The EFHW Shunt Capacitor: A Double-Edged Sword

I do not object to a shunt capacitor. I object to treating a prettier SWR trace as its design proof. Used deliberately, the capacitor can compensate part of a transformer-and-load response. Chosen by folklore or adjusted against one dip, it can trade one problem for another without showing where the power went.

First Mark the Exact Two Nodes

“A capacitor across the transformer” is not a circuit description. It may be connected:

• directly across the low-impedance input terminals;
• across only part of an autotransformer winding;
• between a tap and the winding common point;
• across the high-impedance antenna terminals; or
• at the end of leads whose inductance is already significant.

Those positions do not see the same RF voltage or current, and they do not create the same input admittance. A component directly across a calibrated 50 Ω input plane is a shunt branch at that plane. A component across a winding section is coupled through the transformer’s turns, leakage and distributed capacitance. Its effect must be solved or measured in that topology.

Record the schematic, tap count, winding sense, lead lengths, connector, enclosure and low-side return connection. Without those details, a capacitor value copied from another box has no transferable meaning.

Susceptance Explains the Useful Part

For an ideal capacitor connected directly in shunt at a declared reference plane:

The capacitor adds positive susceptance that grows with frequency. If the transformer and attached antenna present negative net susceptance at that plane, the added branch can reduce the total susceptance near a chosen frequency. If the conductance is also near the value needed for the system reference impedance, the reflection coefficient and SWR may improve.

That is the useful edge. The other edge follows from the same equation: the added susceptance is not constant. A value that improves the high end can overcompensate another band, create a new resonance or anti-resonance with leakage inductance, or move the closest approach to 50 Ω away from the frequency you intended.

A real capacitor adds ESR, ESL, dielectric loss, tolerance, temperature coefficient and voltage dependence. Its leads and mounting loop join the model. Near or above the component-and-layout self-resonant frequency, the branch may no longer behave as the nominal capacitance suggests.

The Transformer and Antenna Supply the Rest of the Network

An EFHW transformer is not an ideal turns-ratio block. Its input includes:

• finite magnetising inductance and core-loss resistance;
• leakage inductance from incomplete coupling;
• winding and connection resistance;
• turn-to-turn, winding-to-winding and winding-to-enclosure capacitance;
• the transformed high-side complex load;
• the connector, enclosure and measurement fixture; and
• the low-side return path, including any deliberate counterpoise and coax-exterior current.

Fair-Rite and Mini-Circuits both show why broadband-transformer response changes at its low and high ends: finite magnetising impedance dominates one region, while leakage, capacitance, conductor loss and core loss increasingly shape another. An added capacitor interacts with all of them. There is no single “transformer inductance” with which it always resonates.

The antenna load is equally conditional. End-feed impedance changes with wire length and diameter, height, bends, ground, nearby conductors, loss, operating harmonic and the available return path. A value selected with one resistor or one installation may not produce the same response on another wire.

An SWR Minimum Is Not the Bare Wire’s Resonance

A VNA or SWR meter at the coax connector reports the complete network transformed to that reference plane. A local minimum in SWR is the frequency where the measured impedance passes closest to the instrument’s reference impedance. It need not be the exact frequency where input reactance is zero, and neither condition identifies the resonance of an isolated wire hidden behind the transformer.

The capacitor does not magically change the natural response of a disconnected radiator. It changes the connected network. Because the matching unit, return path and radiator exchange current and voltage, the installed system may also settle into a different current distribution when the match and boundary conditions change.

This is why trimming only for the lowest connector SWR can be misleading. Record the complex impedance as R + jX or G + jB, define the reference plane, and keep the radiator and return geometry fixed while comparing capacitor states.

Low SWR and Efficiency Answer Different Questions

An ideal capacitor is lossless. A correctly chosen real capacitor can improve input match with very little dissipation, and the reduced mismatch can increase the power accepted from a 50 Ω transmitter. It is therefore wrong to declare that every shunt capacitor reduces efficiency.

It is equally wrong to call an improved SWR proof of higher radiation efficiency. Current in the compensation branch produces ESR and dielectric loss. The changed input condition may increase winding current, core flux, conductor loss or voltage across distributed capacitance. A lossy network can also look well matched.

Keep the power terms separate:

• Mismatch efficiency describes the fraction of incident power accepted at a defined port.
• Transformer-and-network loss describes accepted power dissipated before reaching the intended antenna terminals.
• Antenna radiation efficiency is radiated power divided by power accepted at the antenna input under the IEEE definition.
• Total or realised efficiency also includes mismatch at its stated reference plane.

An SWR trace addresses only part of the first item. It cannot separate ferrite, conductor, capacitor, common-mode or radiation losses.

RF Voltage, Current and Heat Must Be Rated at the Component

For a sinusoidal voltage across a capacitor that is still operating in its capacitive region, a useful screening relationship is:

Use the voltage across the actual two nodes, not transmitter power converted through an assumed 50 Ω load. A capacitor across a tap or high-impedance terminal may see a very different voltage from one across the input connector. Harmonics, modulation envelope and mismatch can raise peak stress beyond a sine-wave RMS estimate.

KYOCERA AVX’s RF-capacitor guidance separates voltage-limited and thermal current limits; KEMET ties heating to RF current, ESR and thermal resistance. The selected part also needs suitable dielectric, RF current capability, voltage rating, temperature range, dissipation factor and self-resonant behaviour in the final mounting geometry.

Measure temperature with a sensor or validated thermal method at declared power, waveform, duty cycle, ambient, enclosure and test duration. A finger is not an RF current meter, cannot see an internal hot spot and belongs nowhere near exposed transmitter voltage.

The Capacitor Does Not Replace a Return-Path Design

The EFHW circuit still needs a return path. It may include a deliberate counterpoise, a defined section of coax exterior, distributed capacitance to the surroundings and station wiring. A low-side shunt capacitor changes differential input admittance; it is not a common-mode choke and does not by itself define where exterior current stops.

Changing the capacitor can change the balance of currents among available paths. That may alter feedline-exterior current, pattern, touch voltage or RF in the station even when connector SWR improves. Measure common-mode current on the feedline and deliberate return conductor, and repeat the impedance sweep after any choke or cable-routing change.

A Controlled A/B Test Makes the Trade Visible

Freeze the Installation and Reference Plane

Keep the radiator, height, bends, ground condition, feedline route, choke position and return conductor unchanged. Calibrate the VNA at the transformer input connector or de-embed the cable and fixture to that plane. Save frequency, power, IF bandwidth and calibration details.

Measure More Than SWR

Sweep every intended band with no added capacitor and with the candidate value. Save R, X, G, B, reflection magnitude and phase. Overlay the traces rather than comparing two screenshots with different scales. Repeat with capacitor tolerance limits and, where material, temperature.

Separate the Transformer from the Installed Antenna

Use a matrix of representative complex loads at the high-side plane to characterise the transformer assembly. A single nominal resistor is a useful starting point, not a multiband proof. De-embed fixtures whose parasitics matter and state all port reference impedances.

Measure Loss Under Representative Loads

Use calibrated incident, reflected and delivered power at declared planes, a justified network method, or calorimetry with an uncertainty budget. A back-to-back transformer result can be useful only when the cascade, terminations, interaction and symmetry assumptions are established; dividing paired loss by two is not automatic.

Apply Power and Watch the Whole Assembly

At each claimed band, test the intended waveform, power, duty cycle, load region, ambient, enclosure and duration. Monitor capacitor, winding, core, connections and enclosure temperature; track impedance drift and nonlinear products. Stop at the lowest component, insulation or thermal limit rather than a universal temperature.

Return to the Installed Current Paths

Repeat the A/B/A comparison on the antenna and measure feedline-exterior current, accessible RF voltage and a repeatable field or received-signal indicator where practical. Propagation makes slow on-air comparisons unreliable, so use rapid switching, a local field setup or repeated observations.

A Tuner Is an Alternative Network, Not an Automatic Winner

A shack tuner, remote tuner and fixed shunt capacitor occupy different reference planes and see different impedances. A tuner may provide adjustment across conditions but can add its own inductor, capacitor, contact and feedline losses. A remote tuner may reduce high-SWR feedline loss while adding outdoor voltage, weather and control constraints. A fixed compensation capacitor may be simpler and lower loss inside a declared operating envelope.

Compare them as complete systems with the same radiator, feedline, return path, accepted power and measurement planes. “Always use a tuner” is no better engineering than “always add this capacitor value.”

Questions, antenna-factor records or height trials to share? Contact RF.Guru.