Cable Shield Termination And Grounding

A cable that enters an enclosure is a conductor with a shield around it, and the shield only works if it is terminated correctly. A shield that is connected at one end and left open at the other is an antenna, a shield that is connected through a long pigtail is an inductor, and a shield that is connected to the enclosure at both ends carries a current that would not otherwise exist. The termination is therefore an electrical design decision rather than a mechanical detail.

This article explains how a shield works, how the termination changes it, how the connection is made to a board, and how the result is verified.

How A Cable Shield Works

A shield intercepts the electric field that would otherwise reach the conductors inside it, and it provides a return path for the current that is induced on the cable. The effectiveness depends on the completeness of the coverage and on the impedance of the path from the shield to the reference. A foil shield with a drain wire gives good coverage at high frequency but the drain wire adds inductance; a braided shield is less complete but has a lower impedance; a combination of the two is used where both properties are needed.

The current on a shield is the reason that a single ended connection often fails. An external field induces a current on the shield, and if the shield is grounded at only one end, that current has nowhere to return and the voltage it develops across the shield appears as a potential difference between the cable and the equipment. Grounding the shield at both ends gives the current a return path and removes that voltage, at the cost of a ground loop current that flows when the two ends are at different potentials.

The shield also has to be considered where the cable passes through the enclosure wall. A penetration that is not bonded to the wall allows the shield current to continue on the inside of the enclosure, where it can couple into the board, and the usual remedy is a connector that bonds the shield to the chassis at the point of entry. Where the connector is mounted on the board instead, the board ground has to be connected to the chassis at the same point, otherwise the shield current flows through the board on its way to the chassis and radiates inside the enclosure on the way.

Termination And Its Effect

The quality of the termination is measured by the impedance between the shield and the reference at the frequency of interest, and a long pigtail is the worst case. A pigtail of a few centimetres has an inductance of tens of nanohenries, which at a hundred megahertz is an impedance of several ohms, and the shield stops behaving as a shield at that point. The practical rule is that the shield should be bonded to the enclosure or to the board ground over as much of its circumference as possible, and that the length of the connection should be a small fraction of the wavelength of the highest frequency of concern.

A connector that provides a full circumferential bond is the cleanest solution, and it is the reason that metal shelled connectors with a conductive gasket or a spring finger ring are used on cables that carry high frequency signals or that leave an enclosure. A plastic connector with a drain wire inside is convenient and inexpensive, and it is adequate for a low frequency signal in a benign environment.

Connecting The Shield To A Board

Where the cable terminates on a board rather than on a chassis, the shield has to be brought to the board ground through a connection with a low impedance. The usual arrangement is a pad or a row of pads on the board that the connector shell contacts, tied to the ground plane with a group of vias placed immediately beside the pad. A single via to a ground plane several millimetres away adds the inductance of that path, which defeats the purpose of the shield at high frequency.

The return path also has to be continuous on the board. A shield that is bonded to a ground plane which is then split, or which has a long narrow neck between the connector and the rest of the circuit, has an impedance that appears in series with the shield. The routing of the return path is therefore part of the connection, and it is worth keeping the ground under the connector as a solid area rather than as a set of traces.

Ground Loops And Their Management

When a shield is bonded at both ends and the two ends are at different potentials, a current flows in the shield. That current is a problem for two reasons: it can couple into the signal conductors through the transfer impedance of the cable, and it can radiate from the loop that the shield and the reference form. The cures are either to break the loop or to make it small. Breaking it usually means an isolator, a transformer, or a differential receiver that tolerates the common mode voltage, and making it small means bonding the shield to the local reference at each end so that the loop area is minimised.

The choice depends on the signal. A high frequency digital or radio signal is usually transmitted in a differential pair or a coaxial line, and the shield is bonded at both ends because the common mode current is smaller than the penalty for leaving it open. A low frequency analogue signal is more sensitive to a ground loop current than to an electric field, and a single ended shield with a differential input or an isolation barrier is the usual answer.

Verification

The connection is verified by measurement rather than by inspection. A transfer impedance or a shield effectiveness measurement on the cable assembly characterises the cable itself, and an emission and immunity measurement on the complete product shows whether the termination is doing its job in the real enclosure. On the bench, a continuity measurement from the shield to the reference at several points confirms that the bond is present, and a close look at the connector shows whether the bond is circumferential or a single wire.

The cable and the board are usually designed by different people, and the interface between them is where the problem appears. A cable drawing that specifies the shield termination, a connector that provides a defined bond, and a board footprint with a ground pad and a ring of vias around the connector together remove the ambiguity. The related board level measures are described under EMI suppression design principles, the grounding of a switching supply under radiated EMI from a switching regulator, and the outline features under board outline and mounting design.

FAQ

Should a cable shield be grounded at both ends? For a high frequency signal, usually yes, because it gives the induced current a return path and removes the potential difference. For a sensitive low frequency signal, a single ended shield with an isolation barrier is often better.

Why is a pigtail termination so poor? Because its inductance is large at the frequencies where shielding matters. A few centimetres of wire can have an impedance of several ohms at a hundred megahertz, which is enough to defeat the shield.

How is the shield bonded to a board? Through a pad or a ring of pads connected to the ground plane with vias placed immediately beside them, so that the path from the shield to the plane has the lowest possible inductance.