Unscreened cables

Many cable-coupling problems are better addressed without screening the cable, which is after all an expensive and inconvenient option from the installer's point of view.

Instead, the equipment designer implements filtering and other forms of interface protection so that, whatever interference is present on the signal conductors at the cable terminal, the equipment can deal with it. In this respect common mode protection is particularly important.

Two particular types of unscreened cabling have relevance to EMC solutions: these are twisted pair (often referred to as Unshielded Twisted Pair, UTP) and ribbon cable.

Twisted pair

The purpose of twisting

At various places in this guide we have emphasized the importance of ensuring that a signal and its return are closely coupled in an adjacent conductor pair. The closer the coupling, the less interference is induced within the differential circuit. Twisting the pair together improves the situation in two ways:

++ the magnetic field coupling with the pair is continually changing orientation. If the cable is viewed as a succession of half-twists, the coupling with each successive half-twist will cancel that from the previous one;

++ electric field coupling is more balanced to each half of the pair than if the pair were untwisted.

The number of twists per unit length may be significant if the cable is passing through an area of rapidly varying field strength, since the reduction in coupling by twisting depends on the area of each half-twist being small by comparison to the field variation.

In an unknown disturbing environment, a high rate of twists per unit length is desirable. The overall performance of a twisted pair cable relates to its "balance", which in turn is determined by how tightly controlled the cable geometry is - uniformity of twist rate, cross- section and dielectric material. This implies that the handling during installation of such cables can noticeably affect their performance, particularly in respect of bending and crushing, which changes the geometry at particular points.

Coupling to and from twisted pair cables

--- Coupling model for twisted pair

The crosstalk coupling to a twisted pair has both capacitive and inductive components.

The equivalent circuit depicts the capacitive coupling as a current source I C onto each conductor half-twist while the inductive coupling is a voltage source V I in series with each conductor half-twist. Each elemental V I pair cancels the next one, leaving only the induction at the ends of the circuit un-cancelled.

The effectiveness of twisting a signal/return pair is not achieved in isolation from the signal circuit carried by the pair. It depends on the impedance and the balance or unbalance of this circuit. For unbalanced circuits, capacitive coupling dominates at high impedances and there is little reduction in overall coupling by twisting. As the circuit impedance drops so differential capacitive coupling reduces and the inductive part becomes dominant, so that twisting becomes progressively more beneficial. Twisting together power conductors (circuit impedances of a few ohms) is therefore good practice.

Balancing the circuit relegates the capacitive coupling to a common mode effect, since the I C induced onto one conductor appears in the same sense and with the same amplitude as that induced on the other, so that the circuit's common mode rejection works to maximum effect. The crosstalk is then determined mainly by residual inductive coupling. This will be sensitive to an odd or even number of half-twists, or more properly to the differences in voltages induced in the enclosed area at each termination.

Twisting the pair does nothing directly to reduce common mode coupling with the cable; the common mode currents that are present will radiate just as effectively, and external fields will induce common mode currents just as effectively, as if the cable was untwisted. The key aspect which is improved by a good twisted pair is the conversion from common mode to differential mode or vice versa, i.e. the LCL .

Structured cabling and the UTP/STP debate

The proliferation of Local Area Network and other data communications applications in the last decade has resulted in a specification for cable performance, originally in EIA/TIA 568 and more recently in ISO/IEC 11801 and EN 50173. These documents refer to "Category 3" and "Category 5" cables, with further categories in preparation. The categories define performance specifications for particular applications, Cat. 3 referring to operation up to 10Mbps and Cat. 5 up to 100Mbps. Key cable characteristics are near-end crosstalk, attenuation and impedance up to 100MHz.

The purpose of these specifications has been to allow a building to be installed with "structured cabling" without a detailed knowledge of the applications that the cable system will support in future. These requirements can be met by UTP cables and this has allowed the implementation of low-cost LAN installations -- in fact it could be said that the large installed base of UTP cabling has driven the design of successive generations of LAN architectures -- but there is as yet no specific requirement for the EMC performance of the cable system.

Since installations nowadays do have EMC compliance issues, this situation has led to a rash of claim and counter-claim by cable suppliers, that a particular cabling system has proven EMC characteristics and is better than some other system in this respect.

This has particularly surfaced in the argument over whether shielded twisted pair (STP) is better or worse than unshielded. In practice, the argument is irrelevant: an equipment designer has to ensure that his product is compliant when connected to a particular cabling system. If that system requires shielded cable, then it is unreasonable to expect a UTP installation to achieve the desired compliance status. If the equipment has been designed to use only unshielded cable, then it will be impossible to terminate the cable shield correctly and a shielded cable installation may well be worse than a proper unshielded one, especially if inadequate ground/earthing arrangements are made. Although it might seem that screened cable will always be better for EMC purposes, this hinges on the methods used to install it. Apparatus that has been specifically designed for a UTP interface can be perfectly acceptable, provided that the cabling system itself is properly installed.

Ribbon cable

Multi-conductor ribbon cable is a very common and popular method of running multiple signals, particularly parallel data buses, between and within apparatus. Its popularity is mostly due to the ease with which it is terminated, using insulation displacement connector (IDC) techniques. Its correct use inside an equipment enclosure is important to the equipment designer but less so to the system designer. But when it is brought outside an enclosure, difficulties can arise.

The disposition of the conductors in a ribbon cable is rarely within the control of the system designer, but it has a serious impact on the interference coupling to and from the cable. The issue is, as always, the enclosed loop area for signal and return currents. All too often, the circuit 0V, which forms the return for all the wideband signals in the cable, is relegated to a single conductor, which in the worst case is located on one edge of the ribbon. --- why this is the worst case: for the example of a 50- way ribbon of lm length, the loop area for the signal on the opposite side of the ribbon is 635cm^2 - which makes a mockery of other good EMC practices such as a well laid- out PCB!

--- Loop area issues in ribbon cable

Leaving a large area between signal and return is inviting differential mode coupling into and out of the ribbon and hence immunity and emissions problems.

Bringing such a ribbon cable outside a protected enclosure is not advisable. If it is going to be run within a cabinet, all the precautions advised on routing within cabinets are essential.

Improvements in EMC performance of ribbon cables can be achieved by three methods:

++ multiple 0V return conductors adjacent to each critical signal conductor. This is easy to implement at the equipment design stage but not thereafter, and is costly (potentially doubling the number of conductors) and difficult in pin-limited applications;

++ using ribbon cable with an integral ground plane layer. This is a better electrical solution than (a), but is difficult to implement adequately in production since a mass-termination method of connection to the ground plane layer is needed. It is generally unpopular;

++ fully shielded ribbon cable. If properly terminated this can give good EMC performance, but all the issues relating to screened cable become relevant.

The manufacturers of shielded ribbon cable normally use the foil-and-drain wire method which forces a "pigtail"-style termination, compromising the available shielding performance. Also, shielding is least effective at the edges of the ribbon, so the most critical signals should be placed in the center;

++ "round and flat" cable rolls a ribbon cable up into a tube and sleeves it with an overall braid screen. Externally it looks just like any other multi-way screened cable, but it allows the use of mass-termination connectors with a proper 360-degree screen termination and so avoids the "pigtail" problems of flat screened ribbon cables.

If any of these precautions are implemented, external runs of ribbon cable are useable.

But bearing in mind their difficulty, all in all, ribbon cable for wideband data transmission outside a protected enclosure should be avoided if at all possible.

--- Ribbon configurations for improved EMC

Next: Installing cable systems

Prev: Unscreened cables

top of page  Article Index   Home