Gert Gremmen, Tim Haynes, Ralph McDiarmid, Ed Price, John Woodgate
EMC performance of a product is likely to vary with age as the physical characteristics change, e.g. caps dry out and metal junctions oxidize, etc. Obviously, the product is designed with the intention of consistent compliance over the life of the product, but are there any requirements or guidance relating to preventing or controlling this change in EMC performance over time?
This question posed recently on a product safety forum garnered some interesting responses from EMC and product safety experts. Interference Technology delved a little deeper and asked a panel of experts their thoughts on the topic:
Can environmental factors that jeopardize EMC over a product’s lifetime result in performance, reliability and even safety implications?
GREMMEN: Yes, but it will need a risk analysis in the design phase to identify the risks associated with aging. That risk analysis should include the whole spectrum of aging aspects, from components, to soldering techniques, contact properties related to oxidation, moisture, vibration, but also need to incorporate equipment properties that are not expected to “age”, such as software.
The analysis also needs to consider changing properties that at first sight do not impact EMC properties, such as enclosures of which radiation patterns and resonance properties may change and have a bigger impact on EMI as initially assumed.
Ultimately, it is impossible to maintain EMC properties over a long time, and therefore a definition of the lifetime of a piece of equipment is required. Very little work has been done on the calculation of lifetime for electronics in general, but a lot of data on failure rates is available, for example, in soldered contacts, and individual components. I have been involved in developing a simple method of creating a lifetime expectation based on collected statistical data for component lifetimes, but that does not include many relevant aspects, such as enclosure lifetime under miscellaneous (hards) conditions.
PRICE: In the military product area, most products don’t live long enough for aging to be as important as the (usually) very hostile physical environment. One further specific instance is damage to cable / backshell junctions due to abuse by users and/or vibration and shock.
HAYNES: Expanding only slightly, corrosion can reduce screening at interfaces of boxes, cables, connectors etc., gaskets can deteriorate, components can age and affect EMI performance.
Can lifetime reliability be better assured by simulating product aging before EMC testing is done? How can lifetime EMC integrity be assured when equipment is still generally tested only once?
WOODGATE: It depends on how realistic the simulation is, given that no simulation can be wholly realistic.
[Lifetime EMC integrity can be assured] by careful selection of components, operating them well below their maximum stresses and careful overall design (e.g., assure continued integrity of seams in enclosures).
HAYNES: Some requirements (possibly customer requirements in defence/aerospace) need environmental testing (heat cycling / vibration) to be completed before EMC tests are undertaken. No “repairs” are allowed between environmental and EMC testing.
PRICE: In a Qualification test scenario, it is our practice to run the temp cycling, vibration, humidity types of tests before the system goes to EMC. True, some tests (high-G shock, blowing rain, transportation shock, high-G centrifuge) are often done on a separate system.
GREMMEN: I am not aware of simulation software with a view on EMC related to aging. I think that in order to incorporate aging as a factor in EMC, we need to lend a ear to environmental testing specialists, and start collecting events, and try to couple this to EMC properties. A knowledge base may be created that allows designers to identify potential risks.
MCDIARMID: In my brief experience in private industry (since 1983) I can recall only one legitimate complaint concerning radio interference from a product. It’s my experience that EMC problems during operational service of a product designed and verified through type testing to comply with the applicable standards is a rare thing, even over its operational lifetime. It may be that most products functionally fail and are removed from service before they become an EMC nuisance.
Many companies control the configuration of their products by keeping a list of EMC critical components, much like a list of safety critical components for use in UL and CSA compliance. If a component, like a power MOSFET, needs to change manufacturer, then all the specifications are carefully checked and an EMC retest (at least for emissions) may be performed before qualifying the alternate source. If a product complies with the applicable RF emission and immunity standards, it will likely never be a problem during its lifetime.
Are there any regulatory requirements to artificially age a product prior to EMC testing? If so, are the requirements industry specific?
WOODGATE: Not for commercial products in Europe, as far as I am aware.
Direct controls of performance are not appropriate for safety-critical applications. Each product has its requirements set by following a procedure, which is described in the multi-part standard IEC 61508.
HAYNES: There may be some implicit requirements, such as in section 6 of the UK Health and Safety at Work Etc. Act 1974 – Duties Towards Articles Used At Work. There may be other such requirements in legislation (possibly in the Provision and Use of Work Equipment?). There may also be requirements in Joint Airworthiness Requirements (JARs) that require a ground-based “golden sample” aircraft to be exposed to more hours of simulated flying than the worst-case flight-cleared aircraft of the type. This is to identify fatigue failures in the ground-based model before they can happen in the flight-cleared aircraft.
PRICE: USA military procurements are controlled by a strict and extremely specific list of contractually obligating “Line Items”, so this is all defined by the contract.
GREMMEN: There are no such regulations I am aware of. At least there are no such for ordinary commercial and residential equipment, not even for high-end laboratory equipment.
Can a well-defined maintenance schedule help ensure regular surveillance of components that will affect the EMC performance of a product?
WOODGATE: Possibly: Some EMC characteristics can be checked simply enough for inclusion in a maintenance schedule (e.g. low-frequency conducted emissions), while others cannot (e.g. high-frequency immunity). Some might think that ESD immunity could be checked in a maintenance process, but this raises safety issues in itself.
HAYNES: Yes – but only where there is a conscientious maintainer. This may be much more important in respect of vehicle maintenance (screening of the engine management unit, braking control, etc.), than it would be to have annual maintenance on your TV or washing machine. However, how many garage mechanics would be trained in “EMC”?
PRICE: Military systems have several levels of maintenance, with the first line generally not going beyond “pull that box that doesn’t work and put in a new one.” First level people are usually not opening boxes, fixing cables or doing other intrusive things. At the depot level, a manufacturer should specify if EMC items should be replaced as part of maintenance (perhaps a particular EMI gasket is good for only a few open/close cycles).
GREMMEN: Applications that require tight control of EMC properties during a defined lifetime definitely need adequate service. Maintenance has traditionally been focused on functional parameters only, even in aerospace and aviation technology. Many EMC-related components can get defective without even being noticed in functional testing, so additional testing has to be defined in order to recognize potential failures of this kind. While emission properties may be quickly established using, for example, a tabletop spectrum analyzer on a comparative basis (compared to a new piece of equipment), immunity problems will remain unnoticed until a full test suite for immunity has been carried out. As many EMC components have a function towards common mode phenomena, traditional test techniques that use differential mode signals are not capable of detecting failures in such components. A surge suppressor, for example, may be connected to the enclosure and not to system signal ground, and may look connected to nothing from the point of view of traditional automated test sets. In addition to that, commercial test setups for components operate on PCB level, and not on equipment or even system level.
Even with regulatory and contractual compliance established at the outset, could changes be made to a product that may compromise its EMC performance?
WOODGATE: If you mean ‘changes during life’, the answer must be ‘yes’. [This can be avoided by] total encapsulation, making the product unrepairable. But then how can the heat be dissipated? Heat pipes? Such ‘heroic’ measures are hardly justified.
GREMMEN: Many EMC properties of equipment are undefined in the design phase. If a problem during initial testing is found, remedial measures such as filters may help controlling emissions. However, the source of the problems is not under control. A change of manufacturer for a microprocessor, for example, may change emission levels to a high degree, even if the component is pin and specification compatible. The EMC properties of such a component are in general not specified in a datasheet. A newer process on chip level, or even change of location of manufacture may impact those properties. The manufacturer is not bound to notice their customers, as EMC properties go undocumented. I lately witness a large number of SMPS related problems in EMI where just the ongoing progress in FET developments results in much faster switching times as before, creating substantial interference problems in the 30-200 MHz range. The manufacturers of the switchers, confronted with the problem, told us that they had updated their switching FETS as older (tested) models quickly became obsolete and the FET manufacturer provided them with equivalents and pin compatible models with a faster switching time. Even software can have an impact on EMC; simple changes in firmware (updates) can create sudden changes in emissions or immunity behavior that are completely outside the view of the software designers. Given the fact that modern software development heavily depends on external modules and libraries, such changes may even be unnoticed to the manufacturer of the product. Again, EMC properties of the building blocks of electronic are most of the time undocumented.
And yes, during testing it is too late. Too many definite choices have been made at that point already.
HAYNES: A rack of equipment installed in an EMC cabinet passes the required tests with the doors closed. The customer can use the equipment with the doors closed, but they are always open because he wants to see the flashing lights and alpha/numeric displays to ensure the equipment is working …
To limit customer-based changes – design products that are actually fit-for-purpose – or should that be fit-for-the-users-purpose. Otherwise implement measures to prevent EMC features being “overridden”.
PRICE: Keep close to your customer so they can accidentally show you all the new and exciting things they do with your product. You will learn the most interesting things.
Is there a benefit to considering environmental and EMC requirements at the outset of testing a product?
WOODGATE: It’s far too LATE at the onset of testing. These requirements must be applied at the outset of DESIGN, where they pay a dividend of around 1000%.
HAYNES: There is much greater benefit to be obtained by considering ALL environment and EMC requirements that will apply during the product lifecycle at the very beginning – at the concept stage of the product/project.
PRICE: I think everybody says it’s good to plan ahead.
GREMMEN: Yes, if EMC is recognized to have a substantial impact on reliability and safety, taking in consideration all these aspects will create benefits for both manufacturers as well as consumers.
Do you have any anecdotal examples that illustrate the above points or would be helpful to fellow engineers?
HAYNES: Some radio equipment was EMC tested after vibration and thermal testing had been successfully completed. The equipment failed EMC on emissions and immunity. The cause was “cracked” semi-rigid coax / connector joints where either vibration or thermal expansion had caused the connector to cable-sheath interface to crack, causing the transfer impedance to increase and allow signals into / out of the coax cable. If the equipment had not been vibration / thermal tested before the EMC – this failure would not have been found.
PRICE: Same scenario; this time finding connector nuts that were improperly torqued and allowed the connectors to loosen.
Gert Gremmen is a senior test engineer in EMC and product safety for electronic and electrical products, director of ce-test qualified testing bv in the Netherlands and an expert in CE marking.
Tim Haynes, electromagnetic engineering specialist at SELEX S&AS, has worked on space systems, avionics, marine and submarine equipment for the defense industry and has employed forensic techniques in resolving problems, in particular EMC issues during design and development of hardware.
Ralph McDiarmid, AScT, has 12 years experience in power electronics circuit design and simulation – converers and inverters to 3kW – and 10 years experience with regulatory international product approvals, including CSA, UL and CE.
Ed Price, a NARTE certified EMC engineer, has worked in the Electromagnetic Compatibility Lab at Cubic Defense Applications in San Diego, Calif. since 1993.
John Woodgate of J.M. Woodgate and Associates has a background in the consumer products and sound reinforcement sectors of the electronics industry, as well as product management and marketing of audio, high fidelity and video products. He became an independent consultant in 1984.
NOTE: Participants’ comments do not necessarily reflect the views of their employers.