History of EMI Compliance
It was 1938 when a critical moment occurred in the history of American EMC (electromagnetic compatibility) compliance testing: the FCC (Federal Communications Commission) introduced its first set of restrictions on transmitter emissions.
According to Com-Power Corporation, it was observed in 1892 that power lines could negatively impact the performance of telegraph cables. It was this observation that brought about the 1892 Law of the Telegraph in the German Empire. With this development, amongst others, the need was born for effective regulation and accurate EMC testing equipment.
As technology evolves to satisfy our needs and wants in areas such as smart grids, smart cars, and smartphones, there must also be advanced corresponding EMC testing equipment to ensure such smart devices are safe to use, from an EMI perspective.
"Let's Wait to See If It Passes"
Unfortunately, "waiting to see if it passes" is indeed an approach taken by some, according to Michel Mardiguian in his book Interference Control in Computers and Microprocessor-Based Equipment (page v). He states that "Coping with this phenomenon [EMI] is often considered to be the domain of highly specialized individuals and therefore is not taken into account during the initial design stages; instead, the resultant test data are awaited 'to see if it passes'."
If you, as an electronics designer or engineer, do not address EMI (electromagnetic interference) concerns or do not have a plan to remedy EMC testing failures, you are placing yourself in a precarious situation. Trying to band-aid a PCB so that it will pass EMI testing may be costly, it may be time-consuming, it may be both, or it may simply not work. In my personal experience as a practicing electrical engineer, many engineers—and far many more managers—simply don't understand how important it is to address EMI concerns from the very beginning of the design.
It's too easy, for them, to say "let's worry about it later". That response always makes me think of the old adage: if you don't have time to do it right the first time, when will you have time to do it over again. Simply leaving yourself the option of incorporating EMI countermeasures into a PCB, a cable, or an enclosure may save you time, money, and stress.
A Little EMI Theory
Many engineers and managers refer to EMI- and EMC-related problems and theory as "black magic". But it's not magic at all. Sure, it may be complicated and it may require some math skills, but if you understand the concepts and/or have the right people (experts) working the problem(s), you should be okay.
The source to victim concept is widely accepted in the world of EMI. When an EMI problem exists, there is always a noise source and a victim where the trouble or problem is occurring. Also, in order for a noise source to cause interference, there must be a coupling path between the source and the victim. As such, EMI can be reduced in one or more of the following areas:
- Source: Interference can be reduced at this level by means of decoupling, shielding, or simply making a less noisy design.
- Coupling path: Interference can be reduced here by spacing and/or shielding if the coupling path is radiation or by using filters if the coupling path is conductive.
- Victim: Reducing interference can be achieved by local decoupling, isolating, or shielding, or by redesigning the circuit/device such that the components are less susceptible to EMI.
Figure 1. Three basic elements of an emitting/susceptibility situation. Image courtesy of Mardiguian (page 1.2)
Noise reduction can be measured in units of dB (decibel). To evaluate the amount of noise reduction as a result of using a filter or a shield, we use the following expression:
dB = 20 log10 (VOUT / VIN)
For example, a shielding device that attenuates voltage by a factor of 10 (a ratio of volts/volts—a dimensionless number) is to say that the shielding device provides a shielding effectiveness of 20 dB. See Table 1 below for decibel values as a ratio.
Table 1. The Decibel as a Ratio
According to Mardiguian (page 1.2, 1.4), the attenuation effectiveness in terms of dB values can be grouped as follows:
- 0 to 10 dB = poor attenuation. A filter that reduces the conducted noise (or a shield reducing the EMI field) by this amount hardly pays for itself. The effect may be noticeable, but it cannot be relied upon to eliminate EMI problems.
- 10 to 30 dB = minimum range for achieving meaningful attenuation. In mild cases, EMI problems would be eliminated.
- 30 to 60 dB = range where average EMI problems can be solved.
- Over 60 dB = range for gaining above-average attenuation—requires special attention and quality in shield and/or filter mountings (surface preparation, gasketing, and bonding). Reserved for equipment which must operate at or near a 100% dependability in extreme environments.
Common EMI Countermeasures
Below is a list of common EMI countermeasures:
Metal Shielding Boxes
- Such boxes mimic little Faraday cages. Their intent is to encase, as much as reasonably possible, all the electrically noisy components.
- See Figure 6 and 9 of my Teardown Tuesday: Leeo Wi-Fi Carbon Monoxide and Smoke Alarm for two examples of shielding boxes.
- Flat ribbon cable: It is ideal, although not always feasible, to separate the digital signals with a ground connection. See Figure 2 below.
Figure 2. Flat ribbon cable EMI reduction options.
- Twisted Pair. Differential signals are twisted with each other, or a single-ended signal is twisted with a return wire. With differential signals this approach is highly effective against received common-mode noise because the differential receiver will cancel out this noise. Generated EMI is also reduced because currents traveling in opposite directions in the two wires will produce fields that balance each other.
- Shielded twisted pair: In the context of idealized differential signaling, shielding is unnecessary, but in real life the coupling between the two wires is not perfect, and the receiver's common-mode rejection is not infinite. So the shielding around the pair further reduces the effects of generated and received EMI.
- In general, an unshielded cable acts as an antenna that receives or radiates EMI. Conductive shielding, often connected to a ground node, helps to reflect and absorb EMI before it can negatively affect a circuit.
Figure 3. Shielded twisted pair wiring
- Ferrite beads (also called ferrite cores, or chokes). Ferrite beads suppress high-frequency electrical signals. When attached to a cable, they help to mitigate the effects of received EMI and reduce the amount of generated EMI. Ferrite bead kits (see Figure 4) are available for EMI testing and troubleshooting, whether it be in-field troubleshooting or testing at an EMC compliance test lab.
Figure 4. Ferrite bead kit. Image courtesy of Laird-Signal Integrity Products at Digi-Key
Final Box Design
A final enclosure should ideally behave as a Faraday cage; that is, it should provide a continuous conductive enclosure. However, this is usually impractical because enclosures need gaps or access ports for ventilation, maintenance, cabling, and user-interface components such as buttons and switches.
Therefore, when designing a final enclosure the following items should be considered:
- Keep the number and sizes of openings to a minimum. See Figure 5 and 6 below.
Figure 5. An unnecessarily large access port. A poor design
Figure 6. A better design. Enclosure openings are properly sized, and cable shields are grounded at entry
- Cover ventilation openings with conductive grids. The fineness of the grid depends on the EMI frequencies involved, with higher frequencies requiring smaller openings.
- EMI gasket material is used to seal gaps in doors, hinged sides, or panels. See Figure 7 and 8 below.
Figure 7. EMI gasket material. Image courtesy of SASIndustries (page 04)
Figure 8. Custom-made EMI gaskets. Image courtesy of SASIndustries (page 06)
EMI/EMC is not "black magic", although it can be rather complicated, especially in high-frequency systems. If you're a design engineer and don't understand EMI, make sure someone on your team does. And if no one does, consider hiring an EMI consultant at the very beginning of your design. Above all else, don't ignore EMI and "wait to see if it passes" EMC testing—this decision may prove very costly and/or time-consuming.