EMC – Low frequency conducted emissions

The truth about IEC61000-3-2 and –3, and their EN clones

How can I claim to tell the truth when all sorts of different stories are circulating?

Because I'm a member of IEC SC77A/WG1 and WG2, responsible for these standards.

Control of low-frequency conducted emissions began with the voluntary IEC standards IEC555-2 and -3. These emissions are of two kinds:

IEC555-2 dealt with the generation of harmonic currents by non-linear loads. These cause capacitors, motors and transformers to overheat. Peak clipping of the voltage waveform reduces the efficiency of switch-mode power supplies and can thus also cause overheating.

IEC555-3 dealt with the amplitude modulation (termed 'voltage changes') of the supply voltage by loads drawing time-varying currents. This causes filament lamps to flicker, which is disconcerting, especially for household consumers. The sudden connection of numerous loads, when supply is restored after an outage, can cause deep amplitude modulation and serious problems, if the loads draw 'inrush currents' far in excess of their steady-state currents. Circuit-breakers and fuses may operate, and damage may even be caused to wiring.

In due course, these standards were reviewed and steps taken to replace them by new standards in the IEC61000-3 series of EMC standards. Unfortunately, it wasn't realised (this was in 1988 or thereabouts) that the effect of the European EMC Directive would be to make these standards quasi-legal documents, instead of the voluntary standards they superseded. This meant that the precision of the language should have been orders of magnitude higher, but it wasn't. Some of us tried to improve the drafts in the early 90s, through the national standards committees, but it was very difficult to have much influence after the work had been in progress for some years.

Both of the initial standards, IEC61000-3-2 and IEC61000-3-3, apply only to equipment rated at up to 16A per phase. For higher currents, IEC opted to produce Reports, IEC61000-3-4 and IEC61000-3-5. However, the European Commission refused to recognise Reports as suitable for implementing the EMC Directive, so it has been necessary to replace them by IEC61000-3-11 (for voltage changes) and -12 (for harmonics) and neither are published yet.

IEC61000-3-2 and -3 (and -11 and -12) are 'product-family standards', so they have authority above the Generic Standards (which refer to them) . But they are exceptional in that:

- they are produced by IEC SC77A instead of a product committee

- the 'product-family ' is 'anything that can be connected to the public low-voltage electricity supply'. Note that that is a restriction: the standards do not apply to something that will not conceivably be connected to the public low-voltage supply. But the standards DO apply to both professional (as defined in IEC61000-3-2) and household equipment. The requirements, however, and not identical, because there is so much more household equipment than professional equipment connected to the public supply. Much professional equipment is actually connected to an industrial MV or HV/LV transformer, and doesn't affect the public supply in the same way as a directly-connected load.

The European standards body CENELEC adopted the IEC standards verbatim, and they were notified in the Official Journal (OJEC) as suitable for demonstrating conformity with the EMC Directive. (Note, products conform, manufacturers comply.) They are due to come into mandatory effect on 01-01-01. However, the pressure for amendment of both standards prevailed and work began in 1998, in IEC SC77A/WG1 and WG2.

Both amendments are being processed as fast as possible in IEC. However, CENELEC decided to use a shorter approval procedure (Unique Acceptance Procedure - UAP) in the case of just the 'harmonics' amendment. That this is illogical must be obvious, but when one is dealing with CENELEC....

It is intended that from 01-01-01, manufacturers will be able to choose, for 3 years, whether to apply the unamended or the amended standards. Almost all will find it better to choose the amended versions. In the case of the amendment to IEC/EN61000-3-3, which will not be published in time, it is likely that the authorities will take no enforcement action in respect of products that would be admissible under the amended standard.

It' s quite impossible to give details of all the changes: that could in fact only safely be done by publishing the amendments. The main changes are:

Class D is no longer defined by a 'special waveform', and applies now only to television receivers and desktop computers. ( Clearly, a switch-mode power supply (SMPS) intended for use in a PC would need to take that into account.) In the future, other products may have to be added, but only after a full and searching investigation.

The change of lower limit of application of Class D limits from 75 W to 50 W is NOT implemented, and will not be implemented unless National Committees vote to include this change in a later edition.

Note: IEC SC77A/WG1 is already working on a new edition, probably timed for implementation in 2005.

Higher limits for voltage changes are introduced for some equipment that is not considered likely to cause annoying flicker or unacceptable voltage dips. For example, some high-power garden power appliances do not normally cause flicker because they are used only during daylight hours.

The method of measuring voltage changes has been clarified in cases where the IEC60868 or IEC61000-4-15 flicker meter is not used.

SMPS are by no means the only source of harmonic current emissions. 'Linear' power supplies draw substantially the same short current pulses and produce the same harmonic spectra. The only difference is that the current pulses of an SMPS are typically shorter and taller, thus producing somewhat stronger harmonics. The main problem is that all the power supplies produce harmonics that add up nearly arithmetically – there is hardly any phase difference between them. And where the SMPS are fed between phase and neutral of a three phase system, the triplen harmonics (multiples of 3) add up arithmetically in the neutral conductor. Some installations had thinner neutrals than phase conductors, but the neutral current can exceed TWICE the phase current (the theoretical value is about 2.8 times!), so it is not surprising that neutral conductors have overheated.

If you don't think that there is a problem with harmonic emissions, examine the voltage waveform of your local residential mains supply (carefully!), especially in the evening when all the television receivers are in use. You will see flat tops on the sine waves, caused by the big current pulses (typically peaking at 60 A) drawn by each receiver.

Inrush current is not only bad news in terms of flickering lamps, it can damage components within the culprit equipment. The inrush current of an SMPS is highest if the switch is closed at a voltage maximum, but for a 'linear' supply, with a transformer, especially a toroid, the situation is more complex. The filter capacitor causes the largest inrush if the switch is closed at a voltage maximum, but the transformer has two possible actions. If the switch was turned off, and the internal arc extinguished, at a positive-going zero-crossing of the voltage waveform, and the switch is subsequently closed at or near a negative-going zero-crossing, there is minimal effect. BUT, if the switch is closed at or near a positive-going zero-crossing, a very large inrush current occurs. Why? Well, the original switch-off actually left the core magnetised. It is a closed magnetic circuit, so even though it is of magnetically 'soft' material, it does retain most of its magnetisation. When the switch is closed at a positive-going zero-crossing, the core is driven hard into magnetic saturation, the primary inductance of the transformer is dramatically reduced and a huge magnetising current flows.

Because of this behaviour, voltage changes are assessed on a statistical basis, taking an average of several measurements using the product's own mains switch (which affects the results). Typically, the average value is about half the worst-case value, and it is this average which is compared with the limit. Note, too, that the limits are expressed as half-cycle r.m.s. values, NOT peak values. This implies that a worst-case peak inrush current of well over 70 A may be tolerable.

Copyright © J. M. Woodgate August 2000

This article may be reproduced as a whole or quoted, provided the author's name is attached.