Shielded vs Unshielded Cable: Braid, Foil and Transfer Impedance

01A shield that is not terminated is an antenna

Get any one of the three wrong and the result is a cable that is heavier, stiffer and more expensive than the unshielded alternative, with no measurable benefit. Two of the three are decided on the datasheet and can be checked before the order is placed. The third — termination — is decided in the cabinet and at the motor junction box, and is where most field failures actually originate.

This article stays on sourced ground throughout. Every transfer impedance figure below is a limit published in a standard or stated on a manufacturer data sheet, and the installation rules are taken from drive manufacturers' engineering guidance and from the PROFINET cabling guideline, not from general practice. Where a common claim could not be traced to a primary source we say so rather than repeat it.

02Foil, braid and served wire: what each construction does

Three screening constructions dominate industrial cable, and they fail in different ways.

A foil screen is a thin aluminium layer bonded to a polymer film, wrapped around the core or around individual pairs. Its defining property is complete coverage: there are no apertures for a field to leak through, which makes it effective against electrostatic and high-frequency interference, and it is why individual pairs inside a multi-pair data cable are foil-wrapped to stop crosstalk between pairs. It is light, thin and inexpensive. Its weaknesses follow from the same construction: the metal layer is thin, so its resistance is comparatively high, and it is mechanically fragile. Under repeated bending the foil work-hardens and cracks. A foil screen is a sound choice for a fixed installation and a poor one for a continuously moving axis.

A braid screen is woven tinned-copper wire. Because it is a weave, it cannot reach complete coverage — a well-made industrial braid is typically specified in the region of 85 % optical coverage, and igus states a maximum of 90 % for its chainflex range. What the braid gives up in coverage it gains in conductance and in mechanical robustness: far lower resistance than foil, and the ability to survive handling, crushing and flexing. Its limitation is frequency. Once the apertures between the woven strands stop being small compared with the wavelength, the weave leaks, and the braid's advantage over foil narrows as frequency rises. For the classic industrial problem — a motor feeder, a drive output cable, or a control circuit running beside power — braid is the usual default and it is usually the right one.

A served or spirally wrapped screen is wire wound helically around the core in one direction rather than woven. Its advantage is that it tolerates twisting, because the winding can open and close as the cable is rotated — which is why cable makers specify a wrapped screen, not a braid, for torsional applications such as robot joints. Its disadvantage is that it offers less coverage and higher resistance than a braid. A point worth correcting, because it circulates widely: a spiral screen is not the standard answer for a drag chain. A linear back-and-forth travel application is served by a braid with a deliberately flat braid angle, and igus states explicitly that the same screen should not be used for linear travel and for torsion. Cables sold for energy chains are commonly braid-screened for this reason.

03Transfer impedance: the number behind the word “shielded”

Coverage tells you how much of the core is covered. Transfer impedance tells you how well the screen performs, and it is the parameter that separates a specified screen from a claimed one. Transfer impedance, usually written Z_T, is the voltage induced on the conductors inside the screen per unit of current flowing on the screen itself. Lower is better.

The value of the parameter is that cable standards publish limits for it by grade. IEC 61156-6, the sectional specification for symmetrical pair and quad cables used in work area wiring, grades screened cable by transfer impedance and by coupling attenuation, and the grades differ by construction rather than by marketing language. The published limits are these.

Three conclusions follow from that table, and each one is counter-intuitive enough to be worth stating plainly.

• A screen does not become better behaved at higher frequency; it becomes worse. Both grades climb monotonically with frequency. From roughly 1 MHz upward a braid's transfer impedance rises at approximately 20 dB per decade, because coupling becomes dominated by the inductance of the weave apertures once those apertures are no longer small compared with the wavelength.
• The Grade 1 to Grade 2 gap is not a constant. It is about 5× at 1 MHz and about 10× at 100 MHz, and it is not a straight line in between. A specification that says “Grade 2 or better” therefore says different things at different frequencies.
• The test method has a usable frequency range, and it is shorter than most people assume. The triaxial method measures only the magnetic component of the transfer impedance; the electrostatic, capacitive component is covered separately, in IEC 62153-4-8. It is generally suitable up to about 30 MHz on a 1 m sample and up to about 100 MHz on a 0.3 m sample, which corresponds to an electrical length below roughly one sixth of the wavelength. A figure quoted at 100 MHz on a 1 m specimen is therefore not the same measurement as a figure quoted at 100 MHz on a 0.3 m one.

One further point is worth carrying into a purchase discussion. IEC 61156-6 notes that a screen DC resistance of 30 mΩ/m or less can be used as an indicator that the Grade 2 transfer impedance requirement is met — but it cannot substitute for the measurement. The distinction matters, because DC resistance is trivially measurable on a bench and transfer impedance is not. A supplier who offers only the resistance figure has given you an indicator, not a qualification.

04Coupling attenuation, and why a grade never travels alone

Transfer impedance is the screen's own performance. It is not the only parameter that decides whether a cable survives an EMC environment, because a screen cannot compensate for a poorly balanced pair. That is what coupling attenuation captures: the combined effect of the screen and the balance of the conductor pair, measured with an absorbing clamp under IEC 62153-4-5, or by the triaxial method. IEC 61156-6 grades it by type, using a value quoted over the 30–100 MHz band with a defined roll-off above it.

The coupling attenuation types, in order of screening performance

Type I is specified as 85 dB, Type Ib as 70 dB, Type II as 55 dB and Type III as 40 dB, each over 30–100 MHz and each following the same frequency dependence above the anchor band. For screened cables, Type II is the minimum requirement; Type III is associated with unscreened constructions, where external crosstalk control is the concern rather than shielding. In construction terms the same standard notes that Type I is met by a braid plus foil design, Type Ib by a braided cable with a lighter braid where maximum screening is not needed, and Type II by a foil screen of specified characteristics. A cable datasheet that quotes a transfer impedance grade and no coupling attenuation type has told you about half of the EMC story.

Converting transfer impedance into shielding attenuation

System budgets are usually written in decibels, and a screening attenuation can be derived from the transfer impedance for an electrically short assembly as a_s = 20 · log (150 Ω / Z_TE), where the 150 Ω is the standardised impedance of the measuring arrangement defined in IEC 62153-4-15. The exercise is worth doing once, because it makes the cost of a poor termination obvious: the screen is only as good as the weakest element in the current path, and the weakest element is very often not the cable.

05Terminating the screen: one end, both ends and the 360° rule

The screen has to carry the interfering current away, and where it is allowed to go depends on the circuit. Two rules cover most industrial cases, and they are opposite to each other for a reason.

Instrumentation and analogue signal circuits are bonded at one end, normally the control-room or panel end. The reason is not shielding theory but loop physics: any difference in earth potential between the field device and the panel drives a current along the screen, that current magnetically couples into the signal pair, and the result appears at the input as differential-mode noise that a PLC's common-mode rejection cannot remove. Bonding at one end breaks the loop. Where lightning protection is required, the far end may be terminated through a capacitor or a surge device rather than left fully open, so that the screen stays bonded at the panel and the protection is provided by a component rather than by earthing the far end.

Drive output cables are bonded at both ends, around the full circumference. This is not a preference. Drive manufacturers' engineering guidance states that screened cable is mandatory between drive and motor for compliance with emission standards such as IEC 61800-3, and that the screen must be connected at both ends — at the drive and at the motor — with 360° termination. The purpose is to give the common-mode current generated by the inverter's fast switching edges a controlled, low-impedance return path, instead of letting it discharge through motor bearings, pipework and building structure. The same guidance is specific about cable construction: a symmetrical multicore cable with low leakage capacitance, and where the cross-section exceeds roughly 10 mm², a cable with three equipotential conductors rather than one. Reinforced or screened single-core cables should not be used for this duty.

Why a pigtail defeats the screen, stated as an arithmetic

This is why the connection is made around the circumference rather than at a point. The PROFINET cabling and interconnection guideline requires 360° shielding on its connectors as a specification item — M12, M8 and RJ45 variants alike — with the stated purpose of screening against EMI from rotating machines. The same geometry is achieved at a gland or a screen clamp: the braid is brought into full peripheral contact with a conductive body that is itself bonded to the enclosure.

Two practical notes apply across all four rows. First, the screen is not a protective conductor. It carries EMC current and may carry fault current, but the protective earth function must be provided by a dedicated conductor with the cross-section the installation standard requires — a screen must never be the only earth path. How the earthing system itself is arranged and sized is a separate design exercise, worked through for hybrid generation sites in earthing and bonding to IEC 60364 and NEC 690. Second, the value of the construction is decided by a measurement at commissioning, not by inspection: a screen-to-earth resistance reading that is unexpectedly high at either end indicates a painted bracket, a folded-back drain wire, or a clamp that is not making contact around the circumference.

06Motion changes the answer: linear travel and torsion want different screens

In a fixed panel the screen is chosen for the interference. On a moving axis, survival comes first, and the screen that survives depends on how the cable moves — which is not one thing.

The distinction is made explicitly by igus in its technical guidance on screens in moving applications. For linear travel in an energy chain, the correct screen is a braid with a deliberately flat braid angle, because a flat weave opens only minimally over the bend radius and keeps the screening window small throughout the stroke. The failure mode to avoid is a braid wound at too acute an angle, which expands over the radius and breaks apart; the broken strand ends then migrate inside the cable and can cause conductor-to-conductor faults. For torsional duty, such as a robot joint, the guidance is the opposite: use a wrapped screen, where the wires are laid in one direction rather than woven, so the screen can open and close as the cable rotates.

This is the point at which the widely repeated claim that drag chains “always use a spiral shield” falls apart. Both statements cannot be true, and the primary source is unambiguous: linear travel and torsion take different screens, and the linear case takes a braid. A cable sold for energy chain duty is normally specified as braid-screened, and the data sheets confirm it — a braid-screened industrial Ethernet cable intended for energy chains states a continuous-flex minimum bend radius of 15 × the outer diameter and a flexing temperature range of −20 °C to +50 °C, against −40 °C to +80 °C for fixed installation. The temperature band is part of the specification for the same reason the screen is: a moving cable is a different product from a static one.

07What an unshielded cable is actually for

A screen earns its place when the circuit is a source of interference that has to be contained, when it is a victim of interference that has to be excluded, or when both are true at once. Drive output cables are the clearest case of the first. Analogue instrumentation running near heavy power is the clearest case of the second. A control circuit inside a cabinet fed from a filtered supply, with its own separation from power wiring, is frequently neither: at that point an unshielded multicore cable with the right conductor geometry does the job, and the money and the installation hours are better spent on separation and on pair twisting than on screen coverage nobody will terminate properly.

Two caveats keep that from being a licence to drop the screen everywhere. First, separation is the alternative, not an additional option — if the screen is removed, the routing has to be reviewed against the installation standard's requirements for physical separation between power and signal circuits, with crossings at right angles where they cannot be avoided. Second, a screen that is specified but terminated badly performs worse than no screen at all in one respect: it adds a path for circulating currents into a system that was not designed to carry them. Where a screen is fitted, the termination rules in section 05 are not optional extras.

For plant wiring where a screen is required, a braided screen over a flexible control-cable construction covers the general case — CY industrial flexible PVC shielded control cable is built this way, with a tinned copper braid over the laid-up core bundle and roughly 85 % optical coverage. Where the circuit is a data channel rather than a control circuit, the construction question becomes a category question, and the balance parameters in section 04 become as important as the screen: Cat6A and Cat8 S/FTP shielded data cable is specified against IEC 61156-5 and -6; the transfer impedance grade limits in Table 2 come from Part 6 of that series.

08What belongs on a shielded-cable datasheet

The word “shielded” on its own carries almost no information. Five items turn it into a specification, and a supplier who cannot produce them has not specified the screen.

It is worth being explicit about where we stand on the last row, because it is the one where claims are least often substantiated. Optical coverage and transfer impedance are measurements; a cycle rating is a test result that depends entirely on the test conditions — travel length, bend radius, speed and acceleration all change the number that comes out. When we state a figure for a flexible construction we state the condition it was measured under, and where a cycle rating has not been run on a specific construction we say that rather than quoting a general figure.

09Six questions for the quotation

• What is the screen, precisely? Foil, braid, served, or a named combination. If the answer is “shielded”, the next four questions are the ones that matter.
• What is the optical coverage of the braid? A percentage, from the construction drawing rather than from a general range.
• Is there a transfer impedance grade or a measured value? For a data cable, a grade against IEC 61156-6; for a control cable, a measured value at a stated frequency, or an honest statement that no such figure is available. A DC resistance figure is an indicator for the Grade 2 requirement, not a substitute for the measurement.
• Where does the screen get bonded, and can the gland make peripheral contact? The construction and the termination hardware are one decision. A braid that will be landed under a terminal screw has been specified as a screen and installed as an antenna. Cable glands designed for screen contact, and the accessories that go with them, are part of the same specification — see EMC cable glands and shield termination for VFD and servo cable for the termination side in detail.
• Does the cable move, and how? Linear travel and torsion take different screens, and a cable rated for fixed installation is not a substitute for either.
• Has the screen been continuity-checked on the finished cable? The screen is a conductive path, and the check belongs in the release test rather than in an assumption. Batch and metre markings belong on the same question, because a screen fault found a year later has to be traceable to the drum it came from.

The screen is a small part of a cable's cost and a large part of its behaviour. Specifying it as a construction, with the numbers that describe it and the termination method that will be used, costs a handful of questions at the enquiry stage and prevents the failure mode that no one can diagnose afterwards — a plant full of cables that are nominally shielded, none of which is doing anything.