The technique of ElectroMagnetic Compatibility (EMC) is, put simply, the engineering process that ensures that your television, video, computer, DVD player, electric drill, mobile phone, vacuum cleaner, kettle and so on can all operate simultaneously without interfering with each other.
Now, the more astute among you will already have noticed that many of these items do in fact interfere with each other. Anyone who has put their phone on the dashboard close to their car radio, or tried to watch terrestrial television while their mother vacuums the living room will testify to this. This does not imply that the whole technique is a futile waste of time; rather it illustrates that EMC is an extremely convoluted, unpredictable and difficult business, and above all is always subject to compromise. It has often been quoted that EMC is more of a black art than a science. This is untrue – it is quite clearly a science, just not a very exact one.
This entry is intended as an introduction to the subject; not a guide to designers. The engineering design involved in mitigating EMC problems can be extremely specialised, is constantly evolving, and should be the subject of another complete entry.
To give an overview of the sort of things we are talking about here, it is probably necessary to provide a few examples of real life scenarios that would be addressed under the heading of EMC, some of which you may already be familiar with.
There were stories a few years back of drivers being locked out of their cars at airports when the tiny signals emitted by their remote keyfobs were completely swamped by the nearby radar transmitter. Similar, although less severe 'breakthrough' problems can be experienced when you here the interference created by a mobile phone, CB radio or by a car's ignition on your transistor radio, or when a radio telescope is effectively blinded by local interference. Strong magnetic fields can erase credit card strips and distort television pictures; the interaction of the solar wind and the Earth's magnetic field can destroy a satellite's electronics or induce huge surges in power lines on Earth. Transient electromagnetic fields, from small electrostatic sparks to lightning strikes to the huge pulses caused by nuclear explosions, can irreparably damage electronic equipment.
Probably the most-studied EMC effect, though, is the ability of radiated noise (interference) from one device to cause another to malfunction temporarily. This is the reason you are not allowed to use mobile phones in the vicinity of safety-critical systems, such as in a hospital or on an aircraft. If you used a laptop computer in the vicinity of a radar, a high-powered communications transmitter or an electric furnace, you would probably find it crashed a lot. By virtue of extensive research and testing, however, devices that are designed to work together, such as PCs and printers or TVs and VCRs, will usually do so with no mutual interference.
Wiggly Amps - The Problem
The problem is electricity, and the rather redundant fact that everything electrical relies on it. A simple direct current flowing down a wire will have predictable effects; the resistance of the wire will probably cause it to heat up, and a magnetic field will appear around the wire. Unfortunately, a direct current is pretty much useless for anything except transmission of power over short distances. Everything else relies on a varying signal, either as a regularly varying or alternating current (AC) power supply signal or an irregular information-carrying analogue or digital signal.
For descriptive purposes, let us imagine a simple sinewave with a single, fundamental frequency f1. Any wire or conductor bearing this signal is capable of acting as an antenna and can radiate a certain proportion of the signal as an electromagnetic wave with the same frequency. Similarly, any nearby piece of wire will act as a receiving antenna and a current with the same frequency will be induced on the wire. The amount of signal radiated or received depends on a multitude of factors, but in general the higher the frequency, the more a signal is likely to misbehave.
In the real world of course there is no such thing as a simple sinewave and most real signals are a hopelessly complicated mish-mash of the fundamental frequency, a load of harmonics2, switching transients3, sidebands4, mutual interference products5, the latest boy band hits from local radio and anything else that happens to get picked up or created by the imperfect electronics in the system.
So much for analogue signals. Now consider a simple digital signal as a stream of alternating 1s and 0s, repeating at frequency f, giving a bit-rate of 2f. Mathematically, the rectangular pulse of the 1 can be considered to be a series of sinewaves, starting with one at frequency f, one at 2f, 3f, 4f and so on. The straight edge of the pulse, if perfectly vertical (ie, zero rise time), represents a sinewave of infinitely high frequency. In practice it is not infinite of course, but if you plotted the frequency spectrum of this simple square wave it would still look like a dog's breakfast. Roughly speaking, the faster a digital signal gets, the wider the slot of the frequency spectrum it occupies6. Unfortunately, these frequencies are not merely mathematical; they are real, and they will radiate and cause interference in the same way as intentional analogue waves.
And more problems
Collectively, the ability of Radio Frequency (RF) radiation to wreak havoc is called ElectroMagnetic Interference (EMI); conceptually a subset of EMC. But on top of all the RF rubbish, you may have very low frequency or DC surges down the supply line. Your local substation may have been hit by lightning, or unusually strong solar activity may have induced huge currents in very long stretches of supply cable. When electrical equipment is turned on, there will be an initial surge as the switch makes contact; the switch may bounce a couple of times causing higher frequency spikes, and then the current will oscillate a few times before it settles down to a steady flow. Heavy equipment drawing large currents and electric motors in particular may cause temporary fluctuations (transients) that propagate for many miles over power supply grids.
Yet another potential problem is electrostatic discharge (ESD). This is the annoying spark that happens when the charge you build up by walking on nylon carpets discharges itself into the metal of a car door. On the scale of a transistor on a microchip, that spark looks like a lightning strike and can cause serious damage. These events are temporary and rare, but have the potential to cause serious damage rather than just momentary interference.
And it's getting worse...
So, every piece of electrical equipment both transmits and receives everything it creates or processes to and from the world at large. This only becomes a problem when the unwanted signals are large enough to swamp the wanted ones, cause temporary or permanent failures or otherwise interfere with the operation of a product (or more frequently a completely different product). The chief reasons that EMC is becoming a more and more important aspect of design are:
- we are using more electrical equipment;
- that equipment is operating at higher frequencies and data rates;
- the equipment is using lower voltages to reduce power consumption, rendering it more vulnerable to interference;
- more and more of the signals are digital, and are cleverly packaged to fit as many separate channels as possible into the available spectrum.
For example, the basic clock frequency of a mid-range personal computer is now higher than transmission frequency used by GSM phones and most radars, never mind the harmonics, and few PCs make any attempt to limit their 'pollution'.
So the world's standards bodies set about producing a set of specifications that all products would have to comply with, placing limits on both the product's emissions and susceptibility to interference. Within Europe for instance, all electrical equipment sold must carry the CE mark7. This means the product meets the requirements of the EC EMC Directive 89/336/EEC8 (as well as the appropriate safety directives) by complying to the most appropriate 'harmonised' standard, eg, EN 50081 for emissions and EN 50082 for immunity.
Internationally, the International Special Committee on Radio Interference (CISPR) is the EMC-related arm of the International Electrotechnical Commission (IEC) and is responsible for maintaining EMC standards accepted by many countries worldwide. In the US, the Federal Communications Commission (FCC) produces equivalents, in Japan it is the Voluntary Council for the Control of Interference by Information technology equipment (VCCI). Many other countries are catching up. As most equipment nowadays is marketed worldwide, it may have to be validated against many of these (and other) standards. The military have their own standards that are predictably much more stringent than commercial ones, in part contributing to the vastly inflated cost of military equipment and ensuring that the armed forces will never have the cheap and plentiful commercial off-the-shelf equipment they so desperately want.
Any particular standard, or set of standards, may require the following aspects of a product's performance to be tested:
- Radiated emissions - the amount of radiation a product can emit, as both electric and magnetic fields, measured at a set distance.
- Conducted emissions – the amount of garbage a product can place onto any wire or cable connected to it, for example the mains power cable or a network connection.
- Radiated susceptibility or immunity – the level of radiation a product can resist before it falls over, again both electric and magnetic fields.
- Conducted susceptibility or immunity – the amount of interference a product can withstand on its power, control or signal interfaces before it falls over or loses the signal. This may cover power surges as well as high frequency interference.
Each test will specify maximum limits that the equipment must cope with. These limits will probably be different for different frequency bands, and will be chosen to reflect real-world scenarios. Some of the more extreme tests may not require the product to carry on working without error, but merely survive. A typical test regime may also cover other electromagnetic problems, such as electrostatic discharge, electromagnetic pulse (EMP) protection (lightning strikes or nuclear explosions), radiation hazards to personnel (RADHAZ) or specific absorption rate (SAR) testing, unintentional emission of compromising signals (known as TEMPEST or van Eck phreaking), mutual interference between emitters and the danger of emissions igniting nearby flammable substances or explosives.
There are many problems with standards compliance though, not least of which is the sheer number of standards. Imagine a personal digital assistant or similar device that uses multiple communications protocols, some of which may require intentional emitters such as GSM/GPRS, Wi-Fi, Bluetooth, etc. Every mode of operation of each of these protocols (which may have multiple power settings or frequency bands) must be verified, often to multiple regional standards and taking into account different configurations of the device and the many accessories that may be connected to it. The number of tests simply mushrooms, and will continue to do so until a global regime is implemented.
A further complication is deciding exactly what constitutes a test failure, particularly with susceptibility testing. All modern data communications protocols include some kind of error checking and correction, so even if interference does cause the loss of a few bits, the user probably wouldn't even notice. A few clicks on an audio telephone line probably wouldn't raise any eyebrows either. Conversely, the loss of a few bits inside a microcontroller in a chemical plant may be catastrophic.
Yet another problem is that it is often left up to the manufacturer to choose the most appropriate standard, which may not be the most stringent one available, or the one that best matches the electromagnetic environment the product ends up in. So, like every other aspect of EMC, standards compliance is a useful risk-mitigation exercise but is no guarantee that a product will actually work.
Traditionally, products have been designed to meet their functional and physical requirements, built, tested and then been sent for EMC testing. The problem with this approach is that in the event of test failure - which frankly is pretty inevitable if EMC has not been considered in the design - the resultant redesign can be quite extensive and very expensive. Conversely, incorporating best EMC practice at the beginning can lead to the product being over-engineered and expensive to produce.
Curbing the problem at source
Protection against surges is relatively easy - if the equipment is deemed sensitive, then a surge suppressor is inserted on the power or signal line. Protection against ESD is pretty much a matter of ensuring that any spark has an easier path straight to ground than anywhere important. Protection against EMI is somewhat more difficult, as all sorts of strange effects become significant at higher frequencies. It is however vital to mitigate against EMI at the design stage. Basically, to limit emissions in the first place you want to:
- Keep internal frequencies as low as you can, eg, by making clock frequencies as low as you can get away with to do the job
- Limit the rate of change of voltage and current (dV/dt or dI/dt in calculus-speak) by, for example, smoothing the sharp edges of digital pulses (slowing the rise time) by filtering
- Separate 'noisy' (eg, power supplies) and 'clean' (eg, sensitive transducers) circuits and wiring
- Keep circuit paths as small as possible, and keep send and return paths together
Locking it away
To protect against external interference, and to contain internal emissions you should use:
- Physical Separation - the easiest and often most effective protection measure. As EM radiation follows an inverse-square law, its signal strength reduces by a factor of four for every doubling in the distance it travels.
- Shielding (or screening) - enclosing the offending item in an earthed metal box will attenuate much of the radiation. Any parasitic currents induced in the metal by the field inside will be channeled straight to ground before they can re-radiate. Cables may also be shielded by a foil or copper braid screen surrounding the conductors and earthed at both ends. If the connectors allow the screen to be bonded to the connecting equipment 360 degrees around the cable at each end then all the better9.
- Filtering - to limit conducted interference along a wire, a filter may be inserted to remove all frequencies except the wanted signal. In practice, this is nearly always a low-pass filter that impedes everything above a certain (cut-off) frequency, on the basis that most parasitic signals are likely to be of high frequency, and the high-frequency ones cause the most trouble. Clearly this has limitations, as any parasitic signals within the pass band will still get through, and a low-pass filter cannot be used if the wanted signal is high frequency itself. A filter will also degrade most digital signals, even if the bit rate is well below the cut-off frequency, as all the high-frequency components that make up the rectangular pulse (see above) will be filtered out.
- Waveguide-beyond-cut-off (WGBCO) - an aperture will act in a similar way to an antenna; that is a signal will freely pass through if the gap is larger than a half-wavelength. Therefore the smaller the gap, the higher the minimum frequency that can pass through. This is known as the cut-off frequency. Providing the maximum frequency you are trying to contain is lower than the cut-off, it won't get out. Now, if you also increase the depth of the aperture walls, relative to the width of the gap, the length the signal has to travel is increased and it is less likely to escape. WGBCOs can be simple metal tubes resembling short lengths of pipe, or take the form of a honeycomb vent over windows or ventilation holes.
Or avoid electrons altogether
Since the problem is electrical, it follows that a non-electrical device will not cause any problems. A purely mechanical computer is probably not a practical replacement for your home PC, unless you have a very large home, but a mechanical governor on a machine in a very noisy environment may be more appropriate than a computerised one. Best of all, an optical fibre is made of glass, doesn't conduct electricity, doesn't radiate and doesn't pick up unwanted interference. What goes in one end pretty much comes out the other, with few if any surprises.
The Harsh Reality
It would be possible in theory to design a vacuum cleaner that didn't disturb the six o'clock news, or a car radio that was immune to a mobile phone right next to it. However, these would be bulky and expensive and nobody would buy them. On the other hand, if you designed an aircraft engine control computer that threw its toys out the pram every time the pilot used the PA, or a pacemaker that went haywire in the presence of an electric toothbrush, nobody would buy them either (or nobody in their right minds anyway).
Modern PCs have the potential to be an EMC nightmare - and as they are effectively cheap mass-produced consumer goods, they are an EMC nightmare. You may have experienced problems due to power surges, and installed a surge-protector in the mains supply. More likely, you have experienced unexplained crashes that you may have attributed to the quality of certain manufacturer's software – it is however also possible that the glitch was caused by interference. The effects of interference are notoriously hard to predict – there are design guidelines and best codes-of-practice - but when it comes down to it there is no way of knowing exactly what will happen in real life.
So to summarise, there are test standards but it is notoriously difficult to know what the problems will be in advance, whether a product passes or fails is often open to interpretation, the limits are geared towards the specific product type and the particular market, and the tests cannot hope to address every possible scenario. At this point you might understandably be considering jacking in your career in electronics to do voluntary conservation work in Madagascar.
But you are made of sterner stuff than that. You arrive at the test house with your exquisitely-designed whojamaflip with its fully screened body, internal components wrapped neatly in tinfoil, conductive glue dripping from every joint and every connector filtered to the limit of your budget, only to find that the plastic fuse holder is transmitting like an airport landing beacon, or that a mysterious 7th harmonic of the unexpected interference product of an internal clock frequency and the mains is resonating strongly enough to make Jodrell Bank think that a gamma-ray burst has just gone off. You may find that if you move a wire a millimetre to the left, a troublesome peak at a particular frequency has been replaced by an equally troublesome peak at a completely different frequency. You may uncoil a cable between two devices and find that the noisy signature of one device now appears as an almost perfect replica on the other. You may find that your whatjamadoobrey is fine on its own, but blows all its fuses when sitting six inches from an empty soup can.
But whatever happens, you will do what you have to do get it through the tests, and then unleash it on the market where it will face thousands of unexpected and untested situations and, fingers crossed, will not cause any air crashes or reactor meltdowns during its lifetime.