance that keeps its voltage swing large relative to a small current, which is exactly the condition for high efficiency. As input power climbs toward peak, the peaking amplifier turns on and injects current that lowers the impedance the main amplifier sees, letting both devices share the load and saturate together at full power. Qorvo, Skyworks, and NXP ship exactly this topology inside base-station power amplifier modules sitting behind nearly every 4G and 5G macro cell, typically built in GaN-on-SiC for the main and peaking dies because GaN's high breakdown voltage tolerates the impedance swings the Doherty load line demands.
The payoff shows up directly on an efficiency-versus-output-power curve, which is the chart every RF engineer pulls up first when comparing amplifier architectures. A conventional class-AB amplifier peaks in efficiency only at its single rated output power and falls off fast below that, so a signal that spends most of its time 6-10dB below peak — true of nearly every modern digitally modulated waveform — runs the amplifier in its worst efficiency region most of the time. The Doherty architecture adds a second efficiency peak at the back-off point where the peaking amplifier hands off, so the curve stays above 50% efficiency across the whole back-off range instead of dropping toward 20%. Base-station vendors including Ericsson, Nokia, and Huawei depend on that second peak to hold 5G massive-MIMO radio power consumption and heat within what an outdoor cabinet can dissipate without active cooling.
Pushing the architecture further costs linearity, which is why nearly every Doherty amplifier ships behind a digital predistortion engine rather than running raw. The abrupt turn-on of the peaking device and the impedance-modulated load line both introduce AM-to-AM and AM-to-PM distortion that a fixed-gain amplifier does not have, so basebands from Analog Devices and Texas Instruments pair the RF front end with a DPD block that characterizes the amplifier's nonlinearity in real time and pre-warps the digital signal to cancel it before the DAC. Three-way and N-way Doherty variants stack additional peaking stages to stretch the second efficiency peak across a wider back-off range for the higher peak-to-average power ratios in 5G NR, trading circuit complexity and bandwidth for an even flatter efficiency curve. Read a Doherty amplifier's datasheet by its efficiency-at-back-off number, not its peak efficiency figure — peak efficiency is what any class-AB stage already delivers, and back-off efficiency is the number that actually determines a base station's power bill.