Why a 49:1 EFHW Sweep Does Not Prove 80–10 Metre Performance

Hook a VNA to a nominal 49:1 EFHW transformer and the trace can be very revealing. It can also be very easy to overread. A pretty input match does not prove low loss, and an ugly unloaded trace does not prove that a correctly loaded transformer is useless. The engineering question is whether the complete network transfers power into the real antenna load, across every claimed band, without unacceptable loss, voltage, current, heat or unintended common-mode behaviour.

The Trace Is Real; the Conclusion Needs a Boundary

A one-port VNA measurement reports the complex reflection coefficient at its calibrated plane. From that, the instrument can display return loss, SWR or impedance. It does not directly report how much accepted power becomes load power, transformer heat, fixture radiation or common-mode current.

If the high-impedance terminal is open, shorted or connected to an arbitrary resistor, the transformer is being tested under that exact condition. An installed EFHW presents a different, frequency-dependent complex impedance and a physical return path. Change the load, lead length, enclosure, capacitor, cable routing or calibration plane and the trace can change.

The Keysight Impedance Measurement Handbook treats the fixture, cables, adapters, signal level and measurement method as part of the uncertainty. That is especially important when a 50 Ω instrument is asked to infer kilohm-order behaviour through a high-ratio network.

What the 49:1 Label Actually Says

For an ideal transformer:

A 7:1 turns ratio therefore corresponds to a 49:1 ideal impedance ratio. In the simplest arithmetic, 50 Ω maps to 2450 Ω. That is a useful starting point—not a statement that every EFHW feedpoint is 2450 Ω on 80, 40, 20, 15 and 10 metres.

The feedpoint impedance changes with wire length and diameter, installation height, bends, nearby conductors, ground, operating harmonic, losses and the conductor connected to the transformer’s other terminal. A 49:1 can be a sensible choice when the real high-side load occupies the intended region. It can be the wrong choice when that load does not.

The current ARRL EFHW kit documentation provides a concrete bounded example: a four-band 40/20/15/10 metre wire described around a 2500 Ω feed region, a 49:1 network and optional primary compensation for the higher bands. That successful implementation is not evidence that every 80–10 metre wire and transformer shares the same load or bandwidth.

The Low-Frequency Limit Is Set by the Loaded Magnetising Branch

At the low end, finite primary inductance appears as a shunt magnetising admittance. If its reactance is too low relative to the transformed load, it diverts current and can increase winding and core loss. The relevant comparison is not “two turns are always too few”; it is the complete primary impedance, core material and geometry, frequency, winding voltage and intended load.

Lower frequency reduces magnetising reactance. More turns can raise inductance, but it also increases winding length and parasitic capacitance. Larger or additional cores can change flux density and thermal capacity. The design is a broadband compromise, not a one-number rule.

An unloaded input sweep can make the magnetising branch look dominant because no intended high-side load is present. That is useful diagnostic information, but it is not the loaded insertion loss. Apply a characterized load matrix that spans the antenna’s measured or modelled complex impedances before declaring the low-band result.

The Upper-Frequency Limit Is a Network of Parasitics

At upper HF, leakage inductance, winding capacitance, lead inductance, enclosure capacitance and coupling between turns become comparable with the intended impedances. Resonances and anti-resonances appear, the transformation moves away from its ideal ratio and voltage distribution can become uneven.

The transformer does not suddenly “stop being a transformer” at one universal frequency. Its equivalent circuit becomes increasingly distributed, and the useful band depends on winding topology, geometry, material, load and compensation.

Fair-Rite’s broadband-transformer technical note separates the low-frequency magnetising branch, mid-band winding resistance and high-frequency leakage/capacitance mechanisms. It also explains the basic design tension: more inductance helps the low end, while fewer turns generally reduce winding loss and parasitics at the high end.

A Shunt Capacitor Can Be Real Compensation

A capacitor across the low-impedance side can counter part of the transformer’s unwanted high-side capacitance as reflected into the input. The ARRL kit explicitly uses an optional primary capacitor for improved behaviour on 15 and 10 metres. Calling every compensation capacitor a cosmetic “crutch” is therefore too simple.

The capacitor changes the complete network. It can improve input match and power transfer over a chosen region; it can also create higher circulating current, sharper resonance or voltage stress if poorly chosen. A flatter S11 trace is encouraging, but it still does not quantify dissipative loss.

Measure S11 and transmission with the intended load, inspect internal voltage/current in a validated model, and repeat at representative power and temperature. The capacitor earns its place when the complete loss, stress and bandwidth results improve—not merely because the SWR line looks flatter.

S11 Is Not Insertion Loss

A well-matched input can hide a lossy network. A 50 Ω resistor proves the point: it has an excellent match and turns virtually all accepted power into heat. Conversely, a low-loss but mismatched network can reflect substantial power without dissipating it.

Keysight’s in-fixture VNA guidance explains why calibration standards, fixture construction, residual error and de-embedding belong to the result. At kilohm-order terminals, a few picofarads and exposed leads are not harmless details.

A Proper Loaded Measurement Tells a Stronger Story

Characterize the transformer in layers:

• Open- and short-circuit diagnostics: estimate magnetising behaviour, leakage and resonances without mistaking them for loaded efficiency.
• Characterized resistive loads: test several high-side values rather than one nominal 2450 Ω resistor.
• Representative complex loads: cover the impedance region measured or modelled for the installed wire on every claimed band.
• Full complex S-parameters: retain S11, S21, S12 and S22 with phase, calibration planes and fixture data.
• Accepted-to-delivered power: separate mismatch from dissipation at the same reference planes.
• Power and thermal sweep: state carrier or modulation, PEP, average power, duty cycle, load, temperature, enclosure and test duration.
• Return-current map: measure current on the intended counterpoise and coax exterior because the bench two-port does not define the installed common-mode path.

The useful result is not “49:1 passes” or “49:1 fails.” It is a frequency-, load-, power- and temperature-bounded operating envelope.

The 80–10 Metre Claim Belongs to the Complete Antenna

For each claimed band, the installed system must answer all of these questions:

• Does the wire support the intended current mode and useful radiation pattern?
• What complex impedance appears at the transformer’s antenna terminals?
• What power does the transformer accept and deliver into that load?
• What voltage, current and temperature occur at the intended power and duty cycle?
• Where does return current flow, and where does the choke define its boundary?
• What loss remains in the feedline, tuner, transformer, conductor and environment?
• What realised gain appears in the azimuth, elevation and polarisation that matter?

A low shack SWR answers only a small part of that list. A lossy transformer or feedline can make the displayed match look better, while an efficient antenna can present a difficult impedance. Multiband convenience and multiband efficiency are different claims.

Why a 4:1 EF-OCF Can Be the More Honest Architecture

RF.Guru’s practical alternative is not “replace every 49:1 with a 4:1 box.” It is to start with a different antenna architecture whose complete current path is explicit. In a one-end EF-OCF, a declared main wire and a deliberate shorter return branch form the antenna, while a separately specified choke defines the end of that intended exterior-current section.

When the measured complex load calls for it, a nominal 4:1 UNUN can perform the differential impedance transformation. The separate choke performs the common-mode-control function. Keeping those jobs explicit makes each one measurable and lets the choke sit where the intended return branch actually ends.

That can reduce the transformation challenge when the installed EF-OCF load occupies a lower impedance region than an EFHW feed. It does not guarantee lower loss, lower voltage, 50 Ω on every band or a better pattern. A well-designed EFHW and a well-designed EF-OCF are both valid candidates; the site, current path, load, feed-system loss and wanted pattern decide.

What Makes the Debunking Defensible

The strongest challenge to an “80–10 metre” promise is not that high-ratio transformers are impossible. It is that the promise is incomplete until the manufacturer or builder publishes the conditions that make it true:

• the exact core material, volume, topology and winding geometry;
• the high-side complex load range on every band;
• the calibration planes, fixtures and small-signal S-parameters;
• accepted-to-delivered power or calibrated thermal loss;
• PEP, average power, duty cycle, enclosure and stabilized temperature;
• voltage and current margins, including compensation components;
• the intended return path and measured exterior-current boundary; and
• the complete installed pattern or realised-gain evidence.

Without those boundaries, the label is a winding ratio followed by a band list. With them, it becomes an engineering specification.

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