Quality Assurance Procedures for EMC Compliance

Ideally, every unit would be fully tested as soon as it came off the production line. But since this is clearly not an economically viable option for manufacturers, some type of quality control is necessary to ensure that identical products really are sufficiently alike as far as their EMC performance is concerned. It is important to note, though, that having a quality system (e.g., ISO 9000) in place does not in and of itself guarantee continuing EMC compliance. Rather, specific quality assurance procedures for controlling EMC compliance must be implemented.

While any number of the many small changes that can readily occur during serial manufacture can ruin EMC performance, in most cases it would be impractical to train all production staff in EMC to a level adequate to prevent them from unwittingly compromising EMC compliance. This article describes three practical and powerful quality procedures that will help maintain EMC compliance without requiring production staff to become EMC experts. Taken together, these strategies can minimize the risk of noncompliance, at minimal cost, in all areas of a manufacturing company. (It should be noted here that a similar set of procedures may also be adopted to help maintain continuing safety compliance in serial manufacture.)

ISO 9000 and other quality systems can be thought of as structural skeletons to be fleshed out by each participating company according to its unique needs and corporate character. The precise form that the quality system itself will take can thus vary widely, as determined by each concern's particular quality policies. There is, however—or at least there should be—some common ground among all such systems.

While common sense suggests that every quality system should seek to ensure that the company's legal obligations are met, too few systems actually perform this essential task. But because compliance with the EMC Directive will almost always be declared, and affixing of the CE mark undertaken, on the basis of analysis and tests done on the first few units of a product model to be manufactured, the quality system must bear the burden of ensuring continuing compliance with EMC laws.

If the legal aspect does not seem motivation enough for maintaining EMC performance in production, consider the following additional arguments:

This, then, is the balancing act the corporate quality system must perform: it must maximize market potential and improve profitability by enhancing staff awareness of commercial costs and benefits, while at the same time protecting the company itself from damaging legal claims.

Unfortunately, once again, the mere existence of a quality system (such as any part of ISO 9000) within a company does not necessarily ensure that the correct actions are being taken to maintain the EMC compliance of products in serial production. Appropriate procedures and work instructions are also needed.

The biggest obstacle most companies face when introducing EMC procedures into their quality systems is the fact that so many apparently trivial or subtle details can affect EMC performance. Some of the design staff may feel comfortable enough with EMC theory, but others may consider it a black art, and training everyone to recognize potential EMC problems would be impractical, if not impossible—they would all have to become experts in the field! Given that constraint, the question becomes, How can continuing EMC compliance be assured without excessive costs?

Full testing, as used to demonstrate the compliance of the original product design, is invariably too expensive to comprise a realistic basis for maintaining compliance. Even a pared-down, self-administered EMC test regime will be too costly if it is the only EMC action taken, because it will be able to detect problems only after the value has been added to the product.

To get a better idea of how and why the procedures described below will work, let's take a quick look at some of the things that can go wrong on the production line.

To appreciate the need for EMC awareness and control, and to comprehend the value of the three procedures presented in this article, we first need to understand just how easy it is for EMC to be compromised in serial production. Here are just a few potential problem areas:

Capacitors. Capacitors of equal capacitance value, tolerance, and voltage rating may not necessarily be equivalent for EMC purposes. For example, fitting 10 nF polyesters in place of 10 nF ceramics, either to save money or because of a parts shortage, can have a disastrous effect on EMC compliance, as shown in Figure 1.

Figure 1. Different types of capacitors and their assembly can have significant effects on EMC (graphs are typical; contact suppliers for correct data).

Assembled Lead Lengths. Excess lead length adds inductance, which can disrupt EMC. For instance, an additional 12 mm of component height above a PCB (e.g., through careless assembly) can result in the following:

Transient absorbers clamping overvoltages 80 V higher on IEC 801-4 fast-transient tests, and 500 V higher on IEC 801-2 electrostatic-discharge tests, causing damage to semiconductors.

A falloff in the self-resonant frequency of a 10 nF decoupling capacitor from 19 to 11 MHz, and a consequent reduction in its suppression of microprocessor emissions (see Figure 1).

A sharp decrease in the effectiveness of a filter above 20 MHz.

Filters. Like capacitors, filters can be similarly specified without being equivalent for EMC. Filters are measured in 50- systems, even though real systems are rarely 50 ; in practice, filter performance can be up to 100 times worse than expected, and identically specified filters can behave quite differently from each other in actual applications. It is impossible to specify filters for EMC using manufacturer-supplied data.

Cable Routing. The proper layout of wiring and cable in a product can be an essential element of EMC compliance. Simplifying a wiring scheme to speed up production or to save tie-wraps can adversely affect EMC performance. (Figure 2 illustrates what can happen if an assembly person tries to save company time and money by "fixing" the wiring layout of a filter.)

Figure 2. Simplifying a wiring scheme to save time can adversely effect EMC performance.

Threadlock Compounds. Almost all threadlocking compounds destroy electrical connections by increasing their contact resistance. Where mechanical connections are important for EMC (e.g., in mounting a connector body), the use of such compounds can preclude EMC compliance.

Process Changes. Variations in the amount of paint overspray applied; surface passivation; the use of colored alochrome instead of plain; and switching from alochroming to anodizing to improve scratch resistance can all affect EMC by preventing good electrical bonds between parts.

Purchased Subassemblies. The EMC performance of dc/dc convertors, display modules, single-board computers, switch-mode power supplies, and other such purchased subassemblies will often be crucial to the EMC of the product as a whole. Subassembly manufacturers are liable to change their designs (along with their EMC performance, hence the EMC of the final product) at any time.

Software. Changes in software can have a significant effect on both emissions and immunity. Even a bug-fixing upgrade can render a product noncompliant for EMC, let alone a major upgrade that enhances performance or capabilities.

Integrated Circuits (ICs). Equivalent ICs from different manufacturers, or different versions of the same IC from a single manufacturer, can exhibit considerable differences in electromagnetic performance. Even within the same IC type from the same manufacturer, emissions and immunity can vary considerably from batch to batch, creating EMC inconsistencies. The EMC characteristics of an IC can also be dramatically altered if the manufacturer does a "mask shrink," even though the new device may still carry the same part number as previous versions.

Applied correctly, these procedures will enable every employee in a company to know immediately and with certainty—and without having to be an EMC specialist—whether or not the part, method, or process he or she is involved with is critical to EMC compliance. This will allow the employee to ensure that any change that may affect EMC compliance, no matter how small or how subtle, is approved by an EMC expert within the company.

Without prior analysis and identification of all EMC-critical aspects of a product's manufacture, the only way to gauge whether compliance is likely to be affected is to have EMC experts vet every single design change, production concession, and drawing iteration. In most cases, this approach is an unworkable one, as it slows down vital procedures and increases costs even as it leaves open the possibility that unnoticed problems (such as changes in wiring routes, or the use of threadlock) may slip through.

EMC-critical identification should take the form of a mark—for example, (E)—applied to the basic identifier, whether it be the part number, the drawing number, or the document number. The mark should also be applied on assembly drawings to indicate those assembly details which are EMC-critical (such as the mounting screws for a connector). This way, if a buyer later picks up an urgent production concession to find an alternative for, say, a capacitor that is out of stock, and finds that the part number has an (E) on it, he or she will know that the company's EMC specialists must be involved in the choice of an equivalent.

The third proposed process, the appropriate use of in-house EMC testing, need not be burdensome if it is carried out sensibly. In both ongoing manufacture and product development, a strong argument can be made for its being more efficient and more cost-effective in the long term than outside test-house testing.

All three of the procedures detailed below are best applied as part of an overall product-development program, since a company's design engineers are the personnel best placed to decide precisely what is EMC-critical and what is not, and to determine the type and amount of in-house testing that will be required to provide for continuing compliance. Existing products will need to be treated retrospectively.

With these procedures in place, the extent of staff training required to ensure that EMC is controlled as completely and as economically as possible will be minimized. EMC will rapidly become part of the corporate culture, maximizing the benefit to the company.

The EMC-criticality exercise is much like the failure-mode analysis required by the EU product-safety directives. Whether a company is registered to ISO 9000 or any other quality-system standard, or whether it has no quality system whatsoever in place, EMC-criticality analysis will serve as a valuable tool for ensuring continuing EMC compliance at reasonable cost.

The determination of what is or is not EMC-critical is usually best left to the members of a company's design team, who should be up to the task by virtue of having to create compliant products quickly and at low cost. The first few times an EMC-criticality exercise is undertaken, it may be expected to proceed rather slowly, as personnel familiarize themselves both with the procedure itself and perhaps with EMC issues in general. During these initial analyses, detailed EMC specifications for common parts (such as decoupling capacitors) will have to be established, and decisions made about usual assembly practices (such as the attachment of connectors to cable screens). The same or similar specifications and practices may then be reused for later projects.

No aspect of design, manufacture, installation, or service should be overlooked in the criticality exercise; everything should be investigated for its EMC implications, and everything found to be important for the maintenance of EMC performance should be controlled using the two quality procedures explained below.

At this stage, it may be helpful for companies to enlist third parties, such as external EMC experts, to give them a sort of "jump-start" by assisting them in identifying the most cost-effective procedures and the most suitable components, methods, and processes. The participation of outside EMC experts can also be useful for formal design reviews: ISO 9000 recommends that such reviews be carried out by persons other than those who did the actual work, which for companies with small in-house EMC staffs may be impossible without third-party involvement.

Once every EMC-critical aspect of a product—from components, through assembly methods and processes, to tests, and everything in between—has been identified (as described above), everything affected needs to be marked clearly and unambiguously, in an appropriate manner.

The mark used to indicate that an item is EMC-critical must appear at the very top level of identification—for example, in the part number of a component. This will ensure that its EMC-criticality cannot be overlooked, even under extreme conditions, by anyone with even the most basic EMC training. Many companies have chosen to adopt an (E) symbol as their EMC-critical mark (sometimes in conjunction with an [S] symbol to indicate safety-criticality) because it is part of the ASCII character set and is unlikely to have any other meaning.

The (E) symbol (or whatever other mark is selected) may need to be added only to the details of parts lists and other paperwork rather than to the actual document numbers, since any EMC-critical parts included in such lists will already carry the symbol in their part numbers. Similarly, while assembly drawings should use (E) marks to clearly identify every detail that has an EMC implication, it may be unnecessary to include the (E) in the drawings' own identification numbers.

Production, installation, and service personnel must all be trained to recognize the EMC symbol and to realize that for those items they will need authority from the company EMC expert before deviating from the specification or drawing in any way whatsoever. It should go without saying that a change-control procedure will also be required, and that any proposed changes to EMC-critical items or techniques must be approved by the on-staff EMC expert. This expert may determine that the change is insignificant given the EMC pass margins achieved and the relative criticality of the product, or he or she may decide that it warrants either partial in-house testing (maybe with a close-field probe) or full EMC testing of a prototype—whatever seems most certain to maintain EMC compliance in serial production.

The EMC training required for production staff is minimal using this approach, and the quality control procedures involved will be easy for anyone on the production team to understand and put into practice. Adherence to such procedures may even be written into job descriptions and/or terms and conditions of employment (a particularly prudent provision given that EMC compliance is now a mandatory legal requirement in the EU).

If correctly implemented, the procedures described above, EMC-criticality analysis and marking, should eliminate the need for much expensive EMC testing while at the same time reducing the risk of noncompliance. Some ongoing EMC testing will still be necessary, however, if only to detect such unforeseeable problems as bad batches of ICs (preferably catching them early in the assembly process), to evaluate the effects of minor changes (e.g., fixing a software bug), or to check that variants of standard products remain EMC-compliant.

If the only EMC testing facility available is a third-party test laboratory some distance away, a company's in-house staff may be tempted to let most small changes go through without rechecking the product's compliance, instead booking time at the test house for every, say, fifth change. Over the long term, this strategy is an unwise one: it will tend to increase the risk of noncompliance, and any money saved by not buying in-house EMC test equipment will eventually be spent on outside test-house fees.

Besides its usefulness for troubleshooting and informal verification, in-house testing can be a real boon for product development, speeding time-to-market, and enhancing cost-effectiveness. Getting to market faster in turn will have a profoundly beneficial impact on break-even time (the point in a project when total expenditure on a new product equals total revenue). Companies wishing to invest in in-house facilities will find a wide range of EMC test equipment to choose from; alternatively, on-site equipment tailored to specific products can be built at surprisingly low cost. So-called golden product comparisons comprise a powerful low-cost testing technique, in many cases requiring nothing more than confidence that nothing untoward has happened during testing.

Properly calculated cost-benefit financial analyses will establish what a company's optimum expenditure on EMC equipment may be; when the math is done, the amount of money that may sensibly be earmarked for in-house test equipment may be a good deal more than expected, even when a payback time of only one or two years is figured in. Too often, however, product designers and financial officers cannot speak each other's language or understand each other's constraints or needs, so projects that would benefit everyone never get under way. In this instance, building the bridge from both sides should pay handsome dividends.

Product EMC compliance can be a fragile thing, vulnerable to the smallest and most commonplace changes in electronic components, assembly, and manufacturing. Without quality control procedures to govern EMC-critical parts and procedures, a pass result on a full EMC compliance test proves only that the unit tested is compliant, and in no way guarantees the compliance of units manufactured subsequently. The procedures set out above can be implemented in any manufacturing environment, in the presence or absence of a recognized quality system, to ensure consistency and ongoing product compliance, with minimal expenditure for training and testing and maximum protection against legal action.

Back to 1999 Annual Reference Guide Table of Contents