Can Coax Between Radials Bypass a Common-Mode Choke?

The myth sounds tidy: radial wires are “RF ground,” so a feedline laid between them cannot participate in the antenna and cannot get around a choke. Real radial wires carry current and develop position-dependent RF voltage. The coax exterior, soil, mast and every connected conductor then form another coupled network. Geometry matters—but geometry is not a verdict.

The Radial Field Is Part of the Antenna

A base-fed quarter-wave vertical is a monopole system. Current on the radiator must close through its earth system and surrounding displacement-current paths. Over an ideal infinite conductor, image theory gives a convenient symmetric model. A practical installation instead uses finite radial wires, lossy soil and nearby conductors.

ITU-R BS.705-2 treats a vertical monopole, the earth constants and a radial-wire earth system as explicit antenna inputs. LLNL’s NEC-5 validation manual likewise models a quarter-wave monopole over a buried radial-wire screen and shows sensitivity to radial-screen size and ground parameters.

That is why “all radials sit at zero RF potential” is not a valid general boundary. Current magnitude and phase vary along each wire. Soil conductivity and permittivity, radial length, count, height or burial, connection resistance and azimuthal symmetry all change the distributed field. The hub may be a useful reference point, but the entire radial field is not one equipotential circuit node.

The Coax Supports Two Relevant Current Descriptions

In the intended coaxial TEM mode, forward and return currents flow on the centre conductor and the inner surface of the shield. For an ideal line cross-section their algebraic sum is zero, and the field is largely confined between the conductors.

The outer surface of the shield can simultaneously carry current with respect to the radial system, soil, mast, station and surrounding field. That exterior current is the common-mode quantity of interest here. It can be launched by feedpoint asymmetry, coupled onto the cable farther along its route or converted at a connector, enclosure, mast bond or other discontinuity.

A good differential SWR does not prove the exterior current is small. Conversely, exterior current can change while the transmitter sees only a modest impedance change. These are different measurements at different modal reference planes.

“Bypass” Is a Stronger Claim Than “Couples”

In a lumped circuit, a bypass path connects across the same two nodes as the component it bypasses. A feedline near radials certainly has capacitance to them and to the soil. It also has inductive and field coupling. But that coupling is distributed along structures whose RF voltage and current vary with position.

Replacing the whole installation with one capacitor directly across one choke is therefore valid only when:

• the electrically relevant structures are short enough for a lumped approximation;
• the two nodes on both sides of the choke are defined;
• the effective capacitance between those same nodes is measured or credibly modelled;
• the choke’s full complex impedance at that frequency is used; and
• the remaining radial, soil, mast, cable and station paths are either included or shown negligible.

Without those conditions, capacitance is still real, but the phrase “sits directly across the choke” may describe the wrong circuit.

The Picofarad Arithmetic Is an Illustration

For an ideal lumped capacitor, magnitude of reactance is:

If—and only if—one of those capacitors is directly across the same two nodes as an ideal 5 kΩ resistive choke, the parallel combination will be dominated by the capacitive branch at these frequencies. That calculation is useful because it shows why small capacitance can matter. It does not establish that a particular coax route has 20 pF or 50 pF, that the capacitance is lumped, or that 5 kΩ is purely resistive and unchanged after installation.

The often-quoted 20 pF example gives about 796 Ω at 10 MHz. The arithmetic is fine. The engineering task is proving the capacitance, nodes and distributed reduction for the actual vertical.

A Choke Is a Frequency-Dependent Complex Component

The exterior-mode impedance of a choke is Z_CM(f) = R(f) + jX(f), not one resistance printed for every band. Ferrite material, core geometry, number of passes, coax dimensions, turn spacing, connectors and measurement fixture all affect that curve. Transmit use adds voltage, current, dissipation and temperature limits.

Every winding also has distributed capacitance. With inductance it produces self-resonant behaviour; above the useful region, impedance can fall and become increasingly capacitive. Würth Elektronik’s measured common-mode-choke models show how even small winding capacitance changes response near resonance. Fair-Rite’s impedance-measurement guidance separately emphasizes vector resistance and reactance plus the influence of the test conductor and fixture.

Keep two capacitances conceptually separate:

• choke self-capacitance, internal to the wound structure and its terminals; and
• installation coupling, distributed between the coax exterior, radial field, soil, mast and other conductors.

Both can limit isolation. They are not interchangeable values.

Feedpoint Placement Usually Helps—but Does Not Prove Zero Current

Putting a suitable choke at the feedpoint often has a strong advantage: it inserts exterior-mode impedance close to the conversion region and minimizes the unchoked coax segment attached to the radial hub. If a choke is placed farther down the line, the segment between feedpoint and choke becomes an explicit conductor in the exterior-mode network.

That segment does not automatically radiate strongly, nor does it become a literal extra radial in every installation. Its current depends on electrical length, route, radial and soil coupling, mast connection, choke impedance and what terminates the cable farther away. At some frequencies it may carry little current; at others it can become a significant part of the installed antenna.

A feedpoint choke also cannot stop every possible excitation:

• finite choke impedance leaves residual current;
• self-capacitance can reduce impedance at some frequencies;
• a metal mast, control cable, protection conductor or enclosure can provide another RF path;
• the field can excite the feedline downstream of the choke; and
• shield and connector imperfections permit finite mode conversion, described for complete assemblies through surface transfer impedance.

IEC 60512-23-3 expresses connector-assembly shielding through the voltage induced inside relative to current on the outside housing. It is a useful reminder that “coax” is not an infinitely perfect wall and that connectors, bonds and accessories belong inside the RF boundary.

There Is No Universal 35-Centimetre Limit

Capacitance per unit length depends on conductor diameters, spacing, dielectric, height, burial, radial density, soil moisture and surrounding metal. Electrical length also scales with frequency and propagation velocity. A fixed physical distance cannot represent all those variables across HF.

The right question is not “Is the choke within 35 cm?” It is:

• How much exterior current exists on each side of the choke across the operating band?
• What common-mode impedance does the installed choke present there?
• How does moving the choke or route alter current distribution, feedpoint impedance and field pattern?
• Does another conductor carry the current that disappeared from the coax?

A short pigtail may be electrically benign in one installation and decisive in another. Measure before turning its length into doctrine.

Model the Structure You Actually Built

A useful electromagnetic model includes more than a vertical wire and an ideal ground symbol:

• radiator length, diameter, base geometry and loading;
• radial count, length, diameter, height or burial and connection topology;
• soil conductivity and relative permittivity over a plausible range;
• the coax exterior as a conductor with its real route and length;
• choke location and measured complex impedance as a frequency-dependent load;
• mast, guys, control cables, bonding and nearby conductive structures; and
• the station-end common-mode termination rather than an unexplained perfect ground.

LLNL’s NEC-5 documentation confirms that wire structures, homogeneous ground, loads, networks and transmission lines can be included and that model currents, near fields and patterns are available. For this problem, inspect exterior-conductor current and pattern as well as input impedance. A neat SWR curve alone cannot validate the modal model.

Use a One-Variable A/B/A Current Test

A clamp-on RF current probe around the complete coax responds to the algebraic sum of the enclosed conductor currents. The intended centre/inner-shield currents largely cancel; the remaining signal is therefore a practical measure of net exterior/common-mode current, subject to probe calibration, loading and frequency range.

Tektronix’s cable-current procedure recommends using the probe’s transfer-impedance calibration and moving it along the cable because standing-wave maxima occur at different locations. For an antenna comparison, convert that into a controlled A/B/A test:

Establish condition A

• Use a low, repeatable transmitter or test-generator level within the probe, choke and site limits.
• Fix frequency, power reference plane, tuner state, radial layout, coax length, station termination and cable supports.
• Mark several probe positions: immediately above and below the choke where accessible, then along the feedline and relevant mast or auxiliary conductors with the correct probe.
• Keep probe orientation, closure, receiver settings, bandwidth and calibration constant.
• Record complex feedpoint impedance or SWR, soil/weather state and the current magnitude at every marked position.

Change exactly one variable for condition B

Move the same choke while preserving coax length and route, or reroute the coax while preserving the same choke and termination, or substitute one measured choke at the same position. Do not move the choke, shorten the coax, change the radial layout and retune the antenna in one step; that experiment has no single cause.

Restore condition A

Return the changed item to its baseline state and repeat the map. If the second A result does not agree with the first within the measurement uncertainty, drift, connector repeatability, temperature, soil moisture or another uncontrolled variable has entered. Resolve that before accepting the B difference.

De-energize before moving a choke, route or bond, and do not remove required protective or lightning bonding for an RF experiment. Maintain controlled access and the applicable RF-exposure boundary; ITU-T K.52 provides the current installation-assessment framework for human exposure.

Interpret the Current Map Carefully

One low reading can be a standing-wave node. One high reading can be local maximum. A useful conclusion comes from multiple positions and frequencies, supported by known probe transfer impedance and repeatability.

If the claim concerns pattern, add a repeatable field-strength or pattern measurement. If it concerns receive noise, hold receiver gain, bandwidth, antenna switching and time as tightly as possible and record wanted signal and noise separately. Exterior-current evidence identifies a path; it does not automatically quantify the system consequence.

The Practical Installation Rule

Start with a measured, power-capable choke at the feedpoint because that usually gives the clearest modal boundary and the shortest uncontrolled exterior segment. Route the coax mechanically and safely, include its actual path among the radials or through the soil in the model, and map the current. Add or move another choke only when the measurement identifies where extra impedance helps.

The coax may run between radial wires and still carry exterior current. It may also run there with little exterior current. The radial route changes the network; it does not decide the result. That is the myth worth retiring.

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