The EFHW Capacitor Trap: Choose the Antenna Before the SWR Dip

The trap is letting the matching trace write the band plan. A shunt capacitor can cancel an inconvenient part of the input susceptance, but it does not give the core new material properties or move a directional null out of the bearing you want to work. If an added band makes the transformer, component stress or installed pattern a poor fit, I would rather leave that band to a better-suited antenna. That is a design decision, not a defeat.

Draw the Capacitor Between Its Actual Two Nodes

“A capacitor across the transformer” is not enough information. The component may be connected across the coax input, across part of an autotransformer winding, from a tap to the common node or across the high-impedance antenna terminals. Those positions see different voltage, current and transformed impedances.

Lead length, mounting loop, connector, enclosure, winding sense and the low-side return connection also join the RF circuit. A value copied from a different winding or box is not automatically transferable.

For an ideal capacitor connected directly in shunt at a declared input plane, the admittance becomes:

The capacitor adds positive susceptance that increases with frequency. If the existing network presents negative susceptance at that plane, the two can partially cancel near a chosen frequency. The input then approaches the line's reference impedance only if the remaining conductance is suitable as well.

A capacitor attached to a winding tap is not simply a shunt branch at the coax connector. Transformer action, leakage inductance and distributed capacitance stand between the component and the measured port. Solve or measure the complete topology.

Several Coupled Responses Can Create One SWR Dip

An installed EFHW matching system combines at least four frequency-dependent structures:

• The wire and its environment: conductor length, diameter, insulation, bends, height, ground and nearby objects set the antenna's complex load.
• The return system: an intentional conductor, a declared section of coax exterior, local capacitance and nearby conductive structures complete the current path.
• The transformer: magnetising impedance, leakage inductance, winding resistance, core loss and distributed capacitance shape its two-port response.
• The compensation part and layout: capacitance, ESR, ESL, tolerance, temperature coefficient and mounting geometry add another branch.

These responses are coupled after connection. It is therefore too simple to say that every new dip is “the capacitor resonating the transformer,” just as it is too simple to call it the wire's natural resonance. The measured minimum is where the input impedance of the complete connected network passes closer to the reference impedance.

Measure resistance and reactance—or conductance and susceptance—not only SWR. A Smith-chart trace can show whether the capacitor mainly cancels susceptance, shifts a loop, changes the resistive crossing or creates a narrow pole-zero interaction.

Higher Wire Modes Are Not Automatically Inefficient

On a frequency where a long wire supports several half-wave current regions, its pattern develops more lobes and nulls. The end impedance and the transformer's load can also change strongly with frequency and installation. That makes multiband matching and directional coverage harder to generalise.

Multiple current maxima do not, by themselves, mean the wire is lossy or that little useful radiation occurs. A higher-order mode can radiate efficiently while placing deep nulls in inconvenient directions. Conversely, a broad and attractive input match can coexist with substantial loss. Efficiency, pattern and match are different results.

For upper-HF use, inspect the complete installed current distribution and pattern. Ask whether the required bearings fall inside useful lobes, whether the matching network handles the actual load and whether common-mode current has changed. Do not infer those answers from the band label or transformer ratio.

What the Capacitor Can Improve

A well-chosen compensation capacitor can reduce net input susceptance over a useful part of the operating range. Depending on topology, it may counter a leakage-inductive trend, reshape a high-frequency roll-off, move the closest approach to 50 Ω or reduce the amount of correction required from another matching element.

That can be a real improvement. If the new operating point lowers winding current, reduces core excitation or raises accepted power without adding significant component loss, complete-system efficiency may improve. The direction is not universal; it must be measured with the intended complex loads.

A fixed part is most defensible when the required bands, load envelope, component tolerance, layout and powered limits are declared. It is not a universal extension that turns an arbitrary transformer and arbitrary end-fed wire into an efficient all-band system.

What the Same Capacitor Can Make Worse

The same frequency-dependent branch can overcompensate another band, create a sharper input resonance, increase sensitivity to wire height or moisture, raise voltage at a winding section or increase circulating current. A match that improves at room temperature may drift when the core, winding or capacitor warms.

A real capacitor has loss. While operating below self-resonance in a suitable series model, two useful screening relationships are:

Use the voltage across the actual capacitor nodes, not transmitter power converted through an assumed 50 Ω load. A high-impedance winding node can see far more voltage than the coax connector. Modulation peaks, harmonic content and mismatch can raise stress beyond a simple sine-wave estimate.

The component also needs adequate RF current, dielectric, voltage, temperature and self-resonance margin in the final mounting arrangement. Lead inductance and enclosure coupling can move the useful frequency response.

Bandwidth Can Grow Through Correction or Through Loss

A compensation network can broaden the range over which the input stays inside a specified SWR limit. That bandwidth can be useful, but it does not identify the mechanism. Reduced reactive error can improve useful transfer; increased dissipation can also flatten a reflection curve.

State bandwidth together with its criteria: frequency limits, maximum SWR or return loss, complex load range, minimum transfer, maximum temperature and permitted input power and duty cycle. A claim such as “covers 10 through 80 metres” says little without those boundaries.

Component tolerance and environment matter too. Sweep the nominal capacitor and its tolerance limits, then repeat across the expected temperature and installation states. A response that depends on one exact value can be difficult to reproduce.

SWR Describes Reflection, Not Where Accepted Power Went

At a declared reference impedance Z₀, the input reflection coefficient is:

A smaller |Γ| means more incident power is accepted at that plane. It does not say how the accepted power divides among radiation, transformer loss, capacitor loss, conductor loss, soil loss and unintended common-mode paths.

A lossy network can therefore produce a low SWR, but that does not mean every capacitor-induced dip is loss. An ideal capacitor is lossless, and a suitable RF capacitor may dissipate very little. The correct conclusion comes from a power balance or a validated loss measurement, not from suspicion in either direction.

The Return Path Remains Part of the Network

An end-fed wire still has two current terminals. The return may use a deliberate counterpoise, a defined section of coax exterior, distributed capacitance and nearby structures. A capacitor across a differential port is not a common-mode choke and cannot define where exterior feedline current stops.

Changing the capacitor can change input voltage and current, which can redistribute current among available return paths. That can alter pattern, touch voltage and RF in the station even if connector SWR improves. Measure coax-exterior current and intentional-return current with and without the capacitor.

If a choke is used, select its location from the intended current boundary and verify its complex common-mode impedance and powered limits on every required band. Moving the choke changes the antenna-side return geometry, so the compensation value may need to be reconsidered.

A Measurement Sequence That Exposes the Trade

These checks answer a practical decision: keep the compensated configuration, restrict its operating range, or use the alternative below. They are not a reason to postpone choosing an antenna indefinitely.

• Freeze the installed system: wire, height, bends, transformer, enclosure, return conductor, coax route and choke position remain fixed.
• Declare the reference planes: calibrate or de-embed to the transformer input and, when possible, characterise the antenna-side port separately.
• Measure the uncompensated network: save R, X, G, B, reflection magnitude and phase across every intended band.
• Add the candidate part: record its exact nodes, value, part family, tolerance, lead length and mounting geometry; repeat the sweep on the same scales.
• Test representative loads: use a matrix of the complex loads the transformer will actually encounter, not one nominal resistor or an open-circuit trace.
• Measure transfer and loss: use a justified two-port method, calibrated power balance or calorimetry with the reference planes and uncertainty stated.
• Apply power: verify voltage, current, temperature, impedance drift and nonlinear behaviour at the intended waveform, duty cycle, load, enclosure and ambient.
• Return to the antenna: map common-mode current and compare useful field at equal accepted power with rapid A/B/A switching or an equivalent controlled method.

A back-to-back transformer test can help, but dividing cascade loss by two is valid only when the two networks, intervening impedance, terminations and interaction support that assumption. The installed complex antenna load remains the final test.

The Better Alternative Is a Better-Fitted Job

For one priority band or a deliberately chosen pair, I favour an EFHW transformer designed around those loads and frequencies. A narrower job lets the designer concentrate on the required magnetising behaviour, winding arrangement, parasitic response and power duty instead of forcing the same compact assembly to satisfy widely separated extremes. That does not prove any particular unit is efficient; it removes unnecessary conflicting requirements.

If several upper-HF bands are the real objective, I would also consider an off-centre-fed arrangement whose installed loads permit more moderate transformation, or a separate radiator sized for the priority band. The off-centre option can reduce the required high-side voltage ratio and winding burden when the complete current geometry supports it. It still needs its intended return, a suitable choke boundary and a useful pattern. A different label or a smaller ratio alone is not the advantage—the better-fitting load and current system are.

• Keep the capacitor when it provides useful compensation across the required loads without unacceptable loss or stress. A properly designed compensated transformer is legitimate engineering.
• Restrict the EFHW to its useful bands when a difficult extra band compromises an otherwise good installation. There is no need to discard an effective antenna because it does not do every job.
• Change the antenna or feed arrangement when the wanted band requires a more suitable load, better current placement or different directional coverage. Do not keep adjusting an input dip to answer a pattern or loss problem it cannot solve.

My recommendation is not to ban upper-band EFHW operation. A suitably designed short EFHW or useful higher-order mode can work well there. It is to stop treating an electrically stretched all-band claim as a requirement the hardware must somehow obey.

High-Impedance Feed Regions Demand RF-Safety Margin

EFHW matching networks and wire ends can develop substantial RF voltage. Keep exposed terminals inaccessible during transmission, use suitable insulation and creepage/clearance, provide mechanical strain relief and de-energise before touching or changing the capacitor.

Increase power in controlled steps while monitoring the entire assembly. A cool core does not prove the capacitor is cool, and a cool enclosure does not prove there is no internal hot spot or excessive electric-field stress. Complete the applicable RF-exposure assessment using actual power, duty cycle, pattern and accessible geometry.

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