Rail EMC case study: intermittent comms dropout

A communication link that works on the bench but drops out intermittently on a train is rarely suffering from a random fault. More often, the disturbance path has not yet been identified.

This rail EMC case study presents an illustrative investigation based on problems commonly encountered in rolling stock and trackside equipment. It is not a report of a named customer project and contains no claimed test result. The focus is the engineering method: reproduce the comms dropout, identify the coupling mechanism, apply a controlled mitigation and retain enough evidence to support the wider compliance assessment.

Typical scenario

Consider a distributed rail control system containing a central controller, remote input and output modules and a long communication harness routed through a vehicle. The link remains stable during workshop testing. Once installed, it occasionally loses communication when traction equipment changes state, contactors operate or another high-current subsystem is energised.

The first temptation is to blame software timing. Watchdogs are extended, retries are increased and diagnostic logging is added. These changes may make the symptom less visible without reducing the coupled disturbance. A receiver that is repeatedly driven outside its common-mode range can still corrupt messages, even if the protocol successfully retries most of them.

The applicable requirements depend on where the equipment is installed and its intended function. Within the EN 50121 family, rolling-stock apparatus, signalling equipment and fixed power installations do not automatically follow the same product-specific requirements. EN 50155 may also be relevant to electronic equipment used on rolling stock, while contractual rail specifications can add further tests or operating conditions. Engineers should check the latest active editions, product scope, test levels, port definitions, performance criteria, cable configuration and customer test plan rather than treating EN 50121 as one universal test method.

For the illustrative system, the likely investigation would combine functional monitoring with conducted and radiated immunity techniques selected from the applicable product standard. A product standard may call up basic methods such as IEC 61000-4-6 for conducted RF immunity or IEC 61000-4-3 for radiated RF immunity. Transient immunity methods may also be relevant where switching activity is implicated. These methods are not interchangeable, and none should be selected merely because the equipment is available.

A CDN may be suitable for conducted RF immunity on a defined cable port when the called-up method and cable arrangement permit it. Where a rail, automotive, military or aerospace procedure specifies current injection, a correctly characterised BCI probe may be required instead. Using a convenient clamp in place of the prescribed coupling device changes the injected quantity, common-mode impedance and calibration method. The resulting dropout may be repeatable, but it will not necessarily represent the required test.

EMC Hire can support this type of work through conducted immunity equipment hire, practical setup advice, pre-compliance engineering, test-facility access and on-site testing. Where formal compliance testing is appropriate, the work can be planned around the applicable product requirements without implying that testing alone completes the manufacturer’s conformity obligations.

Reproducing the failure without changing it

Intermittent faults become harder to diagnose when the test setup is altered too quickly. Moving the equipment to a bench, shortening the harness and replacing the vehicle supply with a laboratory power source can remove the very impedances that allowed the disturbance to couple.

Start by capturing the installed configuration. Record harness length, shield termination, connector backshells, chassis bonds, supply topology, communication speed, termination resistance, unused conductors and the operating states of nearby equipment. Photographs should include cable height, separation and crossing points. A statement such as “standard vehicle harness used” is not enough to reproduce the test six months later.

Functional monitoring must distinguish a physical-layer disturbance from a higher-layer recovery event. Useful observations may include:

• receiver error counters and frame-check failures;
• bus-off, link-down or loss-of-carrier indications;
• common-mode and differential voltages measured with suitable bandwidth and probing;
• controller resets, brownout flags and watchdog activity;
• the timing relationship between dropout and operation of switching loads;
• recovery behaviour after the disturbance is removed.

Probe choice matters. A long oscilloscope ground lead can add enough loop inductance to create ringing that is not present at the receiver pins. Conversely, a low-bandwidth differential probe may hide a fast common-mode event. The measurement system must have adequate voltage rating, bandwidth, common-mode rejection and isolation for the circuit and environment. Manufacturer documentation for the transceiver, connector, cable and probing equipment should be checked rather than relying only on nominal interface labels such as CAN, RS-485 or Ethernet.

Separating the likely coupling paths

Common-mode current on the communication harness

Long rail harnesses can behave as effective receiving structures. RF current coupled onto the cable may return through chassis bonds, parasitic capacitance, shields and connected equipment. If shield termination impedance is high at the disturbance frequency, common-mode voltage can appear at the transceiver despite acceptable low-frequency continuity.

A pigtail shield connection may measure close to zero ohms with a multimeter yet present significant inductive impedance at RF. Replacing it with a short, circumferential termination can materially change the current path. That change should be evaluated mechanically as well as electrically, particularly where vibration, connector design and maintainability constrain the installation.

Disturbance entering through the DC supply

A communication failure does not prove that the disturbance entered through the data cable. A supply transient can upset the transceiver, processor or isolation circuitry while leaving no obvious permanent reset record. Monitoring supply rails at the affected electronics, rather than only at the laboratory source, helps expose voltage disturbance across wiring and input filtering.

Power-port immunity testing must use the coupling network and waveform required by the applicable method. A LISN should not be substituted here. LISNs are used for conducted emissions measurements on relevant power ports, not as conducted immunity injection devices.

Radiated coupling into the enclosure or PCB

If cable-focused conducted immunity does not reproduce the symptom, radiated RF immunity may need investigation. Enclosure seams, display apertures and poorly bonded panels can allow fields to couple into internal traces. The applicable field strength is expressed in V/m, with frequency range, modulation, dwell and performance criteria taken from the relevant product standard and test plan.

Near-field probing can help locate susceptible circuitry during development, but it does not reproduce a calibrated radiated immunity test. It is a diagnostic technique. Treating a hand-held probe result as formal evidence gives false confidence because field distribution, distance and delivered level are not adequately controlled.

Diagnosis through controlled changes

Change one variable at a time. If a ferrite, shield bond, filter capacitor and software retry modification are applied together, the failure may disappear but the engineering team learns very little. That creates trouble when cost, mass or space pressure later forces one of those changes to be removed.

A disciplined sequence may compare the original configuration with a small number of reversible modifications:

• improve the shield termination while retaining the original cable route;
• change harness separation from known switching conductors;
• apply an appropriately characterised common-mode suppression component;
• improve local high-frequency bonding between enclosure sections;
• evaluate supply filtering at the affected module;
• inspect communication termination and receiver biasing against manufacturer recommendations.

Component current rating, voltage rating, insulation, temperature range and rail environmental constraints remain relevant. A ferrite that suppresses RF on a prototype bench may saturate, overheat or fail mechanical requirements in service. Mitigation cannot be judged solely by whether the dropout stops during one sweep.

Once a plausible coupling path is identified, repeat the test across representative operating modes and cable arrangements. Record both failures and non-failures. Negative results are useful only when the applied level, frequency, modulation, injection point, EUT state and monitoring method are documented.

From pre-compliance diagnosis to defensible evidence

Accessible pre-compliance testing allows the engineering team to investigate a prototype before a formal programme is booked. It can reveal susceptibility bands, operating modes and harness configurations that deserve closer attention. Calibrated engineering data also provides a better basis for design reviews than an undocumented trial with an RF generator and clamp.

Pre-compliance results do not automatically prove compliance. Formal work should follow the applicable product or product-family standard, agreed test plan and required configuration. Depending on regulatory, contractual or programme requirements, final rail testing may need to be performed by an appropriately accredited laboratory.

EMC Hire uses test equipment with calibration traceable through an appropriate ISO/IEC 17025 accredited calibration provider. Suitable, traceable calibration supports measurement accuracy, repeatability and comparison between development and formal testing. It also strengthens the evidence available to engineering teams, corporate stakeholders and regulatory reviewers. This wording relates to calibration traceability and does not mean that the physical instrument itself is accredited.

For commercial products following an applicable self-certification route, robust test records may support the technical file, EMC risk assessment, mitigation evidence and Declaration of Conformity. The manufacturer or responsible economic operator remains responsible for determining the legislation, standards, conformity assessment route and documentation required. EMC Hire’s commercial CE marking test support can form part of that engineering process where relevant.

Rail projects may also be governed by interoperability, safety and contractual obligations beyond general product legislation. The UK Government publishes railway interoperability guidance, but project-specific legal and contractual requirements should be confirmed with the responsible bodies.

When to Hire EMC Equipment

An intermittent comms dropout often demands concentrated testing for a short period, followed by little or no need for the same RF equipment. Purchasing a generator, power amplifier, directional coupler, coupling device, monitoring instrumentation and calibration accessories for one investigation can tie up capital without ensuring that the equipment suits the next programme.

Hiring is technically attractive when the required frequency coverage, power or coupling method is defined by a particular test window. It allows teams to scale capability during prototype integration, vehicle commissioning or a formal test campaign without accepting long-term storage, servicing and calibration overheads.

Correct system matching still matters. The amplifier must provide adequate linear output across the required range, and the coupling device must suit the cable type and method. Monitoring and protection arrangements must tolerate reflected power and changing load conditions. EMC Hire’s conducted immunity system information provides a useful starting point, but the final configuration should be checked against the current test plan.

Rental also reduces the risk of buying a system optimised for one rail platform that is unsuitable for a later project. Equipment can be selected for a defined investigation, supported with practical setup guidance and returned when the engineering work is complete.

Common EMC Testing Mistakes to Avoid

Testing a shortened harness

Reducing cable length may change resonances, common-mode impedance and coupling area. A stable short cable does not demonstrate that the installed harness will be stable. Where the prescribed configuration differs from the vehicle installation, both compliance and diagnostic objectives need to be understood.

Moving cables between test runs

Small routing changes can alter injected current and radiated coupling. If cable position is not controlled and photographed, an apparent mitigation may simply be a geometry change. Supports should be non-conductive where required and consistent with the relevant method.

Using the wrong coupling device

A CDN, BCI probe and capacitive coupling clamp serve different immunity methods. Substituting one for another invalidates the assumed injected quantity and calibration. Selection must follow the called-up method, cable type and product standard.

Monitoring only application-level communication

A pass indicator that remains green can hide repeated physical-layer errors corrected by retries. This may satisfy one chosen performance criterion but still expose inadequate margin or unacceptable latency. Error counters, link state and recovery time should be monitored where technically available.

Ignoring bonding impedance

DC continuity does not characterise an RF bond. Long straps, painted surfaces and narrow pigtails add impedance that redirects disturbance current through electronics. Bond geometry and surface preparation should be documented, not merely described as grounded.

Failing to record test parameters

Frequency range, level, modulation, dwell, coupling position, calibration data, EUT mode and observed behaviour are needed to reproduce a result. Without them, the evidence trail cannot reliably support diagnosis, formal comparison or a technical file.

Frequently Asked Questions (FAQs)

Should the first investigation use conducted or radiated immunity?

Base the choice on the suspected coupling path, installation evidence and applicable product requirements. Correlation with cable-borne switching events often justifies conducted investigation first, while sensitivity to enclosure orientation or nearby transmitters may point towards radiated testing. Both paths may contribute.

Can a BCI probe be used instead of a CDN?

Not automatically. A BCI probe injects current around a cable, while a CDN establishes a defined coupling and common-mode impedance for suitable ports under methods such as IEC 61000-4-6. Use the device specified by the applicable method and test plan.

Does passing IEC 61000-4-6 demonstrate rail compliance?

No. IEC 61000-4-6 is a basic conducted RF immunity method. The applicable rail product standard or contractual specification determines whether it is required, along with levels, frequency range, port applicability, configuration and performance criteria.

Can generic immunity testing be useful during development?

Yes, when its purpose is clearly defined. A service such as generic immunity pre-compliance testing can help identify weaknesses, but a generic standard must not replace a rail product standard where the latter applies.

How should a dropout be classified during immunity testing?

Classification depends on the performance criteria in the applicable product standard and agreed test plan. Record the duration, error mechanism, automatic recovery, operator intervention and any loss of stored state. Calling every recovered packet loss a pass can conceal behaviour that the project considers unacceptable.

What evidence should be retained after mitigation?

Keep the original and modified configurations, photographs, schematics, harness details, calibration references, test parameters, software versions, functional logs and engineering rationale. This supports later design changes and provides a defensible link between the observed problem and the chosen mitigation.

Planning the next test window

Intermittent rail communication faults are solved fastest when the harness, bonding, power system and protocol monitoring are treated as one coupled system. Reproducing the symptom under controlled conditions is more valuable than applying several unverified fixes and hoping the fault has disappeared.

To discuss an equipment hire quotation, on-site investigation, pre-compliance support, formal compliance testing where appropriate, or space at the EMC Hire test facility, contact the engineering team on +44 (0)1462 817111 or email sales@emchire.co.uk. Providing the applicable standard, expected frequency range, harness details, interface type and target test window will help the team assess the required setup.