Class E Power Amplifier Design: Theory, Math, and RF Bench Reality

A Class E power amplifier is a highly efficient tuned switching circuit where the transistor operates as a strict on/off switch and a specialized reactive output network ensures that drain voltage and current waveforms never overlap, theoretically achieving 100% efficiency. In a real RF installation, swapping a linear Class AB stage for a Class E stage eliminates the massive finned heatsinks and drops the DC current draw by more than half. This allows you to push 5W of continuous RF out of a tiny TO-220 MOSFET running off a standard 12V battery, fundamentally changing the thermal and power-supply requirements of your transmitter.

The Core Concept: Switching Without the Overlap

Linear amplifiers (Class A, B, AB) act like variable resistors. The transistor is partially on, meaning it has both voltage across it and current flowing through it simultaneously. Since Power Dissipation = Voltage × Current, this overlap generates massive amounts of heat. Class E abandons linear operation entirely. The transistor is driven hard into saturation (fully ON) or cutoff (fully OFF).

The magic of Class E lies in the output load network—a specific arrangement of a shunt capacitor, a series inductor, and a series capacitor. This network is mathematically tuned to shape the waveforms so that Zero Voltage Switching (ZVS) and Zero Derivative Switching (ZDS) occur. When the MOSFET turns ON, the voltage across its drain has already fallen to zero. When it turns OFF, the current through it has already dropped to zero.

Where You Meet Class E in Practice

You will not find Class E amplifiers driving your living room speakers; they are inherently narrowband due to the tuned output network. Instead, you meet them in applications where high RF efficiency is critical:

• Ham Radio QRP Transmitters: Popular kits like the QCX and uSDX use Class E final stages to squeeze 5W of RF from a 12V supply.
• Industrial RF Heating: Plastic welding and induction heating rely on high-power Class E (or Class D) topologies at 13.56 MHz or 27.12 MHz.
• NFC and RFID Readers: The 13.56 MHz driver stages in proximity card readers use low-power Class E to maximize magnetic field strength without draining the host device battery.
• MRI Machines: The RF transmit coils in medical imaging are often driven by high-power solid-state Class E modules.

The Math: A Worked 14 MHz Design Example

Let’s design the output network for a 5W Class E amplifier operating at 14.2 MHz (the 20-meter ham band) using a 12V DC supply. We rely on the classic Nathan Sokal design equations, which remain the gold standard for this topology (detailed further in The ARRL Handbook for Radio Communications).

1. Calculate the required load resistance ($R_{load}$):
   The switch needs to see a specific resistance to deliver the target power. The formula is $R_{load} = \frac{0.5768 \times V_{cc}^2}{P_{out}}$.
   $R_{load} = \frac{0.5768 \times 144}{5} = $ 16.6 Ω.
2. Transform the antenna impedance:
   Our antenna is 50 Ω, but the FET needs to see 16.6 Ω. We must design an L-match or pi-network to transform 50 Ω down to 16.6 Ω before the signal reaches the Class E network.
3. Calculate the shunt capacitor ($C_{shunt}$):
   This capacitor sits directly across the MOSFET drain and source. $C_{shunt} = \frac{0.1836}{2 \pi f R_{load}}$.
   $C_{shunt} = \frac{0.1836}{2 \pi \times 14,200,000 \times 16.6} \approx $ 124 pF.
4. Select the DC feed inductor ($L_{RFC}$):
   This RF choke must have a reactance at least 10 times higher than $R_{load}$ to prevent RF from leaking back into the power supply. At 14.2 MHz, a 33 µH toroidal inductor provides over 2,900 Ω of reactance, which is more than sufficient.

Bench War Story: When Parasitics Blow the FET

Theory assumes ideal components. The workbench does not. Here is a scenario that highlights why Class E design fails if you ignore datasheet parasitics.

The Setup: I was building a 50W Class E amplifier for 7.0 MHz (40m band) using a classic IRF510 MOSFET. The design called for a 50V drain supply.

The Numbers: Following the Sokal equations for 50W at 7 MHz, the math dictated an external shunt capacitor ($C_{shunt}$) of exactly 470 pF. I soldered in a 470 pF high-voltage silver mica capacitor.

The Outcome: I keyed the transmitter. The ammeter spiked to 8A, a sharp crack echoed from the bench, and the IRF510 gate popped, shorting internally. The FET was destroyed in milliseconds.

What Went Wrong: I ignored the MOSFET’s output capacitance ($C_{oss}$). The IRF510 datasheet lists $C_{oss}$ at roughly 360 pF (measured at 25V $V_{ds}$). At RF frequencies, this internal parasitic capacitance acts directly in parallel with my external 470 pF capacitor. The total shunt capacitance was actually 830 pF. This over-capacitance ruined the ZVS timing. When the gate drive turned the FET ON, the drain voltage hadn't finished falling yet; there was still about 30V across it. The resulting $V \times I$ overlap caused instantaneous thermal runaway.

The Fix: You must subtract the FET's $C_{oss}$ from your calculated $C_{shunt}$ value. Alternatively, for frequencies above 10 MHz, abandon the IRF510 and use a modern LDMOS device like the NXP MRF101AN, which offers much lower parasitic capacitance and higher gain.

Class E vs. Class D and Class F

People commonly confuse Class E with other switching topologies. Here is how they actually differ in practice.

If you are building a subwoofer amp, you want Class D. If you are building a 14 MHz ham radio transmitter, you want Class E. Class F is generally reserved for commercial engineers designing multi-stage base station amplifiers where controlling the 3rd harmonic is necessary to square off the voltage waveform for even higher efficiency.

Frequently Asked Questions