EMI as a power distribution problem

To an extent, EMI (Electromagnetic interference) has always been a power distribution problem. Harmonics of the supply frequency on AC systems have caused issues for decades, probably since humans first started to harness the power of electricity. The problem has only grown as the number of devices requiring DC, rather than AC, power has increased. As a result, there are a number of standards governing acceptable levels of harmonic distortion. Some of these cover the distorting devices themselves and apply to equipment manufacturers. Others govern distortion present on the network and apply to system operators.

More generally, EMI refers to phenomena that have the potential to interfere with the normal operation of equipment. Ignoring harmonic distortion to the supply, this generally covers radio waves, whether they are generated intentionally or otherwise. The array of standards governing EMI cover a very broad range of frequencies: from 150 kHz to frequencies of several gigahertz. The motivation for this is that these frequencies have historically been used for transmitting and receiving radio signals - anything from long wave radio to RADAR - and even low levels of stray signal could interfere with sensitive receivers.

Harmonic distortion, in the power distribution sense, covers a much narrower band, that starts at or near DC and extends to about 2 kHz. It does not necessarily only include harmonics of the supply frequency, since subharmonics, interharmonics and, indeed, DC all cause issues for AC power distribution and therefore need to be regulated.

This leaves a gap not covered by existing standards between 2 kHz and 150 kHz. The table below summarises the different frequency bands (The observant reader may notice that there is no mention of frequencies above 3 GHz. This is another range where regulation is a concern, but for very different reasons. In any case, this article is not concerned with such high frequencies) - the highlighted column is the gap we are concerned with. Until recently that gap has been largely ignored, but it is becoming increasingly obvious that there is a need to regulate emissions at these frequencies and efforts have been made towards developing new standards to do just that. It is also this range of frequencies that I intend to discuss in this article, first by highlighting why there is an issue to begin with and then by explaining why it is not just a problem for equipment operation but for power distribution as well.

The 2 kHz to 150 kHz range was not historically a major problem: power supplies tended to create interference in the lower harmonic range and other equipment tended to use higher frequencies. Equipment operating at frequencies in this range tended to have naturally high immunity to external interference anyway, since an antenna suitable for transmitting or picking up signals in that range would need to be several hundred meters in length, so creating one by accident is unlikely. That's not to say there was never a problem, it was just a relatively easy problem to overcome through good design. However, advances in power electronics have resulted in switching power supplies with significant power ratings operating within this unregulated band and the number of these supplies in use is increasing rapidly. The ever increasing popularity of power line communication technologies has also contributed to a rise in interference in this band. All this new interference is conducted, rather than radiated, and so there is no need for an antenna to pick up the signal, it simply travels along the power line it is connected to. The lack of regulation does not help. In fact, it creates a perverse incentive to operate equipment in this range: I personally know of at least one power electronics engineer who chose a switching frequency very close to 149 kHz for the purpose of avoiding the requirements set out by standards. I am sure there will be many other examples of similar, cynical design decisions being made.

As I've already pointed out, much of the interference in this unregulated frequency range is the result of equipment injecting, either deliberately or incidentally, on to the cabling used for power distribution. Personally, I am not in favour of the widespread use of power line communications for precisely this reason: they use cables for a purpose they were never intended for, with little concern for the unintended consequences. However, I suspect that the technology is unlikely to go away on the basis of my personal view. But it should not be too difficult to see why this interference might be considered a power distribution problem. I've been told of cases where cabling has failed as a result of overloading, not at the supply frequency, but at much higher frequencies. There is also the potential for these unwanted signals to effect the behaviour of existing protective devices which were never intended to cope with such interference. In fact, it a problem that has existed for quite some time, since substations have made use of power line communication for quite some time for control purposes; the difference is that in these cases, the infrastructure operator is in control of the equipment being used and can plan accordingly.

But there's more. Skin effect, a phenomenon which causes the resistance of cables to increase at high frequencies, is significant at these unregulated frequencies. The effective resistance of copper cabling could be more than ten times higher at 10 kHz than at the supply frequency of 50 Hz. Higher resistance tends to mean higher losses and an increase in operating temperatures. The problem is worse when the interference reaches transformers, which suffer from eddy currents in the iron core and a number of other frequency-related phenomena, all causing additional heating. Although individual sources of interference are unlikely to contribute much to losses on their own, the aggregated contributions of thousands of devices is going to be much more significant.

It will be interesting to see how power distributors address the problem as the already large number of interference sources increases. I suspect that they will require some significant upgrades to infrastructure, something which is already necessary as a result of changes in how we generate and use electricity. There's already a push for new standards in this area and I'm sure that will continue, although the conflict between cost and convenience to users and reliability and maintainability of the network from the operator's perspective will no doubt continue in one form or another