A Class F amplifier is a high-frequency RF power amplifier that uses harmonic tuning networks to shape voltage and current waveforms, minimizing their overlap to achieve theoretical efficiencies up to 100%. Unlike a standard Class AB amplifier that relies on a broadband 50-ohm matching network, a Class F topology fundamentally changes the output circuit by inserting multi-resonant harmonic traps. These traps force the drain (or collector) voltage into a flat-topped square wave while maintaining the drain current as a half-sine wave. Because power dissipation is the product of instantaneous voltage and current, eliminating the overlap between these two waveforms drastically reduces heat generation in the active device.
The Core Concept: Waveform Shaping in Class F Amplifiers
In a traditional linear amplifier (Class A, B, or AB), both voltage and current are sinusoidal. They overlap significantly during the conduction angle, meaning the transistor must dissipate the overlapping energy as heat. This limits practical efficiency to around 50-78%.
The Class F amplifier solves this by manipulating the harmonics generated by the transistor's non-linear switching action. By presenting specific impedances to the harmonics at the drain terminal, we sculpt the waveforms:
- Odd Harmonics (3rd, 5th, etc.): The network presents an open circuit at the drain. This reflects the odd harmonic currents back, flattening the peaks of the voltage waveform into a square shape.
- Even Harmonics (2nd, 4th, etc.): The network presents a short circuit at the drain. This clamps the even harmonic voltages to zero, ensuring the current waveform remains a clean half-sine without voltage overlap.
Do not confuse Class F with Class D. Class D also uses square-wave switching for high efficiency, but it relies on Pulse Width Modulation (PWM) and is strictly used for audio and low-frequency motor drives. Class F is exclusively an RF/microwave topology that relies on resonant LC tanks or transmission line stubs, not PWM. It is also distinct from Class E, which achieves Zero Voltage Switching (ZVS) using a single shunt capacitor and series inductor, rather than multiple harmonic traps.
Worked Example: 1 GHz Harmonic Trap Network Calculation
Let’s design the output harmonic control network for a Class F amplifier operating at a fundamental frequency ($f_0$) of 1 GHz into a 50-ohm load. To get the classic square voltage / half-sine current waveforms, we need to control the 2nd and 3rd harmonics.
1. Third Harmonic Open Circuit (3 GHz)
We need a parallel LC tank tuned to 3 GHz to act as an open circuit at the drain. Using the resonance formula $f = \frac{1}{2\pi\sqrt{LC}}$:
- Select a high-Q RF capacitor: $C_3 = 1.0 \text{ pF}$ (e.g., Murata GJM1555C1H1R0BB01).
- Calculate required inductance: $L_3 = \frac{1}{(2\pi \cdot 3\text{e}9)^2 \cdot 1\text{e-12}} \approx 2.81 \text{ nH}$.
- Select inductor: Coilcraft 0402HP-2N8X (2.8 nH, high Q at 3 GHz).
2. Second Harmonic Short Circuit (2 GHz)
We need a series LC trap tuned to 2 GHz to act as a short circuit to ground.
- Select capacitor: $C_2 = 2.0 \text{ pF}$.
- Calculate inductance: $L_2 = \frac{1}{(2\pi \cdot 2\text{e}9)^2 \cdot 2\text{e-12}} \approx 3.16 \text{ nH}$.
- Select inductor: Coilcraft 0402HP-3N2X (3.2 nH).
On paper, lumped LC components work perfectly. On the bench, a standard 0402 MLCC has about 0.5 nH of Equivalent Series Inductance (ESL), and the PCB pads add another 0.2 nH. At 3 GHz, these parasitics will severely detune your 2.81 nH inductor. For frequencies above 2 GHz, experienced RF engineers abandon lumped components for the 3rd harmonic open and instead use a quarter-wave microstrip open-circuit stub etched directly into the Rogers PCB substrate.
Where You Meet Class F in Practice
You will rarely find a Class F amplifier in consumer electronics or audio gear. This topology is strictly the domain of high-power, high-frequency RF transmission where thermal management and DC power consumption are critical constraints.
- Cellular Base Stations (4G/5G/6G): Massive MIMO arrays use dozens of small, highly efficient power amplifiers. Class F (and its inverse, Class F⁻¹) allows these arrays to run cooler, reducing the need for heavy heatsinks and lowering site power costs.
- AESA Radar Systems: Active Electronically Scanned Arrays in military and aviation radar rely on Gallium Nitride (GaN) HEMTs operating in Class F to maximize transmit power per module without melting the T/R (Transmit/Receive) modules.
- Satellite Downlinks: In space, DC power from solar panels is strictly limited, and dissipating heat in a vacuum is notoriously difficult. Class F efficiency is mandatory for high-bandwidth satellite transmitters.
For authoritative deep-dives into the waveform mathematics of these topologies, the Microwaves101 encyclopedia on Class F amplifiers provides excellent visual breakdowns of the drain voltage and current overlaps.
Common Confusions: Class F vs. Class E and Class J
When specifying an RF power stage, engineers frequently debate between Class E, Class F, and the newer Class J. Here is how they differ in practice:
| Feature | Class E | Class F | Class J (Continuous Class F) |
|---|---|---|---|
| Primary Mechanism | Zero Voltage Switching (ZVS) via shunt capacitance | Harmonic peaking (square V, half-sine I) | Reactive harmonic terminations (allows wider bandwidth) |
| Bandwidth | Narrow (< 10%) | Very Narrow (< 5%) | Broad (up to an octave) |
| Max Frequency | Usually < 1 GHz (limited by device $C_{oss}$) | 1 GHz to 10 GHz+ | 1 GHz to 10 GHz+ |
| Typical Device | LDMOS, GaAs | GaN HEMT | GaN HEMT |
Class F is extremely narrowband because the harmonic traps are highly resonant. If your application requires amplifying a wide band of frequencies (like a 5G carrier aggregation band spanning 100 MHz), pure Class F will fail. In that case, you transition to Class J, which intentionally introduces a reactive component to the fundamental and harmonic impedances, trading a tiny amount of peak efficiency for a massive increase in bandwidth. For more on modern GaN implementations, refer to the Qorvo RF Design Hub application notes on continuous-mode amplifiers.
Decision Path: Selecting Your RF Power Amplifier Topology
Use this decision tree to lock in your amplifier class and active device technology based on your specific RF requirements.
| If your requirement is... | Then choose this Topology | Recommended Active Device |
|---|---|---|
| Frequency < 100 MHz, maximum raw power, moderate efficiency | Class AB (Doherty configuration) | LDMOS (e.g., NXP MRFX600H) |
| Frequency < 500 MHz, switching power supply or RF heating | Class D or Class E | Si MOSFET or GaN (e.g., EPC eGaN FETs) |
| Frequency > 1 GHz, single narrowband carrier, max efficiency required | Class F | GaN HEMT (e.g., Qorvo TQP0230) |
| Frequency > 2 GHz, wide bandwidth (e.g., 5G NR), high efficiency needed | Class J (Continuous Class F) | GaN HEMT (e.g., Wolfspeed CGHV1J070D) |
If you are designing a modern, narrowband RF transmitter operating above 1 GHz and need to minimize thermal load, specify a Class F topology using a bare-die or packaged GaN HEMT like the Qorvo TQP0230. Do not attempt to use lumped 0402 components for the 3rd harmonic trap; design quarter-wave microstrip stubs on a Rogers RO4350B substrate to handle the parasitics cleanly.
Frequently Asked Questions
Can I use a Class F amplifier for audio?
No. Class F relies on harmonic resonance networks tuned to specific RF frequencies (typically >1 GHz). Audio frequencies (20 Hz - 20 kHz) are far too low for practical harmonic trap components, and the required inductors would be physically massive. Use Class D for high-efficiency audio.
What is an Inverse Class F (Class F⁻¹) amplifier?
Class F⁻¹ swaps the waveform shapes: it shapes the current into a square wave and the voltage into a half-sine wave. It requires a short circuit at odd harmonics and an open circuit at even harmonics. It is often preferred in modern GaN designs because the device's natural parasitic output capacitance ($C_{oss}$) makes creating an open circuit at the 2nd harmonic easier than creating a perfect short.
