Stop Blaming Your Coax Before Checking the EFHW
Stop Blaming Your Coax Before Checking the EFHW
The coax is easy to blame because its loss appears in a data sheet. The high-ratio box at the antenna deserves the same scrutiny. My preference is a purpose-built single- or dual-band EFHW, not an 80–10 m coverage label accepted as proof of low-loss broadband transformation.
RF.Guru working definition: Common-mode current is the non-cancelling phasor-sum current in a specified set of conductors, evaluated at a defined cross-section and using a declared current-direction convention. In the intended differential transmission-line mode, the outgoing and return currents are equal and opposite, so their phasor sum is zero. When they do not cancel, the remaining current must close through another reference or return path—such as the outside of a coax shield, a mast, equipment chassis, station wiring, nearby structures, earth, the operator, or distributed coupling through the environment.
This broader working definition is especially useful in practical antenna systems. On transmit, non-cancelling current on the outside of the coax can make the feedline and connected structures part of the radiating antenna system unless that path is intentional, clearly defined and properly controlled—for example by providing the required return path and placing a suitable common-mode choke at the correct boundary.
When the signal is disappointing, the feed line often gets blamed first. Yet replacing a sound run of coax cannot recover the power being dissipated in a poorly suited EFHW transformer. A neat SWR curve does not acquit the matching box. That is the point I want operators to keep in view before buying thicker cable.
I am not arguing that coax loss is negligible. A long, thin, damaged or badly mismatched run can be the main problem. I am arguing against scrutinising every fraction of a decibel in the cable while treating one high-ratio transformer as equally well suited to 80 through 10 metres. Choose the transformer’s job deliberately, then account for the actual losses.
My practical recommendation: for an EFHW, start with a well-defined single band or a deliberately engineered two-band grouping. That reduces the frequency and load range that one ferrite assembly must accommodate. If broad multiband coverage is essential, require evidence for the complete high-ratio assembly—or consider a feed arrangement that presents a less extreme matching problem. Fewer advertised bands can be an engineering advantage, not a missing feature.
RF safety: an end-fed half-wave has a high-voltage feedpoint. At 100 W into a purely resistive 2450 Ω, the simple sinusoidal value is about 495 V RMS and 700 V peak; real mismatch and transients can raise local stress. De-energize before changing the transformer, wire, counterpoise, choke or test fixture, and use components, enclosures and clearances rated for the actual voltage, power, duty cycle and environment.
Put the Matching Box Back in the Power Budget
The useful question is not whether coax or EFHWs are guilty as a class. It is whether the matching system has escaped the scrutiny already applied to the cable. Follow the power through successive boundaries:
Transmitter → tuner/filters → coax differential mode → matching transformer → antenna terminals → conductor/ground/environment → radiation
A parallel path may also carry common-mode current on the outside of the coax shield, station wiring or another conductor. That path can radiate, dissipate heat, change the intended pattern, couple noise on receive and create RF-safety or EMC problems.
| Quantity | What it means | What does not prove it |
|---|---|---|
| Matched line attenuation | Conductor and dielectric loss of a specified cable, length, frequency and condition when terminated in its nominal impedance. | SWR alone or a generic cable-family name. |
| Additional line loss under mismatch | Extra conductor/dielectric dissipation caused by the higher voltage/current distribution on the lossy line. | The interface mismatch-loss formula by itself. |
| Transformer dissipation | Core, winding, dielectric and connection loss at a stated impedance, frequency, power, waveform and temperature. | A low input SWR or an unloaded VNA sweep. |
| Mismatch loss | The fraction of an incident wave not accepted at a specified reference plane: −10 log10(1 − |Γ|²). | Heat in that interface; reflected energy is not automatically dissipated there. |
| Radiation efficiency | Radiated power divided by power accepted by the antenna structure. | A good SWR, a distant signal report or one field-strength direction. |
| Realized gain | Efficiency, mismatch and radiation pattern combined in a stated direction and polarization. | Transformer temperature alone. |
Do not double-count: if a measured end-to-end insertion result already includes mismatch, fixture and connector effects, those terms cannot simply be added again. Define the input and output reference planes before adding dB.
Coax Loss Depends on Cable, Length, Frequency and Load
Matched attenuation is set by the exact cable construction, length, frequency, temperature and condition. The official Times Microwave LMR-400 data sheet, for example, specifies typical attenuation of 0.7 dB per 100 ft at 30 MHz for a 1.0:1 termination and 25 °C. That leaves about 85% of the input power after 100 ft. It says nothing about RG-58, a 10 ft jumper, a 100 m tower run, water ingress or damaged connectors.
Loss in dB converts to a power-transmission ratio through:
η = 10−LdB/10
Thus 0.1 dB transmits 97.7%, 0.5 dB transmits 89.1%, 1 dB transmits 79.4% and 3 dB transmits about 50.1%. These percentages are useful only after the dB term has been tied to a real component and test condition.
Mismatch changes line dissipation
A lossy line has baseline matched attenuation. With mismatch, forward and reverse waves change the current and voltage distribution along the line, so total conductor and dielectric dissipation generally increases. The added loss depends on the line’s attenuation, electrical length and complex load reflection—not SWR magnitude alone.
That extra line dissipation is distinct from interface mismatch loss. At a load plane, 1 − |Γ|² is the fraction of the incident power accepted on that encounter. What happens to the reflected power depends on the source, tuner and intervening lossy network. “Multiple passes” can be a helpful story, but a steady-state transmission-line calculation is the reliable method.
Loss also makes a shack-end SWR look deceptively better because the reflected wave crosses the feed line twice. Rohde & Schwarz’s antenna-system return-loss note shows that feeder attenuation and cable mismatch can corrupt an antenna return-loss measurement made at the bottom of the line. A low shack reading therefore does not prove low feedpoint reflection or low line loss.
The Band List Is a Much Bigger Promise Than the 49:1 Label
A nominal 49:1 impedance transformation corresponds to a 7:1 turns ratio in the ideal model. It maps 2450 Ω to 50 Ω. A real EFHW terminal impedance is not fixed at 2450 Ω: wire length and diameter, height, shape, frequency, nearby objects, ground, the opposite-terminal path and the transformer itself all change the complex load.
High-ratio broadband transformation is demanding, but “49:1” does not determine the loss. Core material and volume, primary inductance, turns, winding resistance, interwinding capacitance, leakage inductance, construction, load, frequency, power, duty cycle and temperature all matter.
An 80–10 m brief spans more than three octaves; a 40–10 m brief still spans more than two. A list of usable amateur-band windows is not continuous broadband antenna behaviour. Even a genuinely broadband transformer would still feed a wire whose complex impedance, current distribution and pattern change between those windows.
Fair-Rite’s official broadband-transformer guidance divides response into low-, mid- and high-frequency regions. Low-frequency roll-off is linked to falling shunt/magnetizing impedance; the high-frequency region is limited by leakage inductance and parasitic capacitance. The cutoff frequencies belong to the individual design—not to every transformer with the same ratio.
That is the conflict I prefer to reduce at the design stage. More turns can provide useful low-frequency magnetising impedance, while their winding length, leakage and capacitance can make the upper end harder. Reducing turns to improve high-frequency behaviour can leave the low-frequency end short of inductance or voltage-per-turn margin. Material choice and core geometry change this trade; neither a #43 nor a #52 label resolves it on its own. Their published material curves are not a finished EFHW transformer rating.
A common winding/material combination can therefore be limited at one end of an 80–10 m span, or at an edge of a 40–10 m span. That is not a universal frequency at which all EFHWs fail, nor proof that no ferrite assembly can provide useful multiband windows. It is a reason to ask what that particular construction gives up to reach its endpoints.
Measured designs can differ substantially
Wes Hayward, W7ZOI, published a controlled EFHW transformer measurement that used a tunable L-network to return the high-impedance side to 50 Ω. He measured the L-network component Q, calculated its loss and subtracted it from the cascade. His small-core QRP examples are not ratings for modern commercial units, but they show how strongly loss can depend on a particular construction:
| Measured construction | Load | 7 MHz loss | 14 MHz loss |
|---|---|---|---|
| FT-114-61, 3:27 turns | 4.4 kΩ | 0.10 dB | 0.21 dB |
| FT-114-43, 3:27 turns | 4.4 kΩ | 1.30 dB | 1.30 dB |
At 1.3 dB, about 74% of accepted input power remains; at 0.1 dB, about 98% remains. These 3:27 windings have an ideal turns-ratio-squared transformation of 81:1, not 49:1. The comparison does not establish a universal core-material ranking: it applies to Hayward’s turns, cores, load, frequencies, fixture and low-power conditions. Transformer loss therefore needs measurements or specifications for the exact construction and operating point.
The MyAntennas EFHW-8010 product page specifies a three-core transformer for 3.5–30 MHz and claims less than 0.4 dB average loss. Treat this as a design-specific manufacturer claim, not independent validation. It is a testable statement for that model and does not establish the performance of other EFHW matching units. An average also does not specify the worst loss at a band edge, at a different complex load or after the enclosure reaches operating temperature. I want those boundaries, not just the attractive average.
A Shunt Capacitor Is Not a Free Bandwidth Upgrade
A shunt or compensation capacitor can resonate part of the transformer’s leakage inductance or adjust the input match over a chosen range. That is legitimate network design. It can also carry substantial RF current, see high voltage, add dielectric/ESR loss and create a narrow or load-sensitive response.
For a capacitor across two stated nodes, the shunt susceptance is B = 2πfC. Its influence changes with frequency; the branch current also depends on the RF voltage across it. Compensation that helps a chosen region can overcompensate another, shift poles or zeros, or change circulating stress. A shunt branch is not the same thing as a complete LC network deliberately solved for a specified source and load.
This is why I will not accept “we added a capacitor and the SWR improved” as the end of the argument. Evaluate the complete network against its insertion-loss, return-loss, voltage, current and temperature limits over the intended loads, bands and power. Keep compensation when it earns its place in that design. Do not ask it to conceal an overambitious frequency brief.
The Long Wire Adds Another Upper-Band Compromise
A wire that is a half wavelength on its lowest band becomes several half wavelengths long on harmonic bands. For an 80 m fundamental, it is approximately 3 wavelengths on 15 m and 4 wavelengths on 10 m; for a 40 m fundamental, approximately 1.5 and 2 wavelengths. End effects and installation shift the exact resonances.
As electrical length increases, the radiation pattern develops more lobes and nulls. That can make one path much stronger and another much weaker without a corresponding change in total radiated power. “Pattern stability collapsed” and “efficiency collapsed” are therefore different findings:
- Efficiency needs accepted versus radiated power, conductor/ground-loss modeling, calorimetry or a validated field method.
- Pattern needs a validated electromagnetic model or controlled angular field measurement.
- One signal report includes propagation, remote antenna, polarization, noise, fading and the selected lobe/null.
For a station that values dependable coverage of chosen paths, a dedicated shorter radiator on an upper band may be a better starting geometry than relying on whichever lobe the long low-band wire happens to provide. That is a pattern choice, not an automatic efficiency gain. A dual-band EFHW still needs both of its patterns checked; limiting the band brief makes that a smaller job.
Lawrence Livermore National Laboratory’s NEC is an established method-of-moments tool for wire antennas and ground interaction. A useful EFHW model must include the actual wire geometry, conductor, ground, transformer equivalent circuit and common-mode feed structure; a perfect source placed on an isolated wire cannot determine transformer or feed-line loss.
The Return Path and Common-Mode Boundary Must Be Defined
An end-fed radiator is not a one-terminal circuit. Displacement current and current on a counter-conductor complete the electromagnetic system. Depending on the installation, that counter-conductor can include a deliberate wire, transformer capacitance, the coax exterior, equipment bonding and nearby structures.
That does not make all outside-shield current “heat.” Common-mode current can:
- radiate and alter the intended pattern;
- dissipate in a lossy choke, soil, building wiring or resistive conductors;
- change feedpoint impedance and shack-end SWR;
- couple local noise into the receive system; or
- create RF exposure, touch-voltage and interference problems.
A choke changes that current path and itself has a complex common-mode impedance and a power/temperature limit. The 2024 ARRL article “Common-Mode Chokes” recommends measuring feed-line current and retesting after installation; it also shows why choke response is frequency-dependent and can fall above self-resonance.
No universal 0.05λ rule: a fixed counterpoise length or choke distance can become an intentional part of one design, but it is not a general cure. Choose the common-mode boundary from a model or current sweep, install a choke with adequate complex impedance and thermal margin over the required bands, then measure again.
For Broad Multiband Use, Consider Changing the Feed Problem
Off-center-fed dipoles, loops and verticals do not share one nominal 100–200 Ω feedpoint or one matching topology. Their feedpoint impedances and balance properties depend on geometry, feed position, ground/radial system, height and frequency:
- For an off-center-fed installation whose transformer ports and load arrangement are intentionally unbalanced, my practical default is an appropriate UNUN for impedance transformation plus a separately specified choke for common-mode control. Choose the ratio from the actual complex load and the choke boundary from the intended return path. A suitable current balun remains a valid alternative for a balanced installed load; the antenna’s label alone does not decide the choice.
- A full-wave loop’s feedpoint impedance changes with shape, height, polarization and feed position.
- A quarter-wave vertical over an adequate radial system is normally below 50 Ω before matching; other verticals can be much higher or reactive.
- A 9:1 random-wire transformer does not create resonance. If a shack tuner matches through a high-SWR coax run, that line can add substantial loss; whether moving the tuner helps depends on the actual line and remote matching conditions.
A suitably chosen feedpoint can present a more moderate impedance than a kilohm-class end feed. That reduces the required voltage transformation: in the ideal model, 4:1 impedance means 2:1 voltage/turns, while 49:1 means 7:1. The useful advantage is a less extreme transformation problem when the antenna actually supplies the appropriate load. Fewer turns alone are not proof of lower core loss, and a 4:1 box is not a drop-in replacement for a 49:1 end-feed transformer.
Lower transformation ratio can simplify one transformer design, but it does not guarantee a lower-loss antenna system. Ground loss, loading, feed-line length, tuner topology and radiation pattern can outweigh the ratio. I would rather select an appropriate feed arrangement—or accept a well-placed tuner with known loss—than assume another shunt adjustment has made a high-ratio box equally suitable everywhere.
Why I Choose a Smaller Job for the EFHW
For a station centred on one band, choose the radiator, transformer and return arrangement around that band. For two related bands, deliberately design and qualify the pair. Groupings such as 160/80 m, 80/40 m or 40/20 m are examples of a bounded brief, not a promise that any long wire and transformer will automatically cover both.
The benefit is concrete: the same winding does not have to satisfy the low-frequency magnetising requirement and the upper-HF parasitic limits of an 80–10 m assembly at once. Material, turns and compensation can be chosen for a narrower set of loads, and the thermal and voltage corners are easier to identify. That gives the designer more room to optimise what the operator actually uses.
A single-band label is not a quality certificate, and a two-band design can still be poor. But it removes part of the compromise before trying to repair it. If you genuinely need many bands from one wire, treat that convenience as a declared system compromise and require the measurements to cover every band you intend to transmit on.
How to Measure an EFHW Transformer
- Define the ports and efficiency: specify whether efficiency means output power divided by accepted input power or transducer gain including mismatch.
- Use the intended complex load: a 2450 Ω resistor is one reference, not every EFHW. Characterize the resistor’s RF resistance and parasitics over frequency.
- Calibrate to the real reference planes: adapters, lead-ins, PCB traces and high-voltage fixtures must be characterized or de-embedded. Rohde & Schwarz’s fixture guidance explains why a coaxial VNA calibration stops at its calibration plane.
- Measure S11 and transmission: return loss alone cannot separate transformation from dissipation. Correct raw transmission for fixture, load and mismatch contributions.
- Treat back-to-back tests carefully: halving cascade loss in dB assumes comparable, reciprocal transformers and a controlled intermediate interface. It is a useful cross-check, not an automatic proof.
- Verify at operating power: low-level VNA data do not establish heating, voltage breakdown, core nonlinearity or duty-cycle rating. Use calibrated forward/output power or calorimetry with a controlled load and temperature.
- Report uncertainty: include source and sensor calibration, directivity/dynamic range, load tolerance and parasitics, fixture repeatability, mismatch, connector loss, temperature and stabilization time.
Dan Koellen, AI6XG, demonstrates the accounting problem in his EFHW transformer measurement study: raw S21 through a transformer-plus-load fixture includes transformer, load and mismatch terms, and assuming an ideal high-value resistor can even produce impossible efficiency results.
A Station-Level Test Plan
| Question | Measurement | Useful result |
|---|---|---|
| Is the coax dominant? | Measure or calculate loss for the exact cable and connectors at frequency, then include the actual load mismatch. | Matched and total line loss with reference planes. |
| Is the transformer dominant? | Calibrated/de-embedded transmission and return loss into representative loads; confirm thermally at power. | Loss versus frequency, load, power, duty cycle and temperature. |
| Is common-mode current significant? | Clamp-on RF current sweep along the feed line before and after the proposed choke/counterpoise change. | Current magnitude/distribution and change—not an assumed watt value. |
| Is the antenna dissipative? | Validated model including real ground/conductor loading, or controlled efficiency/field measurement. | Accepted-to-radiated efficiency with uncertainty. |
| Is the path sitting in a null? | Pattern model or controlled multi-angle field comparison using unchanged power and propagation conditions. | Gain by direction and polarization. |
Fix the Dominant Loss—and Stop Asking One Box to Do Everything
Account separately for the coax, high-ratio transformer, matching network and any common-mode path. Include cable mismatch in the transmission-line calculation, define the common-mode boundary with current measurements, and keep directional pattern nulls separate from dissipated power. If the cable is the largest measured loss, improve the cable. If the transformer is the problem, another spool of low-loss coax cannot repair it.
A warm matching box is evidence that some power is being dissipated there, not a measurement of how much and not proof that the coax is lossless. Conversely, a cool box in a brief low-power test does not establish its high-duty rating. The loss budget tells you where improvement is available; the declared band and load range tells you whether the design is being asked the right question.
My conclusion: I favour a purpose-built monoband or bounded dual-band EFHW when predictable matching and a manageable design envelope are the priorities. For broad multiband operation, either demand complete evidence for the high-ratio system or choose a feed arrangement that reduces the transformation burden. Do not let a reassuring SWR reading make the matching box untouchable while the coax takes all the blame.
Mini-FAQ
- Why do you prefer single- or dual-band EFHW designs? They give the radiator, transformer and return system a smaller frequency/load problem. That reduces the conflict between low-frequency magnetising requirements and upper-frequency parasitics, without guaranteeing the quality of any particular build.
- Does an 80–10 m band list prove broadband operation? No. The span exceeds three octaves, and separated usable antenna windows are not continuous broadband response. The matching assembly needs loss, voltage and thermal evidence across the actual intended loads and bands.
- Is coax loss affected by SWR? Yes. A lossy coax has baseline matched attenuation, and mismatch generally adds conductor and dielectric loss by changing the voltage/current distribution. The increase depends on line loss, electrical length and complex load—not SWR alone.
- Does low SWR prove an EFHW transformer is efficient? No. Return loss says how much wave is reflected at one plane. A lossy transformer can show a good input match, so transmission or calorimetric evidence is also required.
- Is a shunt compensation capacitor always a bad idea? No. It can improve a chosen response region, but it changes current, voltage and frequency dependence. Better input match is not by itself proof of lower loss or wider safe operating bandwidth.
- Do all 49:1 transformers fail on the upper HF bands? No. Their useful range depends on material, core geometry, winding, parasitics, load, power and temperature. The objection is to assuming all-band performance without qualifying that particular assembly.
- Is common-mode current on the coax automatically wasted power? No. It may radiate, dissipate, alter the intended pattern, couple noise or create EMC and RF-safety problems. Current and field measurements are needed to classify its effect.
- Where should an EFHW common-mode choke go? There is no universal distance. Choose a boundary using the intended antenna model or a feed-line current sweep, use adequate complex choke impedance and thermal margin, then verify the installed current.
- How should coax and transformer loss be compared? Put both in the same calibrated power budget at defined reference planes, frequency, load, power and temperature. Keep matched line loss, extra mismatch-related line loss and transformer dissipation separate.