VHF/UHF Hotspot and PCB Noise: A Hands-On EMC Investigation

A hotspot is a radio, a digital computer, several clocked buses, a power-conversion system and a bundle of cables sharing a very small space. A bare transceiver PCB adds the installation itself to that system. I troubleshoot it as a source–path–victim problem, one controlled change at a time.

Name the Symptom Before Opening the Toolbox

“Noise” is too broad for diagnosis. Record what the receiver actually shows or reproduces:

• Discrete carriers or a regular comb may track a clock, switcher, data burst or mixing product. Spacing is a clue, not identification.
• Broadband hash may come from fast edges, switch-mode conversion, digital traffic, arcing or several overlapping sources.
• Hum or buzz may be baseband coupling, supply ripple, rectified RF, periodic load current or a ground-reference problem.
• Pulsing interference can follow network activity, display refresh, processor load, transmit cycles, thermal control or an unrelated appliance.
• Desensitisation without an obvious spur can be receiver blocking or gain compression caused by a strong nearby signal.

Fix receiver frequency, mode, bandwidth, gain, squelch and antenna position. Save an audio sample, spectrum trace or signal-strength record with the device state, cable arrangement and power source. An automatic-gain-control system can make two very different RF conditions sound deceptively similar.

Think in Sources, Paths and Victims

A digital board can create disturbance energy in clocks, processors, switch-mode converters, displays, USB interfaces and network interfaces. That energy matters only when a coupling path delivers enough of it to a susceptible circuit or receiver.

These paths are not mutually exclusive. A converter can place differential ripple on a supply pair; imperfect balance can convert part of it to common mode; the cable can radiate; an enclosure seam can couple that field into the receiver. Treating only the last visible step may move the symptom without fixing the system.

Do Not Assume VHF Must Be Noisier Than UHF

A 2 m installation may show more interference than the same arrangement on 70 cm, but frequency alone does not prove why. The source spectrum, cable electrical lengths, board and enclosure resonances, antenna placement, front-end filtering, receiver linearity and wanted-signal level all change with band.

A one-metre lead is not automatically a half-wave antenna at 144 MHz. Its common-mode electrical length depends on the current path, dielectric loading, terminations and the other conductors connected to it. At 433 MHz, a different portion of the same assembly may resonate or couple more strongly. Measure both bands with the same receiver settings and installation state before calling one inherently cleaner.

Frontend differences also matter. A stronger out-of-band signal can block or mix in one receiver while remaining harmless in another. ITU-R documents this general problem for monitoring receivers: strong nearby fields can produce desensitisation, blocking and false responses. The first check is therefore whether the receiver is still linear.

Prove the Receiver Is Not Inventing the Problem

Insert a known attenuator at the receiver input and repeat the observation. A real external signal should fall by approximately the inserted loss while the receiver remains above its own noise floor. Intermodulation or compression products can change much faster, disappear abruptly or move relative to the wanted signal. That behaviour is evidence of a receiver limitation, not proof that the board has stopped emitting.

Repeat with a suitable band-pass filter or preselector ahead of the receiver. If an in-band symptom improves when a strong out-of-band signal is removed, the original result may have been overload. Keep the preamplifier off unless sensitivity is genuinely needed; a preamplifier can create its own distortion when strong signals are present.

Apply the same caution to a spectrum analyser. Increase input attenuation and disable its preamplifier to see whether a suspected harmonic or intermodulation product changes disproportionately. Protect the input from transmit power and transients. The analyser can generate a convincing spur internally when its mixer is overdriven.

Power Tests Isolate One Path, Not the Whole Environment

Running the device from a battery is a valuable test because it removes one mains-powered supply and its cable from the configuration. It does not automatically provide RF isolation. A power bank may contain its own switching converter, and USB, Ethernet, coax, programming or audio cables can preserve common reference and current paths.

Use a known, current-capable supply and confirm that the device remains within its voltage limits during receive, transmit and processor-load changes. A voltage sag or converter instability can create audio and digital symptoms that resemble external RFI.

When changing the supply:

• keep antenna, receiver settings, cable position and device state fixed;
• record supply voltage and current as well as RF or audio output;
• disconnect other conductive paths one at a time where safe;
• restore the original supply and confirm that the original symptom returns; and
• never defeat protective earth or modify a mains supply as a troubleshooting shortcut.

A cleaner result on battery means the changed power arrangement matters. It does not yet distinguish differential ripple, common-mode current, grounding, cable radiation or a change in physical layout.

Use Near-Field Probes as Locators

An H-field loop is useful near switching-current loops, inductors, high-current traces, cable launch points and enclosure seams. An E-field probe responds strongly near voltage nodes, clock traces, connectors and poorly contained electric fields. Probe size, height, angle and loading all affect the reading.

Start with a larger probe to find a region, then use a smaller probe for spatial resolution. Hold height and orientation constant when comparing changes. A probe maximum near a component does not prove that component is the original source: it may be carrying or reradiating energy generated elsewhere.

Do not place an improvised probe on hazardous mains circuitry. For conducted-emissions work on mains-powered equipment, use appropriate isolation, a suitable LISN or coupling network, rated probes and competent test practice.

Clocks and Harmonic Combs Need Correlation

Fast digital edges contain energy well beyond a clock’s fundamental frequency. Repetition rates, switcher frequencies and their harmonics can create regular spectral families, while data-dependent activity adds sidebands or raises a wider floor.

A frequency match is only a lead. Correlate it with a controlled state change: alter processor load, network traffic, display activity or a configurable clock while keeping receiver and geometry fixed. A candidate source becomes convincing when the spectral line follows that change, the local probe identifies a physical path, and restoration recreates the original result.

Also check whether the measuring instrument aliases or mixes the signal. Change span, resolution bandwidth, sample rate, input attenuation and preselection. A line that behaves with the instrument rather than with the device may be a measurement artefact.

Enclosures Work Through Continuity and Controlled Penetrations

A conductive enclosure can reduce electric-field coupling and provide a return surface for shielding current, but a metal box is not automatically an RF seal. Seams, lid contacts, apertures, displays, ventilation slots and unfiltered cables can dominate. A cable passing through the wall can conduct the disturbance around the enclosure.

Before adding foil or paint, identify whether the field is electric or magnetic, where current crosses the boundary, and whether the enclosure is safely bonded. A temporary seam bridge can be a useful experiment if it is electrically and mechanically safe. Restore it to prove repeatability.

Do not cover ventilation, short circuitry, detune an antenna, compromise insulation or create accessible RF current in the name of shielding. When a cable crosses the boundary, filtering or common-mode impedance usually belongs at that boundary so disturbance current is not allowed to flow across the interior before treatment.

Ferrite Must Match the Current Mode and Frequency

A ferrite around the whole cable acts on net common-mode current. It does not remove differential ripple between conductors inside that cable. A ferrite on only one conductor changes both differential- and common-mode impedances and must be analysed as part of that circuit.

Choose a core from its measured complex impedance over the frequency range of the actual disturbance. Material name alone is not enough; geometry, number of passes, winding capacitance, cable spacing, current, temperature and installation position all matter. More turns often raise low-frequency impedance, but parasitic capacitance and resonance can limit or reverse the benefit at VHF/UHF.

Place the ferrite where it interrupts the measured current path—commonly at a source or enclosure boundary—and verify cable current and receiver response on both sides. A quieter receiver after moving the cable is not proof that the core alone caused the improvement.

A Repeatable Bench-to-Field Sequence

1. Capture the symptom. Record frequency, bandwidth, receiver settings, audio or spectrum, device state and physical layout.
2. Check receiver linearity. Repeat with known attenuation and suitable preselection before treating every line as a real emission.
3. Separate antenna-borne and local paths. Replace the antenna with the correct shielded termination at the receiver reference plane. Interpret the change without assuming that it identifies one source.
4. Map conductive paths. List DC, USB, Ethernet, audio, coax, programming, chassis and protective-earth connections. Disconnect only safe, nonessential paths one at a time.
5. Compare power arrangements. Use a known supply and a battery configuration while holding geometry and operating state fixed.
6. Measure cable common mode. Clamp around the complete cable at marked positions and frequencies; do not infer current from cable presence alone.
7. Localise fields. Scan board regions, connectors and seams with repeatable E- and H-probe height and orientation.
8. Correlate operating states. Change network traffic, processor load, display, transmit state or switcher load and watch whether the suspect spectrum follows.
9. Apply one treatment. Add one filter, shield bond, cable reroute or characterised ferrite at the identified path.
10. Restore the baseline. Remove the treatment or restore the original state. The symptom must return before the change earns credit.
11. Repeat in the installed system. Bench success can disappear after the real antenna, enclosure, cable lengths and nearby transmitters return.

If moving outdoors changes the result, log what changed: location, mains connection, cable layout, nearby emitters, antenna surroundings and receiver field strength. “The room was noisy” is an observation; a source–path–victim chain is a diagnosis.

Primary Technical References

• Rohde & Schwarz, EMI Debugging with Oscilloscopes—source, coupling and emitting-element analysis with near-field and current probes.
• Rohde & Schwarz, EMI Debugging at Board Level—repeatable E- and H-field localisation and the distinction between bench debugging and chamber measurements.
• Keysight, The Journey of a Signal—input attenuation, preamplification, filtering, mixer behaviour, sensitivity and distortion in signal analysers.
• Fair-Rite, technical papers on ferrite EMI suppression and impedance measurement—frequency-dependent common-mode impedance, material and fixture limitations.
• Recommendation ITU-R SM.575-3—receiver desensitisation, blocking and false responses in strong electromagnetic fields.
• IEC 61000-4-6:2023—repeatable conducted RF-immunity test methods and coupling through connected conductors.
• ETSI EN 301 489-1 V2.2.3—EMC test applicability for radio equipment enclosure, power and wired-network ports.
• Directive 2014/53/EU, Radio Equipment Directive—conformity responsibilities for the radio-equipment configuration placed on the market.

The Useful Result Is a Reproduced Mechanism

I do not call a hotspot quiet because it is in a metal box, powered from a battery or wearing a row of clip-on ferrites. I call a change useful when the disturbance falls at the receiver, the relevant path measurement changes with it, the original condition restores the symptom, and the improvement survives the installed configuration.

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