The Fundamentals of Doherty Amplifier Architecture

The Doherty amplifier, first conceived by William H. Doherty in 1936 at Bell Telephone Laboratories, has undergone a remarkable renaissance in modern wireless infrastructure. Originally developed for high-power broadcast transmitters, this architecture now forms the backbone of efficiency-enhancing circuitry in 4G and 5G base station power amplifiers. The core premise of the Doherty amplifier is elegant: it uses two distinct amplifier cells—a main (carrier) amplifier and a peaking (auxiliary) amplifier—working in concert to dramatically improve efficiency over a wide dynamic range, especially at power levels significantly below peak.

Traditional Class AB amplifiers, while linear, suffer from a sharp drop in efficiency as the output power is backed off from the peak—precisely the operating condition that dominates modern cellular signals. The Doherty architecture overcomes this by ensuring that at low power levels only the main amplifier is active, operating at or near its peak efficiency point. As the signal envelope increases, the peaking amplifier turns on, and together they deliver the necessary peak power while maintaining high efficiency across the entire range. This ability to maintain drain efficiency above 40–50% even at 6–8 dB back-off makes the Doherty amplifier indispensable for base stations handling high peak-to-average power ratio (PAPR) signals.

Main Versus Peaking Amplifier

The main amplifier is typically biased in Class AB (or sometimes Class B) to provide good linearity at low power levels. It is designed to handle the majority of the signal power during normal operation. The peaking amplifier, on the other hand, is biased in Class C so that it remains off (or nearly off) for low input levels. When the input signal exceeds a threshold (typically 6 dB below peak power), the peaking amplifier turns on and contributes to the output. The precise turn-on point and the gain characteristics of the two amplifiers are critical to achieving both high efficiency and linearity.

The Impedance Inverting Network

Central to the Doherty amplifier's operation is the impedance inverting network, usually a quarter-wave transmission line placed at the output of the main amplifier. This network transforms the impedance seen by the main amplifier in such a way that at low power (peaking off), the main amplifier sees an impedance that maximizes its efficiency. As the peaking amplifier turns on and delivers current, the impedance seen by the main amplifier changes, allowing both amplifiers to deliver power into the load efficiently. The combiner network must be carefully designed to maintain proper load modulation at all power levels. Advanced Doherty implementations often use offset lines, lumped-element equivalents, or multi-section transformers to achieve broader bandwidth and better performance.

Why Doherty Amplifiers Are Essential for 4G and 5G Networks

The transition from 3G to 4G LTE and now to 5G NR brought about a fundamental shift in the nature of the transmitted signals. Early cellular standards used constant-envelope modulations (e.g., GSM's GMSK) that allowed power amplifiers to operate near saturation with good efficiency. Modern orthogonal frequency division multiplexing (OFDM) waveforms, used in LTE and 5G, produce signals with high peak-to-average power ratios—often in the range of 7–12 dB. To avoid signal distortion and adjacent channel interference, the power amplifier must remain linear over this large dynamic range. Without efficiency-enhancing techniques like Doherty, a conventional Class AB amplifier would waste more than 80% of the DC power as heat when operating at 6 dB back-off. In a base station consuming several hundred watts of RF power, that equates to tremendous energy waste and cooling challenges.

High Peak-to-Average Power Ratio

5G NR signals, especially those using 256-QAM or higher modulation orders, exhibit PAPRs of 8–12 dB. The Doherty amplifier's ability to maintain high efficiency at 6–10 dB back-off directly addresses this requirement. In many macrocell deployments, the average output power is between 20–40% of the peak power, which corresponds to the region where the Doherty amplifier operates most efficiently. Field data from major infrastructure vendors shows that replacing traditional Class AB final stages with Doherty architectures yields 30–50% reduction in power consumption per base station, translating into significant operational expenditure savings for mobile network operators.

Efficiency at Back-Off Power Levels

Efficiency at back-off is the defining metric for base station power amplifiers. The Doherty architecture achieves a theoretical efficiency of 78.5% at full power and still retains 60–65% efficiency at 6 dB back-off, far exceeding a Class AB amplifier's typical 30–35% at the same back-off. This is accomplished through load modulation: as the peaking amplifier turns on, the effective load impedance of the main amplifier is modulated downward, allowing both amplifiers to operate at high efficiency over a wide range. Modern GaN-based Doherty implementations have demonstrated drain efficiencies exceeding 70% at 8 dB back-off for 5G NR waveforms.

Linearity Requirements

While efficiency is critical, signal linearity cannot be compromised. The Doherty amplifier inherently introduces some nonlinearity due to the switching behavior of the peaking amplifier and gain mismatch between the two paths. However, with careful design and the use of advanced linearization techniques such as digital predistortion (DPD), Doherty amplifiers can achieve the spectral purity required for 5G emissions masks. In fact, most commercial 5G base stations combine Doherty final stages with DPD implemented in field-programmable gate arrays (FPGAs) or application-specific integrated circuits (ASICs) to achieve both high efficiency and stringent linearity specifications.

Implementation Challenges in Modern Base Stations

Despite its widespread adoption, the Doherty amplifier presents a set of technical challenges that RF engineers must navigate when designing for 4G and 5G base stations. These challenges become more acute as operating frequencies rise into the millimeter-wave bands for 5G and beyond.

Bandwidth Limitations

Traditional Doherty amplifiers rely on quarter-wave impedance inverters, which are inherently narrowband devices. For 4G LTE bands that span 70–100 MHz, this is acceptable, but for 5G sub-6 GHz bands that may cover 200–400 MHz or more, the bandwidth limitation becomes a significant constraint. Engineers have developed several techniques to extend the bandwidth of Doherty amplifiers, including the use of multiple offset lines, stepped-impedance transformers, and novel combining networks. Recent research has demonstrated Doherty amplifiers operating over 600 MHz of instantaneous bandwidth using coupled-line structures and advanced substrate materials.

Combining Network Complexity

The combiner network must simultaneously provide impedance inversion, proper phase alignment, and low insertion loss. Any deviation results in degraded efficiency or linearity. At millimeter-wave frequencies (e.g., 28 GHz, 39 GHz for 5G), the quarter-wave transmission lines become extremely small—on the order of a few millimeters—and their parasitic effects become significant. This demands extremely tight fabrication tolerances and the use of advanced packaging techniques. Some designers have turned to on-chip Doherty implementations using silicon-based processes (SiGe BiCMOS, 45-nm RFSOI) to co-integrate the amplifiers and combiner network, though the lower quality factor of on-chip passives limits efficiency compared to GaN-based discrete solutions.

Thermal Management

Even with 70% drain efficiency, a 200–300 W peak power Doherty amplifier still dissipates substantial heat. The main amplifier, which operates continuously, experiences different thermal stress than the peaking amplifier, which only activates during high-power bursts. Thermal cycling and uneven temperature distribution across the transistors can degrade reliability and linearity. Base station designs must incorporate advanced thermal solutions such as vapor chambers, liquid cooling, and thermal interface materials optimized for high-power GaN devices. The physical layout of the Doherty cells must also consider thermal coupling between the two amplifiers to prevent one from heating the other excessively.

Advanced Doherty Architectures and Mitigation Techniques

To overcome the inherent limitations of the classic Doherty topology, researchers and industry have developed a host of advanced variations. These innovations allow Doherty amplifiers to meet the demanding specifications of 4G and 5G base stations while maintaining competitive cost and size.

Asymmetric and Multistage Doherty

In a symmetric Doherty amplifier, the main and peaking amplifiers have the same device size and output power capability. However, the load modulation effect is most efficient when the peaking amplifier is sized larger than the main amplifier, typically 1.5 to 2 times larger. This asymmetric Doherty configuration extends the high-efficiency range to deeper back-off levels—often 8–12 dB—making it ideal for the extreme PAPR of 5G. Multistage Doherty amplifiers add one or more additional peaking stages, each turning on at successively higher input levels, further flattening the efficiency curve. While more complex to design and bias, these multistage architectures can achieve 40% efficiency at 12 dB back-off, a remarkable feat.

Digital Predistortion

Digital predistortion (DPD) has become an essential companion to Doherty amplifiers in base stations. DPD works by creating an inverse model of the amplifier's nonlinear behavior and preprocessing the baseband signal to compensate for the distortion introduced by the power amplifier. Modern DPD systems can correct for AM-AM and AM-PM distortion, memory effects, and even cross-modulation in MIMO configurations. When combined with a well-designed Doherty amplifier, DPD enables adjacent channel leakage ratio (ACLR) values of –55 dBc or better, meeting the strictest regulatory requirements. Adaptive DPD algorithms can track temperature and aging drift, maintaining performance over the life of the base station. For further reading on DPD techniques, see the Analog Devices technical article on DPD for 5G base stations.

Use of GaN and Other Wide Bandgap Semiconductors

The adoption of gallium nitride (GaN) high-electron-mobility transistors (HEMTs) has been a major enabler for modern Doherty amplifiers. GaN offers high breakdown voltage, high current density, and low parasitic capacitance, allowing Doherty designs to achieve high output power (100–500 W) at high frequencies with excellent efficiency. The wide bandgap also permits operation at higher junction temperatures, reducing thermal management complexity. Other wide bandgap materials such as silicon carbide (SiC) are used in some high-power designs, though SiC generally lacks the RF performance of GaN. The combination of GaN devices with Doherty architecture has become the gold standard for macrocell and massive MIMO base stations. Companies like Qorvo offer GaN-on-SiC Doherty power amplifiers specifically optimized for 5G infrastructure.

As the wireless industry moves toward 5G Advanced and eventually 6G, the Doherty amplifier architecture will likely continue to evolve, integrating with other efficiency-enhancing techniques and adapting to new frequency regimes.

Integration with Envelope Tracking

Envelope tracking (ET) is a separate technique that modulates the supply voltage of the power amplifier in real time to follow the envelope of the RF signal. When combined with Doherty amplification, ET can further improve efficiency, particularly at deep back-off levels. However, the Doherty architecture already provides excellent back-off efficiency, so the incremental benefit of ET is smaller. Some researchers have proposed hybrid ETPA (envelope tracking Doherty) systems where the main amplifier supply is tracked while the peaking amplifier is supplied from a fixed voltage, achieving over 75% efficiency across a wide power range. Such systems require very fast, high-bandwidth envelope modulators, which are becoming feasible with GaN switches and advanced digital control.

Doherty in Massive MIMO and Beamforming

Massive MIMO base stations use dozens or even hundreds of antenna elements, each with its own transceiver chain. Power amplifiers in massive MIMO systems must be highly integrated and low cost, often leading to the use of highly efficient Doherty amplifiers in the transmitter path. With the trend toward millimeter-wave operation, CMOS-based Doherty amplifiers integrated directly into beamformer chips offer a promising path to 5G radios at 28 GHz and 39 GHz. For example, Keysight Technologies has published design methodologies for Doherty amplifiers at millimeter-wave frequencies, showing that careful electromagnetic simulation and optimized device layout can achieve 40% efficiency at 28 GHz with output powers exceeding 30 dBm per element.

Towards 6G

6G is expected to push into the sub-THz range (100–300 GHz) and require even wider bandwidths (multiple gigahertz). At these frequencies, the Doherty architecture faces fundamental challenges because the quarter-wave impedance inverters become extremely short and lossy. New topologies based on distributed amplification, traveling-wave Doherty, and the use of high-epsilon dielectric materials may emerge. Additionally, the massive bandwidths required for 6G may shift the focus from analog Doherty techniques to fully digital beamforming with highly efficient, linear, wideband amplifiers that do not rely on load modulation. Nevertheless, the underlying principle—efficiently handling high-PAPR signals—will remain relevant, and variations of the Doherty concept will likely persist in hybrid forms.

Conclusion

The Doherty amplifier architecture has proven to be one of the most impactful innovations in RF power amplification for cellular infrastructure. By intelligently combining two amplifiers through load modulation, it achieves unparalleled efficiency at the back-off power levels that dominate 4G and 5G signals. While challenges such as bandwidth limitations and design complexity remain, advanced techniques like asymmetric topologies, digital predistortion, and GaN transistor technology have made Doherty the default choice for high-power base stations. As wireless systems continue to evolve toward wider bandwidths, higher frequencies, and massive MIMO arrays, the Doherty architecture will continue to adapt, ensuring that network operators can deliver the data rate and coverage demanded by subscribers while controlling energy costs. For engineers designing the next generation of base station power amplifiers, a thorough understanding of Doherty principles and their practical implementation remains essential.