Cabling: Coupling to, from and within cables

Cabling between equipment is a principal route for interference coupling, both into and out of equipment. Cables themselves are of course passive, and therefore do not come within the scope of any EMC regulations. Although there are occasional sales claims to the effect that this or that cable type "complies with" the EMC Directive or FCC rules, such claims are entirely spurious. However, use of a particular type of cable may enable a system or apparatus to comply with EMC requirements. Choice of cable is a crucial aspect of system design for EMC. Just as important in a systems context is what you do with the cable in order to install it. Layout, routing and terminations will all affect coupling of interference to or from the cable and can therefore have an impact on the overall system performance.

The equipment designer may need to specify a particular cable type for a particular interface in order to meet EMC requirements for the equipment; the system designer will have to ensure that this is used, and will need to implement the best installation practice.

This section will look first of all at cable coupling, and then explore issues of use of screened and unscreened cables, before finally discussing the questions of classification, segregation, routing and parallel ground/earth conductors.

Coupling to, from and within cables

The distinction between differential and common mode coupling has been discussed in section 4.2.4. Nowhere is this distinction more important than with cables.

Differential mode

In virtually all cases the wanted or intended signal current(s)within the cable are flowing in differential mode. That is, two conductors in the cable are devoted to each signal: one carries the "flow" or "go" current and the other the "return" current. The designers of the equipment at each end have allocated connector pins for each conductor and the source and load circuits are connected to these pins. If the circuit could be considered entirely in isolation, the only current that would be flowing in these conductors would be that due to the intended signal.

This description applies equally to power or signal circuits. Examples of such circuit pairs might be mains live and neutral, DC supply plus and minus, each signal and its 0V in an RS-232 data line, audio or data signals in a telecomms network, RF or wideband data signals in a coaxial cable where the inner conductor is paired with the outer sheath. Whether or not the circuit is "balanced", i.e. the voltages are symmetrical about ground/earth or some other fixed reference, is not relevant to the differential mode classification - although it is important in other respects. The important point is that the current in one half of the conductor pair matches and reflects the current in the other half. In mains filter parlance, this mode is sometimes known as "symmetric" mode because the current is symmetrical across the two halves (in the live line and in the neutral line) of the filter. In a three-phase power circuit, the differential mode is not related to the question of whether the load is balanced across the phases; in the three- phase supply cable, the differential mode currents (including the neutral in the star configuration, or not in the delta configuration) always sum to zero.

It is also the case that in a multi-conductor cable, differential mode can exist quite happily when only one conductor takes the return current for all the other conductors.

This is the case, e.g., in the RS-232 serial data specification. The cable may be carrying anything from two (TXD and RXD) to twenty data signals, depending on the complexity of the application; but there is only one connection for the 0V reference, which takes all of the return currents for each of the signal circuits. As far as the whole cable is concerned though, the differential mode condition is satisfied as long as the net total of all the currents in the cable is zero. Because of the shared return, this is in many respects an EMC-hostile type of connection, which is one of the reasons for the limitation of the RS-232 specification to low data rates and short distances; but its advantage of low cost often outweighs these disadvantages, hence its huge popularity.

Magnetic coupling in differential mode

It was seen that coupling between the magnetic field around a circuit and the current flow within it is related to the enclosed area of the total loop, provided that the current of interest stays within that loop. This is directly relevant to the above description of differential mode currents within a cable. The loop area for a total circuit is dominated by the distance between its cable go and return conductors, multiplied by the cable length. The larger this area, the greater the resulting magnetic coupling.

However, we are usually more interested in the coupling with a small segment of the cable, especially if this runs near to a severe emitter or a sensitive potential victim.

In this case, it is not a question of coupling with the whole loop, but with two closely-spaced conductors carrying equal and opposite currents. If the two conductors were co- located (a physical impossibility, but approached by the coaxial cable configuration, and to a lesser extent by the twisted pair) the net current would be zero and therefore the net emitted magnetic field would be zero. In practice, each conductor generates an equal and opposite magnetic field which is separated slightly in space. As the observation point moves away from the conductors, this separation becomes proportionally less significant and the combined field decays rapidly towards zero. The impact of this geometrical result is easily calculable for any point in a plane normal to the two conductors, at frequencies for which the distances involved are small compared with a wavelength. No magnetic field is coupled in the longitudinal direction, i.e. the direction of current flow.

This characteristic of magnetic coupling results in the advice to keep all cable pairs as closely coupled within a cable bundle as possible.

--- Differential mode current in cables

---Magnetic field at a distance from a conductor pair

---Electric field cable coupling

Electric coupling in differential mode

Electric field coupling is related to the voltages on the conductors rather than to the current flowing in them. Here, we are interested in the voltage developed between a conductor pair as a result of an external electric field applied across them, or conversely the electric field which appears as a result of the voltage applied between them. The coupling is sensitive to the polarization, that is no coupling exists for fields normal to the plane of the wires; maximum coupling exists within this plane. For a constant field strength, the voltage induced is proportional to the conductor spacing, hence again the advice to minimize this separation. It is also dependent on the exposed surface area of the two conductors and on the impedance of the circuit.

The external field geometry and distribution is heavily affected by all other external conductors, whether they are part of an intended circuit or not, and therefore it is far less practical to calculate and predict electric field coupling than it is magnetic. Fortunately, the electric field is far easier to screen against and in practice it only becomes important for unscreened cables or for those whose screens are imperfectly implemented.

High frequency coupling in differential mode

The above descriptions, separating electric and magnetic coupling, can only be applied accurately for low frequencies where the cable length is a small fraction (of the order of a tenth or less) of a wavelength. When the dimensions become equivalent to or greater than a wavelength, the phase variation of the fields with spatial position is important. The resulting phases of the induced currents and voltages vary with their position along the cable. At the point of interest- usually the end of the cable where it connects to the equipment- the quantities can be anywhere from zero to a maximum, depending on the relative phases at each incremental point along the cable, and vary periodically with frequency and the spatial distribution of the fields. In principle the induced interference at any given frequency can be calculated if the geometry and electrical parameters of the cable and fields are known. This is of no practical use for the systems installer though, since these parameters are either unknown, uncontrolled or subject to frequent change. In most circumstances the "envelope" of maximum possible coupling is the most useful aspect that can be predicted.

---Typical variation of cable coupling with frequency cable resonances ; Coupling

Common mode

Cable conductors will also be carrying currents in common mode, simultaneously with the intended and desired signals in differential mode. The common mode currents are almost invariably unintentional or undesired side effects of the required signals, or they are wholly a result of external disturbances. A well-designed cable installation is very largely one which is successful at keeping the common mode disturbances separate from the differential mode desired signals.

Put simply, the common mode currents are those which are flowing on all conductors of the cable equally in the same direction, and therefore returning via some external path. They can be seen as due to two principal causes:

++ I_flow g: I_return in a conductor pair, so that (I_flow – I_return ) :g: 0; the difference between these differential mode currents is the common mode current and is directly related to the signal carried on the cable;

++ external influences, unrelated to the power or signal being carried, couple with the cable equally on all conductors and induce current equally on all conductors - the cable appears as a single conducting structure.

These two causes can be modeled separately but both are usually present in a typical EMC situation.

--- Stray impedances converting differential to common mode

Differential to common mode conversion

The first cause can be seen as a mechanism by which the intended differential currents are partially converted to common mode. The equivalent circuit for this can be represented in which the real-world parasitic components are shown along with the designer's expected circuit. If we ignore the distributed parameters of the cable- that is, the frequency is low enough that the cable length is much less than a wavelength- then each of the three parts of the circuit is modeled as having lumped stray reactances to some external reference. This is often assumed to be " ground/earth", but in reality it can be any indeterminate potential, due to the aggregate of conducting structures that are coupled to the cable and the apparatus at each end.

The plus and minus terminals at the source equipment end will have ZS+ and ZS_; the terminals at the load end will have ZL+ and ZL_. These impedances will be dominated by stray capacitance if the circuit is galvanically isolated, but will be a mixture of resistive, inductive and capacitive if the circuit is ground/earth-referenced at some point. The cable conductors will have stray capacitance CC+ and C C_ along their length, and also will be inductively coupled to the external reference conductors, as shown by LC+ and LC_. Strictly speaking, the cable resistance should be included as well, but it rarely contributes significantly to the common mode conversion mechanism.

If all of these stray components were balanced- that is, if all those with the subscript + equaled their opposite numbers with the subscript - - then no conversion from one mode to the other would take place. This is never the case in reality.

Imbalances between Zs+ and Zs_, and ZL+ and ZL_, are present at each end because of differences in PCB track length and area, connector and circuit asymmetry. Even if the cable were perfect, these imbalances would result in some difference between I+ and I-. This difference is related to the ratio of the desired circuit impedances to the undesired impedances; that is, the lower the value of ZS+/_ and ZL+/_ by comparison to the circuit impedances Z s and Z 1, the greater the effect of their imbalance. If these impedances are mostly capacitive, the imbalance effect becomes more significant at high frequencies.

The difference between I+ and I- is of course the common mode current ICM, which flows through the stray impedances at each end and whose return path is via the nominally " ground/earth" reference. One of the functions of the Parallel Ground/earth Conductor (PEC) discussed later is to provide a controlled path for this return current which is closely coupled to the cable, in contrast to the uncontrolled and chaotic paths that would otherwise be taken without it. These uncontrolled paths still exist even with a PEC in place, but the PEC is designed and terminated to ensure that it provides a preferential, i.e. low-impedance, route for the common mode return currents.

The cable also contributes to the circuit imbalance, through differences in Cc+ and Cc_, and to a lesser extent in LC+ and LC_. These differences will be partly due to imperfections in the cable construction, but mostly to variations in the environment through which the cable passes and with which it must couple - typically, the proximity of other metal objects. If, e.g., a cable passes edge-on near to some conducting structure then one conductor is nearer to the structure, and hence has a higher capacitance and/or partial inductance with respect to it, than does the other conductor.

If the cable geometry could be maintained perfectly flat against all possible nearby conductors then balance could be maintained, but instructions to this effect would not be popular with installation technicians.

Imbalances in cable strays can be negated to a large extent by screening and by twisting the conductor pairs, and these techniques allow a parameter to be developed for any cable type which is called "longitudinal conversion loss" (LCL). This is a measure of the degree of conversion from differential to common mode of the circuit and can be seen as a quality factor for a particular cable - the higher the value, the better. Some cable manufacturers are able to quote LCL figures for some of their range, especially those which are intended as wideband data cables, for which the impact of common mode conversion in controlling RF emissions is important. The LCL can be used to estimate common mode current emissions for a given circuit, as follows:

ICM (OBIaA) = VT(dB~tV)- LCL ( OB)- 20log10. ](2Z o 9 (Zcm + Zct)/(Z0+4Zcm)l

where:

V T is the differential signal voltage,

Z_cm is the common mode impedance of the item having the worst LCL, Zct is the common mode impedance of the item with the higher LCL, Z o is the differential mode characteristic impedance

Separately induced common mode currents

Common mode currents which are related to the signals carried by the cable are usually generated by the conversion mechanism described above. In contrast, there are also common mode currents on any cable which have no relation to the intended signals carried by the cable. These may be generated by the equipment at either end, and then radiate from the cable acting as a disturbance emitter; or they may be induced on the cable by external sources, in which case the cable is acting as the receptor of the disturbance, and conducting it into the equipment at either end.

--- Separately generated common mode emissions

Common mode emissions which have nothing to do with the signals carried by the cable are a common cause of failure to meet compliance requirements, and are often very hard for non-EMC engineers to understand. They are also becoming more prevalent, with the introduction of high-speed digital circuits (microprocessors) into many originally analogue applications. Such emissions can be seen to occur even when a single wire is connected to a supposedly fixed circuit node, such as circuit 0V. --- the equivalent circuit for this mode of emissions generation. The crucial aspect is the presence of an aggregate noise source V N which appears between the point of connection of the cable, and the external reference. This noise source is the result of the normal operation of the circuit, developing potential differences across the mesh of stray impedances that are formed by the PCB tracks and wires within the equipment. Aggregate V N values of only a few millivolts are enough to cause a breach of emissions limits, if connected directly to an external cable. Equipment designers minimize the problem by careful PCB layout to reduce V N, and by careful interface design to prevent it from being directly coupled to an external cable.

Notice that even the apparatus which forms the load for the wanted signal, if it also contains other operating circuits, can cause emissions of this sort (VN2). Power cables, which do not carry any intentional high frequencies, can still offer a coupling path for these emissions.

--- Cable acting as a disturbance receptor

The effect of the second mechanism is that the common mode voltages and currents which are presented at the connectors of the apparatus, may experience conversion to differential mode and then appear at internal circuit nodes differentially and cause malfunctions. This is the most common coupling route in cases of susceptibility. The conversion to differential mode acts in the inverse manner to that discussed earlier, as a result of differential circuit and stray impedances. Good equipment design is aimed, firstly at preventing the common mode signals from penetrating into the circuits, and secondly at laying out the circuits so as to minimize the common mode to differential mode conversion. Because coupling is reciprocal, these are the same techniques as are used to control common mode emissions.

As with the case of differential mode, the coupling can be classified as either via magnetic fields, electric fields or at high frequencies, a combination of the two.

However, the crucial difference is that now the relevant separation distance is that between the cable as a whole, and the reference structure that is carrying the common mode return current, in this respect the internal cable construction is irrelevant; the separation between the individual conductors has no effect on the coupling. The outer of a screened cable will carry common mode currents just as do the conductors of an unscreened cable. In fact the screen is intended to do this. Segregation and routing of cables is more of a vital factor in controlling common mode coupling, since the inverse square law with separation distance that is evident does not operate - the magnetic field around a cable carrying common mode currents decays only proportionally to distance, as does the electric field. For both emissions and immunity, then, the cable is acting exactly like an antenna, and its coupling efficiency is very much greater than in differential mode.

Crosstalk

The issue of crosstalk occupies the never-never land between EMC and signal integrity.

Crosstalk within a cable is not a problem for external (inter-system) EMC in the sense proposed earlier, but it does have a bearing on intra-system EMC, that is, the ability of a system not to interfere with itself. The problem is essentially one of coupling between separate circuits in a cable loom.

Visualize two circuit pairs in a single cable. Along the length of the cable there is distributed capacitance between every conductor and each of the other three conductors. Similarly, there is mutual inductance linking every conductor to each of the others. At frequencies where the cable length is much shorter than a wavelength, the L and C can be simplified to a matrix of discrete reactances, but at higher frequencies it is necessary to assign elemental reactances to an infinitesimally short length and then integrate these over the length of the cable.

--- Intra-cable crosstalk

The mutual capacitance and inductance between the two conductors that form each circuit pair are benign, and determine the characteristic impedance of that pair (Z 0 = {L/C, assuming no losses). In conjunction with the circuit driving and load impedances they will determine the bandwidth capability of the cable/equipment system. But the mutual impedances between the pairs are undesirable. These result in crosstalk interference between the two circuits.

Capacitive crosstalk

The voltages appearing on + and - of pair 1 are coupled by the mutual capacitances onto + and - of pair 2. The amplitude of the induced voltage is determined by the values of the capacitances and the circuit impedances, and the rate of change of source voltage (dv/dt). Balanced circuits, and a balanced cable construction which equalizes the capacitances, will minimize the effective crosstalk since voltages induced on or from the + conductor will be nearly cancelled by those induced on or from the - conductor.

Unbalanced circuits with high dv/dt and high impedances will be the most susceptible to capacitive crosstalk. Screening each pair individually will remove the capacitive crosstalk almost entirely, since the mutual capacitances between pairs are eliminated, to be replaced by mutual capacitance from each pair to its screen and mutual capacitance between screens. (Screens without 100% optical coverage, such as braids, will still allow a small amount of capacitance directly between conductors, through the gaps in the screen.) The screens must of course be connected to a fixed potential (which may be system ground/earth, or sometimes circuit 0V); voltages will still be developed longitudinally along the screens as a result of their resistance, and this along with other screen imperfections is then the limiting factor in capacitive crosstalk suppression.

--- Screening against capacitive crosstalk

Inductive crosstalk

Currents flowing in each conductor will induce a longitudinal voltage in all other conductors as a result of mutual inductance within the cable. The amplitude of this voltage is proportional to the di/dt of the source current and the mutual inductance linking the conductors; circuit impedances do not affect it. If the mutual inductance from one source conductor to each conductor of the other circuit pair were to be equal, then the same voltage is induced in each and the net effect on that circuit is nil. By itself this is not usually the case; but when the opposite sense contribution from the other source conductor is included, in a cable with good symmetry the total contributions can cancel each other. For this reason inductive crosstalk by itself is rarely as serious a difficulty as capacitive.

Distributed crosstalk

At high frequencies the cable must be considered as a distributed structure and the mutual impedances of elemental lengths have to be integrated over the whole length.

The phase differences along the length become significant, and the contributions of inductive and capacitive crosstalk result in constructive interference at each end at some frequencies, and destructive interference (nulls in the coupling) at each end at others. What is more, the sense of the wave travelling down the cable becomes significant, and it is necessary to talk of "near-end" crosstalk (NEXT) as distinct and different from "far-end" crosstalk (FEXT). We will not attempt to go into the detail of distributed crosstalk analysis.

Next: Cable screening techniques

Prev: Architectural shielding

top of page  Article Index   Home