Cable Shielding and Grounding Selection for Cable Assemblies

Begin with the four-layer reference map

Many failed cable specifications use one word—ground—for several electrically different nodes. Replace the word with a map.

1. Signal conductors

These carry the intended differential, single-ended, analog, digital, sensor, control or power-related signal. Record pair assignment, source/receiver topology, impedance and balance requirements where applicable. A differential pair is not automatically immune to interference; imbalance can convert common-mode energy into differential error.

2. Signal reference or return

This is the intentional circuit return or common used by the interface. It may be carried in a conductor, established locally at each isolated endpoint, coupled through a network, or defined by an interface standard. It should not be assumed identical to chassis.

3. Chassis or enclosure bond

Chassis provides a conductive equipment boundary and may receive a cable shield at the connector entry. A low-impedance, mechanically continuous connection at that boundary can be important at high frequency. Whether chassis is coupled to circuit common, and how, is an equipment-design question—not something the cable assembler should invent.

4. Protective earth

Protective earth is a safety function governed by the equipment and installation. It is not an optional noise-control wire. A cable drawing must not tell an operator to remove a protective bond because hum or communication errors occur. A qualified safety engineer must resolve any interaction among chassis, protective earth and cable shields.

For each endpoint, draw where these four layers connect, remain isolated or couple through intentional components. Include connector shells, backshells, panel bonds, drain wires and pigtails. The drawing often reveals that “shield at both ends” actually means a drain wire to circuit common at one end and a shell-to-chassis bond at the other—two very different structures hidden by one phrase.

Identify how interference reaches the circuit

Shield selection should answer a coupling problem. Typical mechanisms include:

• capacitive or electric-field coupling, in which changing voltage couples through stray capacitance;
• inductive or magnetic-field coupling, in which changing current and loop area matter;
• radiated electromagnetic coupling, which becomes important as structures behave more like antennas;
• common-impedance coupling, in which wanted and unwanted currents share a non-zero impedance;
• common-mode conversion, where imbalance at a cable, connector or termination turns common-mode energy into differential error;
• cable-to-cable coupling, affected by route, separation, parallel run length and frequency;
• emissions from the intended signal, where the cable itself becomes part of the radiating structure.

Texas Instruments’ EMI/RFI board-design application report describes cables as possible radiated-emission antennas, susceptibility antennas and cable-to-cable coupling paths. Its Ethernet guidance also emphasizes connector/chassis transitions and common-mode balance. These are engineering mechanisms, not universal cable recipes. A recommendation written for a particular Ethernet PHY or connector should not be copied into an unrelated low-level sensor assembly.

Create a threat worksheet:

If the coupling mechanism is unknown, preserve design options for prototypes rather than freezing a slogan into the drawing.

Match shield construction to the real duty

Foil shields

A foil shield can provide high coverage with low bulk. It is commonly paired with a drain wire because thin foil is not normally treated like a robust round conductor for termination. Its performance depends on foil material, overlap, drain contact, orientation, flex life and how the termination transitions into the connector. Repeated flexing can change a construction that performed well in a static sample.

Do not write only “100% foil coverage.” Ask how coverage is defined, how seams/overlap and bends behave, and how electrical continuity reaches the required termination. The drain wire is part of that answer, not a generic extra ground conductor.

Braid shields

A braid can offer robust circumferential contact and flex performance, but its coverage and impedance depend on strand material, carrier construction, braid angle, optical coverage, termination and frequency. More braid can add diameter, mass, stiffness and termination complexity. A drawing should state the controlled construction and acceptable termination, not rely on a photograph.

Foil plus braid

A combination can address complementary coverage, impedance and mechanical needs. It can also create more layers to terminate incorrectly. Specify whether layers are electrically common, isolated, cut back, folded, bonded to a ferrule or terminated separately—and why. “Double shield” is not a complete engineering definition.

Individually shielded pairs and overall shield

Individual pair shields can reduce internal coupling; an overall shield addresses interaction with the environment. The choice affects diameter, capacitance, flexibility, breakout design and termination. State whether each pair shield, drain and overall shield remains distinct at the connector. Accidental joining at a backshell can defeat the intended architecture.

Twist, balance and routing remain important

A shield does not replace pair balance, controlled geometry or sensible routing. Twisting reduces loop area and helps external fields couple more equally into a pair; receiver common-mode rejection then matters. Separation from noisy conductors and avoiding long parallel routes can reduce coupling before shield current becomes a problem. Validate the combined design.

The termination is part of the shield

At high frequency, a long drain-wire or pigtail connection adds inductance and can leave the enclosure entry electrically open even though a continuity meter reads near zero. A circumferential backshell or clamp may provide a lower-impedance high-frequency transition, but it must be compatible with the connector, enclosure and system architecture. At lower frequencies or in circuits with endpoint potential differences, a DC bond at both ends may allow unwanted shield current. These facts explain why termination cannot be selected from continuity alone.

Record these geometry details on the drawing:

• shield cutback length and permitted variation;
• drain-wire size, insulation and maximum exposed/pigtail length where used;
• ferrule, clamp or crimp method and controlled tooling;
• 360-degree termination requirement, if the system design calls for it;
• connector shell and backshell material/finish and electrical continuity path;
• shell-to-panel contact and coatings/gaskets that could interrupt it;
• unused shield/drain treatment and insulation from neighboring contacts;
• strain relief so mechanical load does not pull on the shield termination;
• corrosion/environmental compatibility of the material stack.

A cable assembler can reproduce a defined termination. The assembler should not decide whether a spare drain goes to signal common, chassis or nowhere when the schematic is ambiguous.

Why one-end versus both-end rules conflict

One-end shield termination is often used to avoid low-frequency circulating current when endpoints sit at different potentials. Both-end termination can provide a more complete high-frequency boundary when the chassis and bonding architecture are designed for it. A hybrid can block DC while coupling higher-frequency energy. Each has failure modes.

The correct reasoning asks:

1. What frequency range must the shield control?
2. Is the shield intended to intercept electric fields, contain common-mode current, or do both?
3. What voltage may exist between endpoint chassis/earth systems?
4. Are signal circuits isolated or referenced between endpoints?
5. What does the interface/product standard prescribe?
6. Does the connector support the intended termination with low, repeatable impedance?
7. What happens during ESD, surge or a fault condition?
8. How will the architecture be verified in the real equipment?

Analog Devices’ application material offers one-end and both-end examples for specific balanced/instrumentation or RS-485 contexts. Those examples are useful because they expose ground-potential and common-mode constraints; they must not be converted into a rule for all signals. A high-speed digital link crossing two chassis can require a different treatment from a microvolt sensor connected to an isolated amplifier.

If a product standard or safety analysis mandates a bond, follow it. Do not insert a capacitor, resistor, fuse or other component into a shield/safety path without system-level electrical, safety, transient and EMC review. A drawing copied from an application note is not a project approval.

Treat the connector and enclosure entry as one boundary

The shield should not arrive at a metal connector shell and then lose continuity through a painted panel, insulating washer or plastic backshell unless the design explicitly intends that break. Conversely, a metal shell bonded to chassis can defeat an intended isolated shield if the cable drawing ignores the equipment interface.

Request an endpoint boundary sketch for both ends:

cable shield → termination hardware → connector shell/selected pin → equipment receptacle → panel/chassis/reference network

List mating connector part numbers, panel finish, gasket/contact features and mounting method. Verify continuity/impedance using the approved method after final assembly, not only on an open bench backshell.

For hybrid assemblies with multiple shield zones, label each zone and its boundary. A shield over a noisy power section may need different routing from a low-level signal-pair shield. Breakout length and exposed unshielded conductors should be controlled because the last centimeters can dominate susceptibility or emissions.

Translate the architecture into a buildable specification

Once a qualified engineer approves the system concept, procurement needs reproducible fields:

• cable core: conductor count/size/material, pair assignment, impedance or balance requirement;
• shield layers: material, coverage/construction definition and drain association;
• layer separation: which shields are common or isolated;
• end A and end B: connector, pinout, shell and shield termination in explicit words;
• chassis/reference map: informative system diagram controlled by the design authority;
• cutback, pigtail, ferrule/clamp/crimp and strain-relief dimensions;
• environmental/flex/diameter constraints and minimum bend rule from the cable design;
• workmanship and inspection standard only when contractually applicable;
• electrical inspection and continuity/isolation test points;
• EMC or functional validation plan at assembly and system level;
• labeling, revision, approved materials/substitutions and change notification.

IPC/WHMA-A-620E is an industry workmanship/acceptability standard for cable and harness assemblies. NASA-STD-8739.4 is an active NASA workmanship standard for interconnecting cables/harnesses. Either can be relevant when a contract invokes it. Neither chooses the correct one-end/both-end grounding topology for every product, and neither should be claimed without reviewing the purchased standard, class/addenda and contract.

Three hypothetical decision cases

The following cases are hypothetical. They are not CoreCavo designs, tests, customers or performance claims.

Case A: low-level sensor beside a variable-frequency drive

A remote sensor carries a small balanced analog signal to an acquisition unit. Both equipment grounds and the isolation scheme are documented. The route runs near a motor drive cable. The team first reduces coupling with pair twist, separation and route control, then compares approved shield termination options using measured noise and the real endpoint potential conditions. It does not automatically bond both ends or disconnect protective earth. The released drawing records which end owns the shield connection and how the drain is insulated at the other end.

Case B: high-speed interface through two metal enclosures

A high-edge-rate differential interface exits one chassis and enters another. The EMC engineer treats connector shells, panel bonds and the overall shield as a continuous high-frequency boundary. The prototype preserves a circumferential termination option and minimizes unshielded breakout. Both-end chassis connection is evaluated together with common-mode current, equipment bonding and immunity/emissions evidence. A DC continuity result alone is not the release criterion.

Case C: moving harness with foil-only prototype

A static prototype passes an initial bench check, but the assembly will flex repeatedly in service. The team recognizes that foil/drain continuity, stiffness and strain at the termination may change. It evaluates a braid or combined construction against flex, diameter, mass and EMC needs, then repeats electrical and system validation after the approved mechanical conditioning. It does not market the new construction as “double shielded performance” without the applicable evidence.

Validate at three levels

Level 1: assembly identity and workmanship

Confirm materials, cable marking, shield layers, drain routing, cutback, termination hardware, shell continuity, pinout, strain relief and traceability against the controlled drawing. Use the invoked workmanship standard and class exactly. A cosmetic inspection cannot prove high-frequency termination performance.

Level 2: electrical assembly checks

Perform the approved continuity, isolation, resistance, bond or transfer-impedance-related measurements that the design requires. The method, test points, fixture, frequency, limit and uncertainty must be specified. A low-frequency ohmmeter result cannot be relabeled as broadband shielding effectiveness.

Any test involving hazardous voltage, protective bonding, energized enclosures, surge, high current or access to exposed conductors must be carried out by qualified personnel with suitably rated equipment, risk controls and the applicable device/standard limits.

Level 3: system EMC and functional evidence

Test the complete equipment and cable in the relevant configuration: endpoint chassis, power supplies, cable route/length, loads, operating modes and nearby noise sources. Use the applicable emissions/immunity standard and accredited laboratory where required. Monitor both formal EMC criteria and the real function—data errors, sensor offset, resets, control loss or thermal behavior.

A passing system test applies only to the documented configuration and change-control envelope. It does not automatically certify a standalone cable, every length or every enclosure.

Diagnose by current path, not by adding layers

Freeze evidence and retest the right changes

The release package should contain the reference map, threat worksheet, approved schematic/drawing, cable/connector specifications, prototype identities, workmanship record, electrical data, EMC/system configuration, raw reports, deviations and signed disposition. Keep photographs that show the termination before it is hidden by overmold where the process permits and the record is controlled.

Review or retest when a change can alter shield impedance, balance, coupling, ground current, safety or mechanical life. Examples include foil/braid material or coverage, drain size/location, pair lay, cable length, connector shell/plating, backshell/clamp/ferrule, cutback/pigtail, overmold, panel finish, endpoint chassis network, isolation, power supply, cable route/bundle, flex duty or applicable standard revision.

The custom cable RFQ guide covers the broader set of electrical, mechanical, environmental and commercial input fields. The cable test and compliance documents guide explains how to bind evidence to the correct product and revision. The custom cable assembly page is the commercial destination; it should not inherit unverified shielding performance from this resource.

For an engineering review, send the signal/reference/chassis/protective-earth map, interface and endpoint drawings, threat/frequency information, route and flex duty, proposed shield/termination, applicable standards, validation plan, quantity and change-control needs through the CoreCavo RFQ page. A submission begins feasibility review; it does not confirm design safety, EMC compliance, certification, production capability, price or schedule until the exact proposal and evidence are approved.