EMC and Coupling

Source, coupling path and victim

Electromagnetic compatibility invariably has two complementary aspects. Any situation of incompatibility must have a source of interference emissions and a victim which is susceptible to this interference. If either of these is absent, there is no EMC problem. There is also a third factor: there must be a coupling path between source and victim, which can be through a direct connection, through proximity, or through radiated energy. The same equipment may be a source in one situation and a victim in another.

-- Coupling via the power line

--- The three parts of an EMC situation

The various possibilities for the nature of the coupling path are explored below.

Direct coupling

Coupling via power or signal lines

The easiest method of coupling to visualize is when there is a direct connection between the source and the victim. The most typical example of this would be via the power supply. Disturbances present at the supply port of the interference source are fed onto the power supply conductors, along which they propagate, and from which they enter the power supply port of the victim. The power supply network itself is usually regarded as a passive player in this scheme. The disturbances can be propagated along it either differentially, that is between its conductors, or in common mode, that is along all its conductors with respect to a remote ground/earth reference. In most coupling situations, both modes occur, though one is usually dominant.

The total coupling path can be modeled as a noise source with its own impedance Z S feeding into the characteristic impedance Z 0 of the power network treated as a transmission line, which then feeds the load impedance Z_L of the victim. Clearly the level of disturbance which ends up at the victim will be a function of these impedances, which are complex and dependent on frequency. For the mains supply network in particular, Z_0 is heavily affected by loads which are attached at other points in the network; between 2 -ohm and 2000 -ohm depending on frequency and time of day can be observed, although the characteristic impedance of the copper wiring itself is reasonably stable and predictable. The impedance of 50 -ohm/50gH between each phase and ground/earth, which has been developed by CISPR to cover the frequency range 9kHz to 30MHz for conducted emissions tests, is an average value which was determined by measurements across a wide range of installations in different countries. (Note that this is quite different from the standardized low frequency source impedance between phases of 0.24 + j0.15 ~ which is used for flicker tests.)

Whether or not the mains network should be treated as a transmission line depends on the length of wiring between source and victim, and the frequency of interest. For longer distances than a quarter wavelength, transmission line parameters are essential; shorter than this, the impedances can be represented by a lumped approximation. At or above 30MHz, this implies that distances longer than 2.5m should be treated as transmission lines. At, say, 1MHz, the appropriate distance is 75m.

The amount of disturbance power that is fed into the network from the source, apart from being determined by the available voltage, is also affected by the impedance match between the source and the line. Maximum power transfer occurs when the source presents a conjugate match to the line; in the context of the CISPR impedances quoted above, that is when the source impedance is 50 -ohm plus whatever capacitance is necessary to resonate with 50~tH at a given frequency. The same will be true of the victim. Of course, in real life interfering sources and victims don’t present such a perfect match, and so maximum disturbance coupling rarely occurs. The purpose of interference suppression filters at the mains ports of equipment, and for that matter anywhere along the power network, is to deliberately increase the mismatch and therefore reduce the power transfer even further, usually by several orders of magnitude. The degree of mismatch is affected by the impedances presented either side of the filter, which is why real life filter attenuation performance rarely resembles that which is presented in manufacturers' data.

Coupling via the power network is not the only example of direct coupling, though it’s the most significant because there is usually no specific relationship between the source and its victim, so that the systems designer has no way of predicting the exact interaction between them. Signal and control lines between different equipment can also carry disturbances, and the same coupling analysis can apply to them.

Common impedance coupling

A second form of direct coupling occurs between two nominally separate modules which share a common impedance. The most frequent example of this is a common ground/earth or power supply connection. --- two modules each of which is ground/earthed to a single busbar, which is then taken to the system ground/earth reference by a further wire. Module 1 may have internal noise V N which is coupled to its ground/earth wire, and referenced externally to the system ground/earth usually by stray capacitance C S or by connections to other modules. The current produced by this noise voltage flows not only in Module l's ground/earth wire, but also in the common ground/earth wire that connects the busbar to the system reference. Since this wire has a finite impedance () a further noise voltage Vin t is developed across its length which then appears between the busbar and the system reference, and hence is coupled to the ground/earth connection of Module 2.

--- Common impedance coupling

This mechanism occurs in all such interconnection cases, and of course very often the actual interconnections are more complex and may involve several modules, which are interconnected to each other via other cables as well as the ground/earth. If problems don’t arise despite this, it’s because the noise voltages generated are small, the modules are well-designed with respect to EMC, and because the ground/earth conductor impedances although non-zero are negligible at the operating frequency of the equipment.

Conductor impedances are not negligible across the whole spectrum of frequencies of interest to EMC, though. Not only that, but the collection of impedances which appear in any given disturbance current path are usually made up of both inductance (due to the ground/earth wires) and capacitance (due to stray coupling between the module and its surrounding structure). Such a mesh of interconnected L and C results in a number of resonant peaks in the total circuit impedance. If any of these peaks coincides with a particularly significant disturbance or susceptibility frequency, the effect of the coupling path is magnified.

To minimize the effects of common impedance coupling:

++ the coupled network must be carefully designed so that high interference currents don’t flow in impedances which are connected to circuits which may be sensitive to the noise; and

++ the impedance of unavoidable common connections should be kept as low as possible over as wide a frequency range as possible; this is one reason for the use of short, wide ground/earth bonding straps rather than wires.

Near field (inductive and capacitive) coupling

Modules don’t need to be directly connected for interference coupling to take place.

Whenever current flows in a conductor, a magnetic field exists around it; whenever a voltage appears between two conductors, an electric field is present. Each of these field phenomena is capable of inducing an interfering signal in a second circuit that is coupled to the field. The principles involved are deliberately used in transformers and capacitors; dealing with a near-field EMC coupling problem is, in principle, nothing more than controlling the unwanted transformers and capacitors in the structure. These are sometimes called "strays" or "parasitics" and become very significant at higher frequencies.

Magnetic or inductive coupling

The voltage induced in a conductor by a current flowing in another conductor in close proximity is

V N = -M di/dt

where M is the mutual inductance of the coupled conductors. This voltage appears in series with the desired signals in the victim circuit, and is unaffected by the circuit impedances. The mutual inductance is determined by the separation of the conductors, the length along which they are in proximity, their geometry, and the presence of any magnetic screening around one or other conductor. The most usual situation in which magnetic coupling is significant is when several circuits are run together in a single cable loom. In this situation the circuits are in close proximity for the length of the loom and their mutual inductance is high. This is one reason for recommending segregation and separation of different cable classes. The quantitative effects of separation distance are shown.

---Magnetic field coupling

Or triples for 3-phase delta supplies, or quads for 3-phase star supplies.

However if a return circuit is run adjacent to the signal (or power) conductor, the magnetic field due to the signal current is nearly cancelled by the magnetic field due to the equal and opposite return current. The greater the distance from the conductor pair, the more completely is the field cancelled. This is why signal and return, or power and return, pairs t should always be run together in any cable loom. The mutual inductance between cable pairs is inversely proportional to the log of the square of their separation distance, whereas that between two individual wires carrying different circuits is inversely proportional to the log of the separation distance directly.

A further situation in which magnetic coupling is significant is when a magnetic component is involved, such as a transformer or motor winding. The magnetic field around the core of such a component can be much higher than around a conductor carrying an equivalent current, but it falls off much more rapidly, inversely proportional to the cube of distance. Problems are usually only significant if a cable carrying a sensitive circuit runs directly past such a component, or if a magnetically sensitive device such as a sensor is placed right next to it. A small amount of separation will show a very large improvement in this situation-- e.g., a change in separation distance from 2cm to 20cm would show a 60dB (1000 x) reduction in coupling.

Magnetic screening is difficult to achieve: most conducting materials such as copper and aluminum are by themselves largely transparent to the magnetic field and don’t offer significant attenuation at low frequencies (see the discussion on skin depth). A magnetically permeable material such as steel or mu- metal is more effective but requires bulk for effective absorption and is difficult to handle. The only realistic solution to many magnetic screening problems is to surround the offending or victim conductor with a conductive tube along its length, within which an equal but opposite current is allowed to flow t, which has the effect of negating the magnetic coupling with fields outside the tube. If the tube is perfectly coaxial with the inner conductor and the currents are exactly opposite, the magnetic coupling is nil. This is one of the main principles by which coaxially screened RF cables work, but it’s only imperfectly effective for other structural geometries - the twisted pair being the next closest approximation.

--This is the principal reason behind the advice to bond cable screens at each end.

---Magnetic field attenuation

---Spacing between conductors

Electric or capacitive coupling

The complement to magnetic coupling involves capacitance between structures and is mediated by the electric field between them. When a voltage difference appears between two conductors an electric field is created; this field induces a voltage on the victim conductor of:

V_in = C_c x Z_in. dVs/dt

where V_in is the voltage induced on the victim circuit of impedance Z_in, by an interfering voltage V s (of negligible source impedance) coupled through a mutual capacitance C_c. Notice that (a) in contrast to the magnetic field coupling, the severity of electric coupling depends on the victim's load impedance, so that high impedance circuits are much more susceptible; and that (b) the source and victim circuits need to be referenced together. This is easily seen when the circuits share a common 0V or ground/earth reference, but E-field coupling occurs even between circuits that are totally isolated. In this case, there are two coupling capacitances involved, one between the source and victim nodes, and one between the reference points for each circuit.

--- Electric field coupling

The mutual capacitance appears between two voltage-bearing circuit nodes, through which the current is irrelevant, rather than between two current-carrying conductors. Of course, these may be one and the same structure, such as a pair of wires, but the distinction between current and voltage is important. The capacitance is affected by separation distance, geometry (in particular the area of overlap of the two nodes), the nature of the dielectric between the nodes, and the presence of any electric field screening between them. Since area is significant, capacitive coupling tends to be greater between large objects than between small ones, but this is often mitigated because larger structures tend not to carry high levels of dV/dt. Where they do, such as with switch mode power converter heatsinks, capacitive coupling is a serious threat.

Screening against electric fields is very much simpler than against magnetic fields.

Any conductive material will act as a barrier to the electric field; the lower the resistance of the barrier the better, but even a material of a few ohms per square (such as nickel paint) will severely attenuate the field. A partial screen can be effective, although some of the field leaks around the edges of the screen. The electric field is easily distorted even by dielectric materials, which makes for difficulties in measurement when this is needed, since any measuring probe will affect the field it’s supposed to measure. The important aspect of E-field screening is that the screen itself must not carry any interfering voltages, or these will capacitively couple (C_c) to the circuit that is being protected. Thus it must be connected to a point of low interference potential; this is naturally assumed to be "ground/earth", but the difficulty with practical implementation is that many "ground/earths" actually carry significant interference voltages, or there may be voltage differentials between nominally identical ground/earth points.

In detailed system screening design, the point of connection of the screen must be chosen carefully. A conducting structure that is unconnected to any potential is of no use as an E-field screen, though it may have other purposes.

Radiated coupling

So far we have considered magnetic and electric field coupling in isolation. At DC and low frequencies this is quite acceptable- it’s known as the "quasi-static approximation". But any varying electric field between conductors requires a current to cause the voltage to change, and any changing current will develop a voltage differential. Thus AC fields are inherently composed of both electric and magnetic components, and as the frequency rises so it becomes more difficult and less meaningful to treat them separately.

At a sufficient distance from the structure which is carrying the radiating currents and voltages, the magnetic and electric components resolve themselves into a propagating electromagnetic wave. The two component vectors are at right angles to each other and to the direction of propagation, and lie in a plane surface which can be visualized as expanding outwards from the radiator in all directions. In free space, at any point on this plane the ratio of electric to magnetic components is constant and equal to 120.rt or 377fL which is known as the impedance of free space. The amplitudes of the two components will vary in different directions away from the source as a result of the geometry and phases of the various radiating elements.

---The plane wave

The "sufficient distance" beyond which the plane wave region of constant impedance begins can be derived from Maxwell's field equations and is given by L/2rt, or roughly one sixth of a wavelength. Examples of transition distance would be 1.6m at 30MHz, 48m at 1MHz, 16cm at 300MHz and 48km at 1 kHz. Beyond this distance in the far field, coupling is radiative, that is the propagating electromagnetic wave induces voltages and currents in the victim structure as if it were acting as an antenna. The interference potential of the wave can be expressed equivalently as a power density (watts per square meter or milliwatts per square centimeter), or as an electric field strength (volts per meter), or as a magnetic field strength (amps per meter). Conventionally, in the frequency range of interest for EMC the electric field strength is quoted.

The radiated coupling both from a source and to a victim then hinges on the effectiveness of either as an antenna. The structures of non-radio electrical or electronic products are rarely designed with this purpose in mind, and they are inefficient converters of radiated energy at most frequencies, which is fortunate; though it’s also possible that, on occasion, a particular arrangement of elements can result in a high antenna efficiency at a particular frequency. Efficient antennas have arrays of elements which are intentionally laid out so that the currents and voltages are in desirable phase relationships at the resonant frequency of the structure, and therefore give maximum transfer of energy at that frequency.

Conversely, good EMC design to minimize radiative coupling consists of deliberately arranging the mechanical layout to stop resonances, damping those which are unavoidable and ensuring that the induced currents and voltages don’t couple well with the internal circuit operation.

Stopping resonances is difficult, especially given that most enclosures are rectangular in shape, so that a resonance will exist when each major dimension is a multiple of a half or quarter wavelength. In the range from 30 to 300MHz, any system enclosures of typical size will exhibit many such structural resonances. Damping them is more successful, and normally occurs anyway with the introduction of parasitic smaller structures (such as PC boards or internal cables) into the main structure; such damping can be enhanced by the deliberate introduction of RF absorptive material such as ferrite sleeves on relevant cables. The main weapon in good EMC design consists of separating the unavoidable interference currents that exist in the structure from the operational currents and voltages that exist in the circuit. This is the purpose of screening, segregation, ground/earth layout, and filtering.

The modes of coupling

One of the most important aspects in EMC is to understand the distinction between the possible modes of coupling. The basis for this distinction is the idea that two separate circuit paths can coexist in the same set of conductors. One of these is the circuit that was intended by the designer- signal and return, or power and return, along which the desired signal currents flow, "differentially", that is in opposition to each other.

The other is the parasitic circuit that is formed between this desired circuit and the structure within which it’s located. This is called the "common-mode" circuit, because the currents in the conductors are all flowing in the same direction. --- these different modes for a generalized apparatus with a mains power supply and a signal line. The arrows in the figure imply emissions; their direction could quite easily be reversed to imply susceptibility.

---Differential and common mode concepts

Conducted differential mode

Conducted coupling in differential mode on the power supply is probably the simplest to visualize. Interference appears between the phases ( L and N, DC + and-, or P1/P2/P3) of the power supply and is carried into or out of the equipment by the phase conductors only. Placing a filter in line with these conductors is the conventional way to attenuate this noise. Typical sources of emissions are switch-mode power supply or switching converter currents, and of incoming interference, fault- or lightning-induced surges are the most common. Conducted emissions tests on the mains port will measure half the differential component on each phase. A similar mechanism operates on the signal lines, although here the threat and method of dealing with it’s different: since in general signal connections are point-to- point, they don’t have the opportunity to pollute a wide area as is the case with power connections. Differential currents in signal lines are principally concerned with the signals themselves. If these are DC or low-frequency (e.g. for sensor or audio connections) then their interfering capability is low; filtering is straightforward and is aimed principally at improving immunity to induced signals. If the signals are wideband (data or video) then there will be both interference capability and susceptibility issues, and filtering is more difficult. Either screened cables or special protection measures at the interface are needed. However, the main problem with such signal lines is radiated rather than conducted coupling.

Conducted common mode

The power supply phase conductors, and the signal line conductors, also carry conducted interference in common mode. In this case the interference does not appear between the conductors. It appears on each conductor with reference to a third point, and the interference currents flow in a loop which includes this third point. In the case of the power supply, there are two possibilities for the third point: it can be the safety ground/earth wire (common mode) or it can be the external structure (common mode). Although the safety ground/earth is usually connected to the external structure at some point, there are differences in these two modes. The most notable difference is that in the first, the currents remain within the power cord- flowing down the phase conductors and returning via the safety ground/earth, whereas in the second, all conductors including the safety ground/earth carry common-mode current which returns via a separate path. Clearly for Safety Class II apparatus (without a safety ground/earth wire) only the second mode applies.

Sources of common mode emissions are much harder to visualize and predict, and also harder to control. They are usually associated with internal high frequency functions within the equipment (such as microprocessor clocks, and also switch-mode supply oscillation) which are not intentionally or directly coupled to the power port but nevertheless appear there through stray coupling. Similarly, incoming common mode interference (typically fast transient burst noise, and radio frequency signals) is coupled into the internal circuits via such parasitic paths. Simple filtering between the phase conductors has no impact on this mode of interference. Mode (A) coupling can be filtered by a common mode choke and parallel capacitors between phase and ground/earth (which worsen the leakage current pollution of the ground/earth network), as is usually found in most bought-in mains filter units. Conducted emissions tests measure the L-E and N- E voltages separately; if there is no differential component then common mode (A) signals are indicated directly. Mode (B) coupling cannot be dealt with by capacitive filtering since there is nowhere to connect parallel capacitors to; common mode chokes in all lines including the safety ground/earth can help, but otherwise the only successful solutions involve structural remedies to the equipment itself.

Common mode noise in the signal lines is equally significant. The comments above regarding filtering are equally applicable, with the added problem for wide-band signals that capacitive filters will affect the wanted signal as much as the interference.

As well as stray coupling within the equipment, another source of emissions occurs for wide-band signals: through leakage between the cable and the environment, some signal current "escapes" from the differential circuit and returns through the structure, thus resulting in a common mode component of the signal. The degree to which this occurs is predictable from the cable parameters, and cables for wide-band applications may quote their performance in this respect as "longitudinal conversion loss".

Conducted common mode interference is more of a problem in general than differential mode, because its coupling paths include physical structures which are normally not designed for the purpose. Consequently, …

++ its effect is difficult to predict and control;

++ it can change with time because of uncontrolled structural changes;

++ it can pollute a variety of unrelated equipment;

++ the currents can flow within a large and uncontrolled loop, increasing their potential for radiated coupling.

Radiated differentia/mode

The same conceptual circuits can be used to visualize radiated coupling.

Because the efficiency of a coupling structure peaks when it’s close to a quarter- or half-wavelength in dimension, at the lower frequencies the major mechanism for radiated coupling is via cables, which are generally longer than other elements of the system. ----- relates the half-wavelength dimension to frequency and to typical sizes of structures. From this it’s clear that equipment enclosures are likely to dominate the radiative mechanism above 100-200MHz; long cables will dominate it below 30MHz. If the coupling is between two structures in close proximity, stray capacitive and inductive (near-field) effects dominate and wavelength plays a less important part, and the coupling can be expected to occur over a wide range of frequencies.

----- of Frequency versus half-wavelength; Frequency 0.5.~ Typical structures 10MHz 15m Long cables; 50MHz 3m Medium cables, large cabinets 100MHz 1.5m Short cables, medium cabinets; 300MHz 50cm Medium cases, 19" rack enclosures, internal wiring 600MHz 25cm; Small cases, PCBs.

In differential mode in cables, the return current path is known and (unless the circuit has been poorly designed) it will be in close proximity to its send path. I.e., the magnetic fields due to the currents from each conductor will tend to cancel each other, and the electric fields from the voltages on the conductors will tend to be concentrated between them. Radiated coupling is therefore minimized when the conductor pairs are kept as close together as possible, as in a cable loom, and best exemplified by a twisted pair. Nevertheless, at high frequencies the small area that remains between the conductors is still capable of radiated coupling and this puts a limit on the amplitude and frequency of wide-band signals that can be transported along cables of a given geometry.

Within equipment it’s often harder to maintain the optimum geometry for all circuits; internal wiring runs and the constraints of PCB layout tend to compromise the rule of minimum enclosed loop area. Thus radiated differential coupling with the internal circuits takes on a greater importance, particularly for frequencies where the structures are resonant, typically above 100-200MHz. A screened enclosure creates an intentional barrier to this coupling mode.

Radiated common mode

The common mode current path, by its very nature, breaches the optimum layout requirement of close proximity of signal and return. Often there is no control over cable routing, and therefore no control over the enclosed area of the common mode loop, which can become very large. Only when a cable is run for its whole length against a structure that intentionally carries the common mode return current, such as a properly bonded cable conduit, is the coupling loop area controlled to a minimum. In other circumstances the cable acts as a reasonably efficient antenna, with maximum efficiency at its resonant frequency. Worse, it’s entirely possible (and indeed normal)

for common mode currents to flow in the screen of a screened cable. The design of the screen should ensure that internally-generated common mode currents don’t transfer to the outside of the screen and that conversely, incoming interference currents that inevitably flow on the outside don’t transfer to the internal circuits. The quality of the screen and its termination at either end are crucial factors in this respect.

Radiated common mode coupling is not limited to cables, although because of their length they assume a great importance. Any conducting structure will carry common mode currents and will act as a radiator or receptor. Current flows and voltage differentials occur on the outside of metal enclosures, and at the frequencies at which they are resonant, they are most efficient at radiating them. Since metal structures can't be avoided in most systems and installations, good EMC design requires care that they are not excited by interference currents generated by the system, and that the currents which are developed by external interference sources are not then transferred to the system's operational circuits.

This description of common mode coupling between structures and circuits, which occurs both within cables and within the equipment itself, should emphasize that limiting transfer between modes is an important goal of interference control. The transfer impedance of a given structure is a measure of this parameter and is discussed in more detail in the next few sections, on grounding or ground/earthing, enclosures and cabling.

Protection measures

In the context of electronic products, EMC protection can be applied at the circuit level, at the interfaces, and at the enclosure. The desired degree of protection can be obtained by trading off techniques at each of these levels against each other to achieve a cost-effective optimum.

The system designer does not have this luxury. System components are "black boxes" whose internal workings cannot be modified, and whose interface parameters are already fixed, which is why EMC aspects are so important in purchasing. The only variables the system designer has to work with- assuming that the black boxes themselves are already specified- are the interconnections between the component modules, the physical layout of the modules, and the possibilities of electromagnetic containment. All of these impact the coupling to, from and between modules, rather than the performance of the modules themselves.

Subsequent sections in this guide will deal with each of these areas in detail. Here we simply enumerate the measures which can be taken at the system level to minimize the effects of the coupling paths that have just been discussed. These can be conceptualized as:

++ creating zones of different degrees of interference potential (partitioning the system);

++ erecting barriers between the zones;

++ applying protection at the electrical interfaces across the barriers;

++ deliberately designing the ground/earth reference network as a sub-system in its own right.

Zoning and barriers

The concept of zoning is very simple: the operating environment of a system or installation is partitioned into a number of different classes of zones, within which different levels of EMC protection apply. Two zones is the minimum for this approach; three is quite reasonable, but more than three would be unusual and only necessary for special cases. --- the concept graphically.

--- The concept of zoning

In this example, Zone 0 is essentially the unprotected environment- the "outside world", so to speak. In an industrial control system this would be the shop floor, or in a building management system it would be the outside, or parts of the building without electrical services if this were a typical office or hotel, say. The significant aspect of this zone is that no special EMC precautions are taken within it. Equipment located in Zone 0 must comply with all the EMC requirements for emissions and immunity that are applicable for the general environment.

Zone 1 is a protected area within the general environment of the system. Within this area lower levels of immunity can be tolerated, as can higher emissions, since the effect of the zoning is to provide some level of isolation from the environment. Alternatively, a higher degree of reliability is offered to equipment with a standard level of compatibility. The nature of the protection can vary, but two typical approaches would be to create an area within which the ground/earthing system is properly controlled, and/or to screen the walls, floor and ceiling of a single room. A Zone 1 area might be used for the control room of an industrial plant, or a safe room for the main servers of an office computing network.

Zone 2 is a smaller area of even greater protection. It could be located within Zone 1, or more rarely it could be an area within Zone 0 which has a greater degree of protection than Zone 1. The most usual approach to implementing a Zone 2 area is to provide a fully shielded enclosure such as a racking cabinet. Within this could be housed very noisy or very sensitive equipment such as motor drives or medical or scientific instrumentation - not both together, of course! The zoning principle requires that the boundaries, or barriers, of each zone are defined, and that the protection offered by each zone is quantified. The screening effectiveness of an enclosure, e.g., could define the difference between Zone 1 and Zone 2. Given a known or assumed level of disturbance and susceptibility in the external environment (Zone 0), the corresponding levels within each zone can be determined by the attenuation across the zone boundary. This can then be used to define the requirements for the equipment that will be located in the zone.

It should be clear that whatever services, electrical or otherwise, cross the boundary must be treated to maintain the required attenuation. This is the function of interface protection (filtering and suppression).

The various EMC phenomena may not all be attenuated equally at a zone boundary.

E.g., a controlled ground/earthing regime may not be as effective as a shielded enclosure for reducing field strengths, but can reduce the amplitude of transients and surges within the zone. It’s entirely reasonable, and good practice, to match the zoning philosophy and techniques to the needs of the equipment within the zones and the environment in which they exist.

Interface protection

The attenuation across a zone boundary must be applied to the electrical services to the same degree as is offered by the physical arrangement, such as screening or ground/earthing.

Filtering and/or transient suppression across the appropriate frequency range for every unscreened cable that crosses the boundary is necessary. This is not difficult to do for power cables, but signal and control lines may pose a greater problem. For this reason it can often be helpful to partition the whole system so that the minimum number of difficult-to-treat cables cross the zone boundaries.

Screened cables don’t need filtering, although surge protection may occasionally be required on internal conductors. However, it’s essential to bond the screen to the ground/earthing structure at the boundary, whether this is a screening wall or an ground/earthing bar. This ensures that the interference currents on the screen are returned to the ground/earth rather than penetrating from one zone to another, which would compromise the attenuation otherwise offered at the zone boundary. This bonding requirement applies not only to electrical screens but to any conductive service that crosses the boundary, such as pipework, conduits or ducting.

Grounding (aka ground/earthing)

A well designed ground/earth structure can contribute greatly to reduced interference coupling within a protected zone. This is because the grounding/earthing controls the common mode currents. Interference currents that remain don’t generate large potential differences across the zone, so that each item of equipment within the zone is subject to lower electromagnetic stress. Conversely, noise currents developed by equipment within the zone are returned directly to the source and don’t propagate significantly beyond the zone structure.

In order to achieve this function the ground/earth structure must be regarded as a component of the system in its own right, and designed as such.

Next: EMC and Mains harmonics

Prev: Electromagnetic environment: Phenomena

top of page  Article Index   Home