HF Baluns: Why I Usually Separate Matching and Choking

For decades, current-balun transformers have been offered as universal “antenna balancers.” My objection is practical: the antenna is not suspended in a textbook. Ground, nearby metal, unequal coupling and the route back to the shack become part of the system. After roughly twenty years working with HF baluns and UNUNs, I do not assume that a tidy drawing has survived intact in the garden.

My usual approach is to separate impedance transformation from common-mode control. When the installed load calls for a 4:1 ratio and the transformer ports are intentionally unbalanced, I use an UNUN for that impedance step and specify a 1:1 current choke for the return-path boundary. When no ratio is needed, the 1:1 choke may be the only interface required. I do not add a transformer merely because an antenna is called balanced.

Why Keeping the Jobs Separate Helps

The wanted signal needs a suitable differential impedance path. The outside of the coax, mast and station wiring create a different problem: where unwanted or deliberately used return current can flow. Those two problems need not have the same optimum circuit or physical boundary.

Separating the jobs lets me choose the impedance ratio without assuming that the transformer's common-mode response is also right for the installation. I can specify the choke's useful frequency range, installed position and voltage/current duty independently. Where a defined coax-exterior section is part of an end-fed return, the choke marks its intended end; where the feedline should not participate, the boundary belongs at that transition instead. This is flexibility in the design—not permission to scatter chokes at arbitrary distances.

The pair can also form a valid hybrid for a genuinely balanced load when the interconnection, isolation and complete assembly support that function. An UNUN plus a choke somewhere along an arbitrary cable is not automatically such a hybrid. Nor are two boxes automatically lower-loss than one: connections, parasitic coupling and both components' stress limits still count.

A good current balun can itself provide transformation and common-mode isolation, including in an unbalanced application. Tom Rauch/W8JI explicitly discusses that capability in Properly Testing Baluns. He also warns that a low-SWR test does not establish balance. That supports looking at the complete current path; it is not his endorsement of my exact UNUN-and-choke arrangement.

So my preference is not “current baluns cannot do both.” It is “I want both jobs specified rather than assumed.” If one verified current-balun assembly meets them, it is a legitimate solution. In the irregular multiband installations that prompted this discussion, separating the jobs is often the more useful starting point for me.

“Current Balun Transformer” Bundles Two Different Jobs

Balun describes a balanced-to-unbalanced port function. It does not identify one winding. A 1:1 current balun or common-mode choke adds impedance to an unwanted common-mode path while passing the wanted differential mode. A Guanella transmission-line transformer can also connect several line sections in series and parallel to provide an impedance ratio. Those are related current-mode ideas, but they are not interchangeable specifications.

Guanella’s high-frequency matching-transformer patent and Ruthroff’s 1959 broadband-transformer paper are primary reminders that topology, line impedance and electrical length create the behaviour. The word on a case cannot replace the circuit.

A Ratio Is an Impedance Relationship, Not a Load Guess

An ideal impedance transformer with ratio n maps a load Z_L to:

A 4:1 impedance transformer maps 200 Ω to 50 Ω. It maps 300 Ω to 75 Ω, not 50 Ω. That may still be a perfectly acceptable 1.5:1 SWR on a 50 Ω system, but it is not an exact match. A nominal 6:1 impedance ratio would map exactly 300 Ω to 50 Ω in the ideal resistive case; whether that is the best physical network is a separate bandwidth, loss, balance and stress decision.

Real antenna impedance is complex and frequency-dependent. If Z_L = R + jX, the transformer maps both terms and its own non-ideal input impedance, leakage, line delay and capacitance join the result. Select a ratio from the measured load envelope at the device plane—not from one resistance remembered from a handbook.

Folded Dipoles: A Natural Case for a Ratio Transformer

For a thin, closely spaced, equal-conductor folded dipole in the same ideal environment as a simple dipole, the familiar result is an input impedance about four times the corresponding simple-dipole impedance. That is why values near 280–300 Ω are common teaching numbers. Guertler’s primary folded-dipole analysis also shows why unequal conductor diameters change the transformation ratio.

Installation moves the number. Conductor diameter and spacing, feed gap, electrical length, height, ground, nearby metal, boom or mast, added elements and weather all change R + jX. A folded driven element in a Yagi can be deliberately used to transform a low array feed impedance; it need not resemble an isolated 300 Ω folded dipole.

This is why the single-band folded dipole remains one of my clear examples: its feed impedance can make a ratio transformer a compact, purposeful choice, rather than an accessory added by habit. A 4:1 Guanella device can provide both the useful impedance step and common-mode impedance when its load and operating range suit. Balanced line to a tuner or another matching network remains an alternative when the installation asks for something different.

Full-Wave Loops: A Ratio Transformer or a Matching Section

A resonant full-wave loop often presents a resistance in the broad neighbourhood used in the familiar 100–120 Ω example. Shape, feedpoint position, conductor size, height, ground, support and nearby structures can move both resistance and reactance materially. The current distribution and input impedance are part of the solved antenna geometry, as shown in the modern primary loop-antenna theory and validation.

A 2:1 impedance transformer maps an ideal 100 Ω load to 50 Ω and an ideal 120 Ω load to 60 Ω. That may be useful across a declared band, but it does not prove the loop stays within that range or that the transformer remains balanced and cool.

A 75 Ω quarter-wave section offers another bounded case. At exactly 90° electrical length on a lossless line:

Between 50 Ω and 100 Ω, the ideal section is 70.71 Ω. A practical 75 Ω section is exact for 50 Ω to 112.5 Ω. Its bounded arithmetic is:

Those are lossless, exact-quarter-wave, purely resistive cases. A reactive loop load remains reactive after transformation, cable loss changes the result, and both loop impedance and electrical length vary with frequency. Keysight’s RF design treatment gives the ideal Z_t = √(Z_inZ_L) relationship and shows why multiple sections can broaden a specified response.

For a suitable single-band loop, I like the practical clarity of a 75 Ω matching section plus a separately specified choke: the line supplies the impedance transformation, while the choke controls the common-mode boundary. That keeps ferrite out of the impedance-transforming section, although not necessarily out of the entire feed system. A single coax section does not itself supply the needed balanced-to-unbalanced behaviour. A ratio current balun is the compact alternative when its combined performance suits the loop.

Cut the Matching Section for Its Electrical Length

For a line with manufacturer-verified velocity factor VF at frequency f, the first physical estimate is:

At 7.1 MHz with VF = 0.85, this gives approximately 8.97 m of physical cable: the 10.56 m free-space quarter wavelength multiplied by 0.85. Manufacturer VF is nominal; connector launch, cable construction, temperature and frequency introduce uncertainty. Cut with allowance, measure the section at its installed reference planes and trim only after the intended load and fixture are defined.

A quarter-wave section is frequency-selective, but “narrowband” is not a complete rating. Usable bandwidth follows the impedance ratio, permitted reflection, load dispersion, line loss and whether one or multiple sections are used. Likewise, a ferrite transmission-line transformer can be broadband: its low-frequency end is often limited by magnetising/common-mode impedance and its high-frequency end by electrical length, leakage and capacitance. Neither family owns a universal bandwidth verdict.

Low Loss and Cool QRO Operation Must Be Measured

A coaxial matching section contains no ferrite core, so it avoids ferrite loss and magnetic nonlinearity. That does not guarantee it “runs cool” or has more QRO headroom. The cable still has conductor and dielectric loss; standing waves create voltage and current maxima; connectors, bends, bundling, sunlight, altitude and ambient temperature change the limit.

A transformer can be limited by copper loss, core loss, insulation, common-mode dissipation, connectors or enclosure temperature. A line section can be limited by dielectric voltage, conductor current, connector heating or environmental derating. Times Microwave’s high-power cable guidance makes its power limits conditional on cable family, frequency, VSWR, ambient, altitude and installation. Fair-Rite’s material guidance likewise treats ferrite impedance as complex and dependent on frequency, temperature and bias.

For either solution, declare:

• forward, reflected, accepted or delivered power at a stated plane;
• frequency, complex load and electrical length;
• waveform, crest factor, average power and duty cycle;
• local differential voltage and current maxima;
• common-mode current, voltage and dissipation;
• connector, insulation, core and conductor limits; and
• ambient, mounting, cooling, temperature rise and test uncertainty.

“QRO-rated” without those conditions is a label, not evidence.

When One Current Balun Covers Both Jobs

A properly implemented Guanella ratio transformer can present useful common-mode impedance as part of its operating principle. It does not automatically require a separate 1:1 choke. Nor does the word Guanella prove that the finished device has enough Z_CM over every required band.

Measure the assembly’s common-mode impedance as a complex quantity:

The loss expression is a first-order sinusoidal estimate using a consistent port-current definition. Real ferrite loss and temperature can be distributed and drive-dependent. Also measure differential insertion loss and match, amplitude/phase balance or mixed-mode conversion, and installed feedline current.

A separate choke is useful when the ratio transformer does not provide sufficient installed common-mode control, when another path boundary is required, or when assigning the functions to distinct components gives the needed operating margin. That is the same separation-of-functions reasoning behind my default. If one current-balun assembly already meets both requirements, adding another component is not a condition of correctness.

Choke Placement Follows the Unwanted Path

The fixed “0.05–0.10λ down the coax” rule cannot serve every OCF dipole, loop, end-fed wire, Yagi or vertical. Moving a choke changes the length and impedance of the exterior-feedline section that remains connected to the antenna. It can move a common-mode current maximum, voltage maximum or resonance without eliminating the complete path.

The ARRL installed common-mode current procedure demonstrates the right loop: measure along the actual cable, change the impedance or placement, and measure again. One convenient current minimum is not proof of a quiet feed line.

How I Apply the Default to Other HF Antennas

Balanced and unbalanced describe port relationships to a reference, not moral properties of antenna shapes. At an isolated two-terminal feedpoint, current entering one terminal leaves the other. Common mode appears when the completed installation provides additional asymmetric paths through feedline exterior, mast, soil, equipment, wiring or displacement current.

• Off-centre-fed dipole: my usual starting point is an appropriate UNUN plus a separately specified choke when the installed port/return arrangement is intentionally unbalanced and the measured load supports the ratio. A 4:1 current balun can also be a valid implementation; W8JI describes such an OCF example. Unequal arms are a reason to examine the environment and return path, not a proof that one topology must fail.
• End-fed half-wave or non-resonant wire: use the transformation appropriate to that particular high or complex feed impedance, together with a defined return/counterpoise and common-mode boundary. This is a natural application of the separate-jobs approach; neither 49:1 nor 9:1 belongs on every wire simply because it is end-fed.
• Yagi: if the driven-element matching system already provides the required impedance, I concentrate on the 1:1 common-mode interface rather than automatically adding a ratio transformer. Folded, gamma-matched and other feed arrangements have different requirements; preserve the actual design.
• Vertical: the radiator-to-radial/ground port is usually unbalanced. Provide matching only as needed, retain the intended radial or ground return, and control unintended feedline-exterior current. The word “unbalanced” does not mean no choke is needed.
• Multiband centre-fed wire: a suitable balanced line and tuner can carry and transform the changing load, with common-mode control at the appropriate transition. A fixed ratio chosen from one band need not help on the others. A broadband current transformer is still a valid option when its complete load and stress envelope fits.

The practical benefit is that each component has a reason to be there. I am not asking a ratio transformer to repair the geometry of a garden, and I am not asking an UNUN to eliminate a return path. I assign the wanted impedance transformation and common-mode control explicitly, then check that their interaction produces the intended system.

Use a Function-and-Evidence Decision

Test small-signal behaviour first, then increase power in controlled steps using representative frequency, load, waveform and duty cycle. De-energise before moving the line or opening an enclosure. Stop on arcing, odour, rapid heating, unstable match, unexpected common-mode current or electrical drift. Recheck the small-signal data after the power test.

Practical Conclusion

In most of the real HF installations I work on, I start with a suitable impedance transformer—often an UNUN—and a separately specified current choke. If no transformation is needed, I do not invent a reason for a ratio transformer. That is the practical position: the ground, feedline and surroundings rarely preserve the symmetry suggested by the antenna sketch, so transformation and return-current control deserve individual attention.

Single-band folded dipoles and full-wave loops remain two clear examples where an impedance-transforming current balun can be an elegant choice. They are not an exhaustive list, and a good current balun may satisfy both jobs in other balanced or unbalanced installations too. What I reject is the universal “antenna balancer” assumption—not a circuit that actually works.

Choose the ratio because the load calls for it. Choose the common-mode boundary because the installed current path calls for it. Then verify the combined loss and stress. That leaves me with a reasoned design choice, not just another box on the coax.

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