Monoband EFHW Matching on 17–10 m: Choose the Network From the Load

A single-band matching network can be optimized around one measured load and frequency. That can reduce compromise compared with a broadband multiband design. It does not make a 49:1 transformer unnecessary, a tuned network lossless or a resonant wire maximally efficient. Those are separate engineering questions.

Start With Electrical Length, Then Trim the Installed Antenna

The exact speed of light in vacuum is 299,792,458 m/s, as documented by NIST. The free-space half wavelength at frequency f is c/(2f). A real wire is shortened by diameter, insulation, end geometry, supports, height, bends, nearby objects and ground coupling. Calling one empirical shortening factor a wire “velocity factor” hides those mechanisms.

The table gives four example design frequencies. Its 0.95 column is 95% of the free-space half wavelength; treat it as a construction starting point, not a guaranteed resonant length. Start slightly long enough to permit trimming in the final geometry.

Those frequencies are examples, not permission to transmit. Choose a design frequency inside the segment authorized for the operator, emission and jurisdiction; current regional guidance is available from the IARU Region 1 band-plan page. The 10 m allocation is much wider than 17 m or 12 m, so one fixed high-ratio match should not be assumed to cover every desired 10 m segment.

Do not trim the bare wire to a supposed resonance and then assume the matcher will leave it unchanged. The transformer, L-network, two-wire section, counterpoise or exterior coax, choke, enclosure capacitance and nearby supports become part of the installed structure. Adjust the assembled antenna from low-power measurements at a defined reference plane.

The End Impedance Is a Result, Not a Constant

A half-wave wire has a current minimum and voltage maximum near its open end, so its feed impedance is high. It is not always 3–5 kΩ, purely resistive or identical at both ends. The value changes with:

• electrical length and the exact feed position relative to the current minimum;
• wire radius, insulation, bends and end hardware;
• height, ground parameters and nearby conductive/dielectric objects;
• the transformer box, line section, counterpoise, mast and feed-line exterior; and
• frequency, loss and measurement reference plane.

The official ARRL EFHW kit guide, for example, describes one multiband implementation as approximately 2.5 kΩ and uses a 49:1 network. That does not make 2.5 kΩ or 49:1 universal. Model the complete wire and environment with a validated method such as LLNL’s Numerical Electromagnetic Code, then confirm the installed feed impedance with calibrated low-power measurement.

Four Conditional Matching Structures

1. Ferrite impedance transformer

An ideal 7:1 turns ratio gives a 49:1 impedance ratio. It reflects 2.45 kΩ as 50 Ω, 2.5 kΩ as about 51 Ω and 3.5 kΩ as about 71 Ω before winding reactance, leakage, capacitance and loss are included. The radiator and any compensation are commonly adjusted so the complete input is acceptable.

Real transformer performance depends on exact material, core geometry and volume, turns, winding arrangement, lead length, compensation, load impedance, frequency, power, waveform, duty cycle and temperature. Fair-Rite’s broadband-transformer guidance separates low-frequency magnetizing/shunt limitations from high-frequency leakage and parasitic limitations. A nominal ratio alone gives neither insertion loss nor power rating.

Wes Hayward, W7ZOI, published a controlled example in Transformers for the End Fed Half Wave Antenna: his FT-114-61 3:27 transformer into 4.4 kΩ measured 0.21 dB loss at 14 MHz, while an FT-114-43 example of the same nominal ratio and load measured 1.3 dB. Those are small-core, low-power, 14 MHz results—not specifications for 17–10 m or another build. They demonstrate why exact material, winding and test conditions matter.

2. Tapped shorted two-wire section

This J-pole/Zepp-family structure uses a roughly quarter-wave two-wire section shorted at one end, joined to the radiator at the other and tapped by the coax. The radiator and matching section load each other. It is not equivalent to inserting a uniform quarter-wave transformer between two resistors.

A starting line factor of 0.98 produces the following unloaded physical quarter-wave starting lengths. Use the measured or manufacturer-specified velocity factor of the actual line; the final length changes with end effects, conductor spacing, weather and radiator loading.

No universal tap distance follows from “600 Ω,” the band and an assumed 3.5 kΩ endpoint. Tap position, line length, radiator length, spacing, wire diameter, feed transition, choke and surroundings are coupled variables. A build dimension therefore needs the complete geometry, a model or derivation and installed measurement.

Build adjustability into a chosen, fully specified geometry. Model the complete conductors, choose a conservative initial tap near the short from that model, and use a VNA calibrated to the intended coax reference plane at low power. Adjust radiator length, matching-section length and tap iteratively; do not assume that moving the tap always changes resistance monotonically in the installed resonant structure.

3. Balanced open-line coupler

A classic end-fed Zepp arrangement uses a balanced two-wire feeder with one conductor connected to the radiator and the other open at the antenna end. The feed line, coupler and radiator form one tuned system. A balanced link-coupled network or other suitable tuner can match it, but “10–70 pF” or “2–3 turns” is not a design until the measured complex input impedance, component Q, voltage/current and coupling are stated.

This approach can have low conductor loss when built from suitable open wire and operated within its design envelope. It can also develop high standing voltage, radiate from an imbalanced feeder or couple strongly to its surroundings. Keep the parallel line spaced from conductive objects, preserve symmetry and verify feeder current balance rather than assuming the name “Zepp” guarantees it.

4. A tuned L-network at the feed point

A two-element L-network can provide a compact single-frequency match. Component values must come from the installed complex load. For the limited textbook case of a 50 Ω source and a purely resistive 3.5 kΩ load, the minimum loaded Q is:

For exactly that idealized load, the calculated component examples are:

These are calculation checks, not finished construction values. If the load is 2.5 kΩ, 5 kΩ or has reactance, both components change. Parasitic capacitance and inductance become significant at the small capacitances used on 12 and 10 m. The Analog Devices L-network derivation and its complex-load matching calculator are useful cross-checks, but the installed measurement remains decisive.

A high-Q coil and low-loss capacitor can make this network efficient, but not lossless. Estimate insertion loss from component Q and layout, then verify it. The fixed high transformation ratio also makes the match frequency sensitive. Loss may broaden an SWR curve while converting more power to heat, so wider low SWR is not proof of better bandwidth or efficiency.

High End Impedance Means High RF Voltage

If the load were purely resistive at 3.5 kΩ and accepted power P, the load voltage would be:

Those are ideal load-terminal values, not guaranteed maxima. Reactive mismatch, standing waves and matching-network circulation can raise local voltage elsewhere. PEP governs waveform crest stress; average power and duty cycle govern much of the heating. Select capacitors, inductors, wire, connectors, spacing, supports and enclosure for the calculated worst case plus a justified margin—never from transmitter output alone.

Compare Matching Loss Under the Same Conditions

A line or L-network can beat a particular transformer, and a well-designed transformer can beat a poorly built tuned network. Compare insertion loss with suitable high-impedance RF loads, calibrated or de-embedded fixtures and a stated uncertainty. At operating power, add temperature-rise or calorimetric evidence and inspect waveform and duty-cycle dependence. SWR alone cannot separate transformation, reflection and dissipation.

Use the same complex load, frequency span, reference planes and environmental state for every candidate. Record wanted-band insertion loss, usable matched bandwidth, component voltage and current, exterior-feedline current, temperature rise and post-warm-up frequency shift. Repeat after enclosure closure, cable routing and wet or dry exposure representative of service.

Resonance, Efficiency, Gain and EIRP Are Different

Resonance means the input reactance is zero at a defined reference plane. Matching means the transmitter sees the intended impedance. Neither establishes radiation efficiency or gain. A complete power accounting is:

The wire, ground/environment, matching network, feed line and unintended common-mode path can all change the result. Geometry and height set the radiation pattern; matching cannot force power into a desired direction. A lossy network can show a beautifully broad low-SWR response while reducing radiated power.

For a comparison, keep radiator geometry, site, coax reference plane and accepted power common. Measure network insertion loss with an appropriate high-impedance fixture, map exterior-coax current, and use controlled field or rapid A/B observations over multiple paths and times. One distant report or one SWR sweep does not establish gain or efficiency.

Feed-Line Common Mode Must Be Designed, Not Assumed Away

An end-fed radiator needs a complete RF current path. Depending on topology, return/displacement current can involve a deliberate counterpoise, the second conductor of a matching section, the coax exterior, mast, station wiring and surrounding capacitance. Removing ferrite from the impedance transformer does not automatically remove common-mode current or receiver noise.

A coax transition at a tapped two-wire section can still mode-convert if the structure or connection is asymmetric. A choke at the tap is not universally required or sufficient; its location and complex impedance should follow measured exterior current and the intended antenna boundary. Use the ARRL measure–install–retest common-mode workflow across the actual bands and cable route.

Likewise, a counterpoise is not a fixed 1–2 m accessory. Its electrical length, routing and coupling make it part of the antenna. Specify the intended current path, model it, then verify exterior current, feed impedance and pattern after installation.

A Practical 17–10 m Design Sequence

1. Choose the actual operating segment. Use the current national authorization and regional band plan, not merely the band label.
2. Define the geometry. Record wire, insulation, height, orientation, supports, ground/environment, matching enclosure, counterpoise or two-wire section and coax route.
3. Start the radiator slightly long. Use the 0.95 × λ/2 values only as rough construction starts.
4. Model and measure the complex end load. State the reference plane and include the intended return structure.
5. Select the topology from constraints. Compare space, loss, bandwidth, tuning access, weather sensitivity, voltage, power/duty cycle and common-mode control.
6. Calculate from the measured load. Do not use a generic tap distance or ideal 3.5 kΩ L-network values as final dimensions.
7. Prototype at low power. Make radiator, tap/line and reactive components adjustable; tune the assembled structure without touching energized conductors.
8. Verify more than SWR. Check insertion loss, component temperature, exterior current, useful bandwidth, pattern/field behaviour and stability in expected weather.
9. Increase power in stages. Use the real waveform and duty cycle, observe temperature remotely and stop if match or temperature drifts.

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