Inferensys

Glossary

Harmonic Termination

The intentional presentation of specific short-circuit or open-circuit impedances at harmonic frequencies to the transistor's intrinsic current source to shape voltage and current waveforms for enhanced efficiency.
Stylish WeWork-like workspace with hot desks and document wall, professional searching through enterprise knowledge base on a mounted ultrawide display, warm industrial pendants overhead.
WAVEFORM ENGINEERING

What is Harmonic Termination?

Harmonic termination is the deliberate engineering of specific impedance conditions at harmonic frequencies to sculpt transistor voltage and current waveforms for maximized efficiency.

Harmonic termination is the intentional presentation of precise short-circuit or open-circuit impedances at the second, third, and higher harmonic frequencies to the transistor's intrinsic current source plane. By controlling the phase and magnitude of reflected harmonic energy, designers reshape the overlapping voltage and current waveforms to minimize dissipated power, enabling theoretical efficiency approaching 100% in Class-F and inverse Class-F amplifier modes.

In a Doherty amplifier, harmonic termination networks are integrated into the output matching and combiner structures to suppress harmonic content while maintaining the correct fundamental-frequency load modulation. Proper termination at the current source plane—de-embedded from package parasitics—is critical for achieving the voltage peaking and current shaping required for high back-off efficiency in modern GaN HEMT designs.

Waveform Engineering

Key Characteristics of Harmonic Termination

Harmonic termination is a deliberate impedance control strategy that shapes the intrinsic transistor current and voltage waveforms at harmonic frequencies to maximize power-added efficiency and output power.

01

Class-F Harmonic Termination

Class-F operation uses harmonic resonators to shape waveforms toward ideal square-wave (voltage) and half-sine (current) shapes, minimizing overlap and theoretically achieving 100% drain efficiency.

  • Second harmonic: Short-circuit termination forces voltage to zero at 2f₀
  • Third harmonic: Open-circuit termination flattens voltage waveform at 3f₀
  • Odd harmonics: Present an open circuit to square the voltage
  • Even harmonics: Present a short circuit to shape the current

Practical implementations typically control only the 2nd and 3rd harmonics due to diminishing returns and matching network complexity at higher orders.

100%
Theoretical Max Efficiency
2nd-3rd
Practical Harmonic Control
02

Inverse Class-F Termination

Inverse Class-F (Class-F⁻¹) reverses the termination scheme: open circuit at the second harmonic and short circuit at the third harmonic. This produces a half-sinusoidal voltage waveform and a square-wave current waveform.

  • Advantage: Lower peak voltage stress on the transistor for the same output power
  • Benefit: Particularly suited for GaN HEMT devices with low knee voltage
  • Trade-off: Higher peak current requires robust current-handling capability
  • Efficiency: Matches Class-F theoretical maximum of 100%

The choice between Class-F and Inverse Class-F often depends on the specific transistor technology and its voltage versus current handling limitations.

GaN HEMT
Preferred Technology
03

Continuous Mode Harmonic Control

Continuous Class-B/J and Class-F modes expand the single-point optimum impedance to a continuous impedance space on the Smith chart, maintaining efficiency across a range of reactive harmonic terminations.

  • Class-J: Second harmonic termination varies from short to open with appropriate fundamental reactance
  • Benefit: Significantly broadens bandwidth compared to fixed harmonic shorts/opens
  • Mechanism: Reactive harmonic terminations introduce phase shifts that reshape waveforms while preserving the zero-voltage, zero-current switching condition
  • Application: Critical for wideband Doherty amplifiers where fixed harmonic traps limit bandwidth

This theory demonstrates that efficiency is maintained not at a single impedance point but along a continuous design trajectory.

> 50%
Fractional Bandwidth Possible
04

Harmonic Load-Pull Measurement

Harmonic load-pull is the systematic measurement technique used to characterize the optimal harmonic impedances for a specific transistor under large-signal drive conditions.

  • Fundamental load-pull: Varies Γ_L at f₀ to map power and efficiency contours
  • Harmonic load-pull: Actively controls Γ_L at 2f₀, 3f₀, and sometimes 4f₀
  • Active tuning: Uses injected signals to synthesize reflection coefficients > 1 at harmonic frequencies
  • Output: Identifies the optimum harmonic impedance constellation for maximum PAE

This data directly informs the synthesis of output matching networks that present the correct harmonic terminations at the transistor's intrinsic current source plane, not just at the package reference plane.

2f₀-4f₀
Typical Harmonic Sweep Range
05

De-Embedding to the Current Source Plane

A critical challenge in harmonic termination design is presenting the intended impedance at the intrinsic current source of the transistor die, not at the external package leads or connector.

  • Parasitic network: Transistor package parasitics (bond wires, lead frames) rotate impedances significantly at harmonic frequencies
  • De-embedding: Uses S-parameter models of the package and transistor parasitics to mathematically translate reference planes
  • Cds and Lout: The drain-source capacitance and output inductance form a low-pass filter that attenuates harmonic voltages at the external terminal
  • Design implication: The external matching network must pre-compensate for these parasitics to achieve the correct intrinsic harmonic termination

Failure to account for this plane shift results in suboptimal waveform shaping and degraded efficiency despite apparently correct external terminations.

Cds
Critical Parasitic Element
06

Harmonic Termination in Doherty Amplifiers

In Doherty architectures, harmonic termination interacts with the load modulation mechanism, requiring careful co-design of the combiner network to maintain proper harmonic control across the dynamic impedance range.

  • Carrier path: Harmonic terminations must remain effective as the load impedance varies from high-efficiency back-off to saturation
  • Peaking path: Harmonic terminations engage only when the peaking amplifier activates at high power
  • Combiner interaction: The impedance inverter at f₀ does not necessarily provide the correct transformation at 2f₀ or 3f₀
  • Solution: Post-matching topologies place harmonic control networks before the Doherty combiner, isolating harmonic shaping from load modulation effects

This co-design challenge is central to achieving high-efficiency Doherty operation across the full dynamic power range.

Post-Matching
Preferred Topology
HARMONIC TERMINATION

Frequently Asked Questions

Essential questions and answers about harmonic termination techniques for optimizing power amplifier efficiency and linearity in Doherty and other advanced architectures.

Harmonic termination is the intentional presentation of specific short-circuit or open-circuit impedances at harmonic frequencies to the transistor's intrinsic current source to shape voltage and current waveforms for enhanced efficiency. By controlling the impedance at the second harmonic (2f₀) and third harmonic (3f₀), designers can engineer the overlap between drain voltage and current waveforms. In a Class-F design, odd harmonics are terminated with open circuits and even harmonics with short circuits, squaring the voltage waveform and reducing dissipation. In inverse Class-F, the roles are reversed to flatten the current waveform. The technique works by preventing harmonic power from being dissipated in the transistor, instead reflecting it back to shape the fundamental-frequency waveforms for minimal overlap and maximum power-added efficiency (PAE).

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.