This paper was originally presented at EMC `96 Roma, International Symposium on EMC, Rome, September 17-20 1996.
This paper discusses the EMC Directive 89/336/EEC from the perspective of experience gained in its application to date. The intent is to analyse this experience so as to provide a pointer to areas of possible improvement. A review of the purpose and salient features of the Directive is offered. Three broad areas of perceived deficiencies are then analysed. These are:
Particular illustrations of the issues are provided.
The EMC Directive 89/336/EEC has been in full operation now for approaching a year. It seems timely to review its application and consider how well, or otherwise it is succeeding in meeting its primary aims. Where there are difficulties, a clear discussion of these will help to progress the Directive's continuing implementation and improvement.
The author has been involved in advising electrical and electronics companies for several years on how to apply the Directive and its associated regulations to their own businesses and products. A very wide spectrum of types of company (from one-man bands to major multinationals) and product complexity (from swimming-pool pumps to electron microscopes) has been covered in this time.
The EMC Directive, one of a number of "New Approach" Directives, was initiated for two reasons:
Some legislative action to ensure product EMC, at least for emissions that may affect an innocent third party, is essential. Given the structure of the European Union, it is clear that such action needs to be taken at a pan-European level. Hence an EMC Directive of some description was inevitable. The EMC Directive as adopted has the following features:
Taking these points together, it is obvious that the most successful application of the Directive should be to items whose exposure to EMC issues is clearly defined, where the phenomena to be considered are well understood, and where the marketing strategy is such that both the point of supply and the point of use can be clearly established. This description applies to such products as domestic white and brown goods, information technology equipment, and professional equipment supplied as individual items.
Because of the cost involved in determining compliance with the requirements, it is also obvious that the Directive will be more successful in those instances where the product in question is made and marketed in volume. The overhead can then be spread over the total sales of the product, and care can be given to correctly assessing compliance. Vice versa, for low-volume or one-off products the cost is much greater, as is the incentive to skimp the compliance assessment.
The application of the Directive is least successful where one or more of the aspects listed above are ill-defined. This applies for example to products for use in unusual or multiple environments, or systems such as industrial plant or building controls where the contractual relationship between supplier and user is complex.
The title of the EMC Directive is "Council Directive on the approximation of the laws of the Member States relating to electromagnetic compatibility". The word "approximation" is important: it implies that member states are required to make their laws similar but not identical. This is indeed what has occurred. As a result of the vagueness of wording of the original Directive, many details of implementation vary between different member states. So the wholly admirable intention of removing technical barriers to trade, has merely resulted in such barriers being re-erected in a different place. Some examples, varying from minor to fundamental, see for instance [1] are:
Also because of the vagueness of the original drafting, there are many aspects, some technical and some commercial or legal, which have to be resolved individually in order actually to apply the Directive. Some of these were the subject of the EC's guidance document of 1993, revised in April 96 [2]. Even this document is equivocal in many places and needs further clarification. This is particularly true of the sections under chapter 4 on the application of the Directive to systems and installations. Despite having been re-written, the revised wording is no clearer on the approach to the conformity assessment procedure for such items. An indication of the Commission's own difficulty in producing this document is given by its lateness, at least six months after its originally intended publication date.
A serious complaint of industry has been that there is a paralysis or a lack of will on the part of the national authorities and the EC when they are called upon to provide a definitive interpretation of specific aspects. The Commission has made clear that it does not see it as its responsibility to provide such interpretation; it is the manufacturer's own responsibility to do so, and if he needs advice, he should seek it from a competent body. This raises questions both of conflict of interest, discussed further in 4.1.2, and of diversity of the resulting guidance. Although there are attempts to develop consensus on some aspects, these are hampered by the variety of views that can be taken by the interested parties.
The Commission's reluctance to take the lead in developing authoritative guidance could indicate its lack of expertise in this area, but it also gives the strong impression that it was inadequately briefed on the consequences of its decisions at the drafting stage of the Directive.
Perhaps the most significant aspect of the EMC Directive is its all-embracing scope. Pre-existing legislation was drafted to cover specific product sectors which were known to be responsible for the majority of actual EMC problems. By contrast the Directive covers all electrical or electronic apparatus capable of causing or suffering interference. It applies equally to series production in high volume and to one-offs made by a small company for a particular customer. It is these latter marginal activities which are most affected, and it is no exaggeration to say that if the Directive is rigidly enforced on one-off products then such producers will go out of business. This was presumably not the intention of the Commission, but it is the inevitable result.
There is of course a technical argument that, particularly for radio frequency emissions, the number of products of a given type introduced into the environment should not influence the need to ensure compliance with the limits: one interfering product in the wrong place can have unacceptable consequences. But this is not a practical argument nor is it supported by the CISPR approach. Practically, all CISPR limits are applied through the 80/80 rule, which effectively allows a small proportion of series produced items to be in breach of the limits and still allow the supplier to claim compliance. Therefore it can be argued that thousands or millions of mass-produced products which comply strictly with CISPR standards still pose a greater threat than a few low volume items which do not. This means that there is clear scope for a relaxation of the compliance requirements for low volume products.
A controversial issue is the insistence on legislation for immunity of all products within the scope. No other EMC legislation has yet attempted to go this far and consequently there is no experience in enforcing it, although there is a clear need for an immunity requirement for certain safety-critical products such as medical equipment and automotive control devices - precisely those products which are subject to their own Directives. Gonschorek has succinctly summarized the issue [3]:
Sufficient immunity is a quality factor of a product. And the question arises: `How far can the authority influence the quality of a product by law?' It cannot be the intention of the directive to require immunity testing of products that never failed in their applications.
There is a common misconception that there is some relationship between emission and immunity requirements, so that in order fairly to distribute the compatibility burden, the requirements for emissions limits and immunity limits must be shared. This idea totally ignores the reality that emissions and immunity limits apply to two quite distinct phenomena. RF emissions limits are placed on unintentional emitters in order to protect radio reception, they are specified in values of microvolts per metre or millivolts, and one cannot expect a radio receiver to be immune from the very signals it is expected to receive. On the other hand, immunity requirements are necessary because of the inherent nature of the operation of radio transmitters and other man-made and natural phenomena, and are specified in values of volts per metre or kilovolts (a million times greater).In other words, the two aspects of EMC address quite different issues and cannot be traded off one against the other.
There are several other Directives which contain EMC provisions. Article 1(2) of the EMC Directive specifically states that such provisions take precedence. Examples are the Medical Devices, Automotive and TTE Directives, and the civil avionics Regulation. However the degree of overlap is left unspecified. As a result it can be hard to tell whether the conjunction of a given product and a given EMC phenomenon is covered by the application of the EMC Directive, one of these other Directives, a combination, or not at all.
The CE Marking is intended as a passport to European sales, nothing else. It should not be used as a distinguishing mark or as an indication of quality. However, this is too novel for both suppliers and customers, who are more familiar with a whole raft of third-party approval marks. Superficially, the CE Marking appears to be just one more. As a result, customers are demanding the CE Marking where it is inappropriate, and suppliers have been using it unnecessarily in an effort to oblige their market.
A further problem with the CE Marking has been well illustrated in BSI News [4]:
First, its blanket nature means it cannot be used for product differentiation and secondly, the fact that it means different things for different products is likely to cause confusion both for the consumer and the installer. The inconsistency across the directives of the independent testing requirement makes it very difficult for purchasers to assess on what basis certification has taken place.
Or, to put it another way [5],
Safety marks came into being because the legal system did not reliably give consumers confidence in a product's safety. The underlying basis for applying the CE mark is no more reliable...
Allied to this, the concept of self-certification is also novel. Because there is no absolute guarantee of compliance - the harmonized standards have their own problems and in any case only provide a "presumption" of conformity, see 3.1 - manufacturers are forced into a strategy of risk assessment when deciding how to comply. Many manufacturers, wholly uncomfortable with the prospect of taking challenging and unknown legal risks themselves, are demanding to be spoon-fed their TCF certificate by competent bodies, many of whom are only too willing to oblige (see also 4.1.2).
The use of CE-marked apparatus in systems and installations is a good example of the technical worthlessness of the CE Mark. The systems manufacturer is expected to use compliant apparatus and to install it correctly. But the presence of the CE Mark does not of itself confirm that the apparatus will conform even if correctly installed. The system builder must seek further information, such as exactly what standards have been applied or what are the conditions within the TCF, before he can have confidence that he can properly claim compliance of his system.
Whilst it was not an explicit aim of the EMC Directive, one of its effects (as with many other Directives) has been to erect barriers where none existed before to imports into the EEA from countries outside it. This has naturally caused alarm in such exporting countries and their national industrial support agencies are keen to have information on the implementation of the Directive. Such agencies are doubly disadvantaged, because of their remoteness from those sources of information that exist as well as the vagaries of that information. There is considerable opportunity for mis-representation of the routes to compliance.
The EC's guidance notes state that
The manufacturer takes all measures necessary in order that the manufacturing process ensures compliance of the manufacturer's products described in the declaration of conformity with the protection requirements in the Directive that apply to them.
No more specific guidance is given as to what these "necessary measures" might be. It is up to the manufacturer to decide what level of assurance he needs that each individual product shipped actually complies with the Directive. CISPR standards state that
Tests shall be made either on a sample of equipment of the type using the statistical method of evaluation, or, for simplicity's sake, on one equipment only. Subsequent tests are necessary from time to time on equipment taken at random from production.
Most companies are thankful enough to have managed to prove that a single item complies. The prospect of ensuring compliance of all shipped units against all phenomena is so daunting that few have tackled it at all, let alone adequately.
Article 10.1 of the Directive allows for self-certification to harmonized standards. Clearly this route is dependent on the availability of suitable standards and CENELEC has been mandated to produce such. The compressed timetable available to CENELEC and IEC committees and the structure of the standards-making process itself, taken together with the technical problems peculiar to EMC standards, has meant that some of the standards so far produced have been unsuitable for the purpose of certification.
It is made clear in the EMC Directive that products that comply with harmonized standards may be presumed to be in conformity with the Directive's essential requirements. This does not guarantee that they are, yet it is commonplace to find industry assuming that compliance with the minimum requirements of the applicable standards is all that is needed to meet the EMC Directive. In fact an offence is committed whenever the use of a product breaches the essential protection requirements, regardless of whether or not the conformity attestation procedure has been properly complied with.
There are numerous instances where the application of standards clearly does not satisfy the essential protection requirements. A manufacturer or user is then unable to meet these requirements and avoid committing an offence, without taking significant further steps over and above what the standard instructs, which places him at a competitive disadvantage with less scrupulous operators.
Article 8(1) of the EMC Directive allows for the Commission's standing Committee to consider whether any harmonized standard does not entirely satisfy the protection requirements. There are many issues for this Committee to consider.
The most important standards that have so far been harmonized are the "generic" standards. These are specifically stated to define the essential protection requirements and are used as the basis for many of the emerging product standards. The phenomena covered by the initial generic standards for the residential, commercial and light industrial environments are
This list of phenomena is essentially restricted, for obvious pragmatic reasons, to those for which test methods have been established. Many other phenomena are actually relevant for the protection requirements. Listed in Schedule 2 of the UK regulations but not included in the generic standards are conducted low-frequency phenomena, radiated low-frequency phenomena, conducted oscillatory transients and CW RF voltages or currents, radiated transients. The problem is that it is inconceivable that, even in the long term, EMC control should be applied on a mandatory basis to all these phenomena.
Article 4 of the Directive (the essential requirements) stipulates that apparatus should not generate disturbances which would prevent "radio and telecommunications equipment and other apparatus from operating as intended (italics added). Historically, radio interference problems were most acute in the broadcasting sector, and CISPR's original efforts were aimed at dealing with these problems. Nowadays there is a much broader spread of susceptible apparatus, yet the inertia in standards development has prevented the new threats from being properly dealt with. This issue has been discussed in [5], where the authors claim
it is the influences of more than ninety years of radio regulation which have inhibited the broad conceptualization necessary to resolve EMC problems.
They list the faults as follows: standards are for testing, not design; not all the relevant spectrum is addressed; equipment characteristics are omitted; particular characteristics of electromagnetic environments are not considered; operating modes and procedures are omitted; the effects of aging are not considered; the susceptibility of the installed base of equipment is ignored; and incident-reporting problems are ignored.
Will the standards being produced to support the EMC Directive improve with time? There is a fundamental structural problem, in that standards are mostly produced at the international level and adopted by CENELEC. Neither structure is suited to producing EMC standards that will become mandatory requirements. Sectors of industry are either over-represented or unrepresented on the standard committees and there is no mechanism to prevent the domination of vested interests. In such a structure, the feedback that would lead to self-correcting improvements functions weakly if at all. Two examples can be quoted of standards appearing to be set by partial interests:
Because there are many different ideas about how to measure electromagnetic phenomena it is not surprising that different standards committees will take various approaches. This has led to many anomalies. One example is the CISPR decision that conducted emissions should be measured below 30MHz, radiated emissions above this frequency. This contrasts with the more flexible approach of TC110/WG2 in approving ENV50140/50141, in which the conducted/radiated breakpoint for RF immunity has been set at nominally 80MHz, with extensions upwards or downwards at the discretion of product committees.
Another inconsistency is the requirement in EN55020 to test immunity of broadcast receivers only up to 150MHz, whereas virtually all other RF immunity tests using IEC 1000-4-3 will extend up to 1000MHz. This is justified at least partly because "statistics of complaints show that for the time being immunity measurements at higher frequencies are not justified" [6]. It does not seem likely that statistics of complaints for other products do justify measurements at higher frequencies.
Goedbloed has argued [7] that basic EMC standards should be provided with a bibliography referring to technical information supporting the various decisions made and making them traceable. He defines "traceability" as "the property of an EMC test, according to which the rationale of each test item can be related to open references, usually international reports and publications." Such material is generally absent from CENELEC harmonized standards. Although this does not affect the use of the test in straightforward cases of declaring compliance, it severely restricts the utility of the document when questions are raised as to its validity and applicability [8]. One specific example of such lack of traceability is offered by the IEC 1000-3-3 flicker standard, which defines flicker limits in a particular way, but with no explanation or justification for these limits or for the method of measurement, which appears to be entirely arbitrary.
It is perhaps unfair to single out any one such standard - most are lacking this type of supporting information..
The normal gestation period for a new standard is around five years. The mandate on CENELEC to produce standards to cover a wide range of products on an accelerated basis, purely to meet the timetable of the Directive, has not allowed sufficient time for the optimum approach to be found for each case. CENELEC and IEC committees have done the best they can, but many standards have been released prematurely and with obvious errors, inconsistencies and omissions. Only a few examples must suffice:
One of the most common complaints from industry concerns the nature of the standards development process: firstly, that because of their deficiencies all standards are in a constant state of flux and revision and so present a "moving target"; secondly, that there is no official route whereby the most up-to-date version (draft or published) of an EMC standard can be found. It is also necessary to know when a new standard is referenced in the Official Journal, since it becomes effectively mandatory from that date. Manufacturers are forced to rely on commercial and costly providers of update services, even though these will normally only offer to update on published standards, not on drafts, or to rely on anecdotal information through contacts on committees or in other companies. For small companies this activity itself forms a significant and unproductive business overhead.
Because products may take years in development and have an extended lifetime, there is a need to anticipate future standard requirements several years ahead. EMC standards are for testing not design, and therefore apply once a product has been developed; it is not possible to "design a product to meet" an EMC standard, as it is with, say, safety standards.
The foreword of several harmonized ENs contains a clause which permits a manufacturer to continue to use a standard for a specified period once it has been superseded or revised, for production only, if he has already certified to it. Manufacturers regard this as helpful in reducing the burden of continued re-certification costs when standards are revised or newly published. Unfortunately the EC has declared this certification clause illegal and has demanded its removal. It remains to be seen what the outcome of this particular move will be, but it clearly adds to the uncertainty and confusion surrounding the use of standards.
Article 10.2 allows the TCF route if the manufacturer cannot or will not apply harmonized standards. The inclusion and specification of this route has significantly increased the confusion surrounding the implementation of the Directive.
In signing a TCF certificate, competent bodies take on the same liability for an offence against the EMC Directive as does the manufacturer. It is therefore only natural that they should be cautious in doing so and should demand a high level of certainty in relation to meeting the protection requirements. This diminishes the attraction of the TCF route for the manufacturer, who perceives a heavy burden of testing and documentation to satisfy the competent body. Some manufacturers have been quoted test and consultancy costs of many tens of thousands of pounds just for a single system. Yet for many types of product, especially large, distributed or unique systems, standards are useless and the TCF is the only viable route. This seriously disadvantages such system suppliers with respect to other industry sectors.
There is considerable confusion regarding the choice between the TCF route or self-certification, and competent bodies are keen to promote the use of their facilities. This can well result in naive manufacturers who consult a competent body for advice, being encouraged to use a TCF when this is not really necessary. It is, after all, not in the competent bodies' interest to promote the self-certification route, although to be fair, many competent bodies take their responsibilities in this respect very seriously. This problem is compounded because the authorities' response to all enquiries regarding matters of interpretation is to refer the inquirer to a competent body. There is a growing view amongst manufacturers that this is equivalent to giving a burglar the keys to the safe.
TCFs are most necessary for unusual or unprecedented product situations. In such cases there is no such thing as the "standard TCF". While the Association of Competent Bodies will be able to promote some degree of uniformity of approach, it is doubtful whether a comparable response from different competent bodies will ever be possible for the more unique applications. The consequence of this variability is that, in countries with many competent bodies, manufacturers will discover the lack of uniformity and treat the views of any one competent body with scepticism, if another has already indicated a contrary view. Alternatively, in situations where only one competent body is available, there is the danger of arbitrary decisions and no check on misuse of authority.
With respect to the actual content of the TCF, the UK DTI has published a suggested content for a TCF and the EMC Test Laboratories Association has similarly published a guide to assessment. These are helpful documents, but as mentioned above, they become progressively harder to apply in unique product situations. Each competent body has a different interpretation of what should be included in a TCF and how it should show conformity with the protection requirements. This leaves manufacturers who are "shopping around" for a suitable competent body understandably confused.
Several other Directives (notably the Machinery Safety Directive and the Low Voltage Directive) require the manufacturer to draw up a "Technical File" which defines what he has done to meet the essential requirements. However, in these cases there is no need to have the Technical File assessed by an outside body; the manufacturer need only produce this file if challenged. This is in contrast to the EMC Directive's TCF. The similarity in name between these two mechanisms has caused considerable difficulty to many companies who have confused or equated the two quite separate entities. It is, indeed, surprising that a non-safety related Directive should require such a unique and stringent assessment mechanism.
From the above catalogue of deficiencies, some are inherent in the Directive, some will be addressed over time. The most serious appear to be:
Elmac Services