EMC is impossible
By J. M. Woodgate B.Sc.(Eng.) C.Eng. MIET SMIEEE FAES Hon FInstSCE MIOA
EMC is impossible
OK, attention-grabbing title. But it's necessary sometimes to look at what we are trying to achieve with all these standards, Directives and Regulations, and how we are trying to achieve it, on the understanding that the headline is unavoidably true. It is impossible to make electrical and electronic products that have infinite immunity and zero emissions. Somewhere, sometime, Electro-Magnetic Compatibility will not be achieved.
Put simply, we want all items of equipment that are likely to be in the same space to work correctly. The 'same space' would be a whole country or even continent in the case of conducted disturbances on the medium or high-voltage power grid, but would be a living room in the case of an LED lamp and a DAB radio. That's a pretty wide range. Another wide one is the frequency range. We don't often have to consider permanent magnetic fields now that cathode-ray tubes have marched into history, but in some countries, railway systems use 16.67 Hz at high power and that produces both electric and magnetic fields. At the other end of the spectrum, most existing standards nominally stop at 300 GHz, but only a few actually set limits above 10 GHz. However, CISPR is planning to look at extending requirements up to 40 GHz, and ITU-T is looking at systems using frequencies of hundreds of gigahertz.
What this means is that a huge range of technologies is required to combat potential interference. Military EMC technology has existed for nearly 100 years, and the main solution is to put things in stout metal boxes, interconnected by screened cables. For the military, this is affordable and acceptable, especially as it makes the equipment more 'soldier-proof'. This leaves only receiver antenna inputs requiring specific immunity measures. But this solution isn't either affordable or acceptable for civilian products, both for consumer use and often for professional use.
We now have, in some but not all countries, control over low-frequency conducted emissions on the mains supply. When first introduced, these were highly controversial, but the development of the switch-mode power supply and automatic power factor correction has largely eliminated the controversy. These techniques bring very significant benefits; higher energy efficiency and higher real power input for a given supply current. In fact, the reduction in harmonic current emissions is almost a side-issue! There is still work to be done on the 2 kHz to 150 kHz 'gap' to keep the relevant experts usefully occupied for a few more years.
It's easier to discuss emissions, because whatever produces it, 'an emission is an emission'. Immunity is much more difficult to discuss in general terms, because it depends very much on the type of product. A washing machine needs quite different considerations compared with a computer, even for ESD, because of its metal case. But since we can't eliminate all possibility of interference, or even for sure reduce it to extreme rarity, we have to accept a compromise that reduces the number of interference cases to an acceptable minimum. Even that is a complex issue: it's known that the vast majority of interference cases that do not have economic consequences, such as interrupting manufacture, are never brought to the attention of spectrum management authorities. Those involving the military, radio astronomy or the civilian emergency services are exceptions. Nevertheless, the present system seems to work well enough.
The big elephant in the room is 'measurement distance' for radiated emissions. Close to a source, there is no fixed ratio between the electric and magnetic field strengths and phases, as there is in the 'far field'. Quite where the far field begins depends on how close you want the ratio and phase to be to their theoretical values at a very long distance. The limit of the near field region is often stated as c/2pf, where c is the speed of light and f is the frequency. Some say that the far field begins at that distance, while others see a 'transition region', maybe up to 2c/f (twice the wavelength) or even more.
Below 30 MHz it is physically very difficult to make measurements far enough away (beyond 10 m) from the source of the emissions, so extreme precautions have to be taken to obtain repeatable measurements in the near field, where the field strength (electric or magnetic) varies a lot from place to place. But even so, separations between potential emission sources and potential victims are very often far less that the measurement distances specified in standards. For a mobile phone, which intentionally emits at 2.4 GHz, the far field stops at around 12 cm, and a victim (which has to be immune to this intentional radiator) may in practice be much closer than that. Manufacturers have to take such issues into account, over and above complying with what is written in the standards, otherwise product sales, or the backflows of rejected purchases, are likely to be disappointing.
Another elephant or, because it's rather shadowy, it might be a yeti, is the maintenance of EMC during product lifetime. The EMC Directive, for example, is silent about its applicability in respect of a product that has been in use for a significant time, such as a few months. Some argue that it still applies, others that it doesn't. Whichever assumption (not 'interpretation' because there is no ambiguous or incomplete wording to interpret) you favour, there is no doubt that emissions can increase over time. For example, drying out of the main filter capacitor of a DC power supply increases conducted emissions. Equally, if immunity is partly secured by a jointed metal enclosure, corrosion at the joints can seriously reduce the screening effectiveness. Manufacturers would be wise to take this into consideration; a filter capacitor with a higher temperature rating might be considered, for example.
You may be assured that most of the people involved in writing EMC standards, or at least those who contribute most to the content, are well aware of these issues and are doing the best they can to make the 'impossible' actually possible. Maybe they don't always get it right; maybe you can see a better test method or a more sensible way to formulate a limit. Don't just hide it, or complain that the standard is poor. Through EMCIA your ideas can be put to the BSI committee for consideration, and there is no 'not invented here' prejudice, but you can expect eagle-eyed scrutiny, which is, after all, completely justified. If your idea has merit, it will be taken up to IEC, CISPR or CENELEC.