Noise Isn’t “E or H” — It’s Coupling

I reject the E-versus-H argument because it starts one step too late. Interference needs a source, a transfer path and a susceptible victim. Remove any one of those—or reduce the transferred signal below the victim’s response threshold—and the symptom changes. That is a model we can test.

Start with Source, Path and Victim

• Source: the device or process producing unwanted spectral energy—perhaps a switching supply, LED driver, solar inverter, Ethernet interface, arcing contact or the station transmitter.
• Coupling path: the complete route by which current, voltage or field reaches the victim, including its return path. It may be conducted, common-impedance, capacitive, inductive, radiated, or a sequence of several.
• Victim: the port or circuit that responds—an antenna and receiver input, an audio amplifier, USB cable and computer, control line, power supply or nearby consumer device.

Frequency overlap matters, but so do level, modulation, timing, impedance and operating state. A source can be strong without causing a symptom when the transfer path is weak or the victim is sufficiently immune. Conversely, a modest signal can cause trouble at a nonlinear or poorly filtered input.

The words “conducted” and “radiated” describe parts of the route, not mutually exclusive worlds. A switching current can couple capacitively onto a mains lead, travel as a conducted current, make the lead radiate, and then arrive as common-mode current on the receiver feed system. The dominant step is the one worth changing.

E and H Describe the Field, Not the Whole Route

Close to compact circuitry, two lumped approximations are often useful:

The first says that changing voltage across a mutual capacitance can drive current into another conductor. The second says that changing current in one loop can induce voltage in another through mutual inductance. Both need a complete circuit: coupled current still requires a return path, and induced voltage matters only through the impedance and loop presented by the victim.

These are quasi-static, lumped models—not proof that a disturbance is “pure E” or “pure H.” Mutual capacitance and inductance depend on conductor area, separation, orientation, surrounding material, shielding, source impedance, victim impedance and frequency. Once dimensions are no longer electrically small, distributed transmission-line and radiation effects join the problem.

There Is No Universal Near-Field Radius for a Shack

IEEE antenna terminology distinguishes reactive near-field, radiating near-field and far-field regions. Their boundaries depend on wavelength and on the source’s electrical size and geometry. The familiar λ/(2π) distance is a useful scale for the reactive field of some electrically small radiators; it is not a universal boundary that makes every object inside it an H-field or E-field problem.

Large apertures can have an extended radiating near-field region. Small loops, short dipoles, long cables, enclosures with slots and house wiring all produce different local E/H relationships. Nearby conductors, earth and walls also disturb the field. In the far field of a wave in a homogeneous medium, E and H are linked by the wave impedance; close to structures they need not have that plane-wave ratio.

So I do not diagnose a whole room from one wavelength calculation. I identify the structure carrying current or voltage and test how geometry, termination and cable state change the symptom.

Keep Conducted and Radiated Paths in the Same Diagram

A handheld near-field probe is excellent for finding a hotspot or comparing A/B changes. Its reading is not automatically calibrated field strength, radiated power or compliance evidence. Probe orientation, height, loading, cable routing, preamplifier state, resolution bandwidth and detector settings all belong in the record.

Common Mode and Differential Mode Need a Reference

On an intended two-conductor path, differential-mode current flows out on one conductor and returns on the other. Common-mode current is the net current of the conductors together relative to another reference or return structure. Real installations can carry both at once.

In coax, the intended TEM signal uses equal and opposite currents on the centre conductor and the shield’s inner surface. Current on the shield exterior closes through the antenna, chassis, earth, other cables and stray capacitance. That exterior path can change antenna pattern, import local noise or export transmitter RF, but its magnitude and consequence are installation-specific.

Asymmetry converts modes. Unequal conductor geometry, an unbalanced termination, shield discontinuity, chassis connection or stray capacitance can turn some differential energy into common mode and vice versa. “The cable is balanced” or “the coax is matched” does not prove that exterior current is absent.

A calibrated current probe around an entire multi-conductor cable responds to the net current through its aperture; equal and opposite differential currents largely cancel. A probe around one conductor sees that conductor’s combined modal current. Frequency response, aperture position, probe transfer impedance, loading and instrument noise floor limit either measurement.

The Victim Decides Which Coupled Signal Matters

A receiver can respond to an in-band signal through its intended antenna port, to out-of-band energy through front-end overload or intermodulation, or to common-mode current that reaches the chassis and converts into differential voltage. Audio, USB and control circuits can rectify RF at nonlinear junctions or fail when coupled RF crosses a logic or reset threshold.

This is why an S-meter or waterfall alone does not locate the path. Keep receiver frequency, mode, bandwidth, preamp, attenuation, AGC, reference level and antenna state fixed. Record wanted-signal level, noise in a declared bandwidth and overload indications. A falling displayed noise floor is not automatically a better SNR if the wanted signal or receiver gain changed with it.

Immunity is port- and state-specific. IEC separates standardised radiated-field immunity from RF disturbances coupled onto cables because the fixtures and reproducible test methods differ. A passing result under one product standard, port configuration and test level does not guarantee immunity in every amateur installation.

A Controlled Coupling-Path Sequence

Define One Symptom and One Metric

Choose something repeatable: noise power in a stated receiver bandwidth, wanted-signal SNR, RF current at a cable location, audio output, packet error count, or the transmit power at which a reset occurs. Photograph or save the instrument state and cable layout. Do not change several controls at once.

Prove the Source with A/B/A

Switch one suspected device between its normal and inactive state, then restore it. Match the change in spectrum and symptom at least twice. Repeat across the frequencies or operating states in which the device emits. A portable receiver or local probe can help, but proximity alone does not prove the route to the station.

For mains-powered equipment, use normal controls or isolate a circuit only when it is safe and permitted. Do not open live equipment, defeat interlocks, lift a protective-earth conductor or remove a grounding pin as an RF experiment. Protective earthing remains a safety function even when it also carries RF current.

Measure at the Victim Boundary

Start where the unwanted energy enters the victim: antenna connector, feedline exterior, DC input, USB lead, audio lead or mains port. Compare the same point and instrument settings in A/B/A states. A 50-ohm termination can distinguish antenna-port pickup from some receiver-internal spurs, while a whole-cable current probe can reveal net cable current. Neither test identifies every path by itself.

Break One Path Without Creating Another

Substitute one element at a time: a known load for an antenna, a battery for an external supply, a shorter cable, an optical link, a shielded enclosure, a defined filter, or a common-mode impedance at one declared location. Record both the symptom and the current or voltage that the change was meant to affect.

Galvanic isolation breaks a DC conductive connection, not necessarily an RF path. Transformer capacitance, cable shields, power supplies and nearby metal can bridge the isolation barrier. Likewise, a filter fitted after the nonlinear or susceptible stage cannot protect that stage.

Change Geometry in Repeatable Steps

Move or rotate one cable, loop, source or probe through recorded positions. Keep cable length, termination and receiver settings fixed. There is no universal “move it 30 cm” threshold: electric-dipole, magnetic-loop and distributed cable fields have different distance and orientation behaviour, especially around conductive structures.

Confirm Across Frequency, Level and Time

Repeat at another band or device operating condition and at the lowest transmitter power that reproduces a TX symptom. If a proposed fix works only at one frequency, map its useful and harmful ranges. The final evidence should show that the intended path quantity changed and the victim response followed it.

Three Station Symptoms, Translated into Paths

Home Noise That Disappears in the Field

The source may be local electronics, but the route might be direct radiation, conducted current on house wiring, re-radiation from that wiring, feedline-exterior common mode or several together. Hold receiver gain and bandwidth fixed. Switch one safe source at a time, compare antenna and terminated-input states, measure whole-cable current where possible, and repeat the winning change A/B/A.

A Computer or Audio Interface Fails on Transmit

The transmitter is the known source. Compare the antenna with a correctly rated dummy load, using the minimum power needed for diagnosis and respecting equipment and RF-exposure limits. Then inspect common-mode current on feedline, USB, audio and DC cables. A ferrite at the victim entry, at the source or elsewhere tests different path segments; placement is part of the result.

A Nearby Consumer Device Demodulates the Transmission

First verify the station’s wanted and unwanted emissions with suitable measurement equipment and a defined load. Then, with the device owner’s permission, test which victim port admits RF. Speaker, mains, antenna and data leads can each be the receiving structure. Cooperation and one-variable changes are more informative than assuming either transmitter fault or victim fault from the audible symptom alone.

Why No Single Choke, Ground or Shield Is the Answer

• A common-mode choke helps only when common-mode current passes through it and its complex impedance is useful at the problem frequency. Winding capacitance, resonance, material, turns, cable geometry, voltage, current and temperature bound the result. It may also introduce unwanted differential impedance or a new resonance.
• A bond or “RF ground” changes a current network; it can reduce one impedance and strengthen another path. Keep protective bonding compliant, then measure RF current and voltage rather than assuming the shortest-looking strap is low impedance at every frequency.
• A shield needs suitable material, continuity, aperture control and a defined termination. An electrostatic screen can redirect displacement current; it does not automatically stop magnetic coupling or cable current.
• An isolation transformer or optical link removes only the paths it actually interrupts. Parasitic capacitance, shields and parallel power connections may remain.
• A filter needs the correct mode, source/load impedance, frequency range, voltage/current rating and physical placement. Bench insertion loss in 50 ohms does not by itself predict an installed common-mode result.

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