11.1 Electromagnetic interference (EMI)

Sources of EMI

Electromagnetic interference (EMI) is the disruption of normal electronic device operation caused by electromagnetic fields from other sources. EMI can be triggered by both natural phenomena and man-made devices, and its effects range from minor signal degradation to complete system failure. Understanding where EMI comes from, how it couples into circuits, and how to measure and mitigate it is essential for designing reliable, compliant systems.

Natural vs Man-Made Sources

Natural sources of EMI include lightning strikes, solar flares, and cosmic radiation. These can generate powerful electromagnetic pulses (EMPs) capable of disrupting electronic systems over wide areas.

Man-made sources are far more common and include power lines, motors, switches, digital circuits, and wireless communication devices (cell phones, Wi-Fi routers). In practice, man-made EMI is what you'll spend most of your time dealing with in design work.

Common EMI Sources

Some of the most frequently encountered EMI sources in electronic systems:

• Switching power supplies generate high-frequency noise due to their rapid switching action. The fast voltage transitions create broadband spectral content that can couple into nearby circuits.
• Digital circuits produce high-frequency harmonics because their square-wave signals contain energy at odd multiples of the fundamental clock frequency.
• Electric motors and transformers can generate both low-frequency EMI (from 50/60 Hz power) and high-frequency EMI (from commutation arcing or core saturation).
• Other common sources include fluorescent lights, microwave ovens, and medical equipment (MRI machines, electrosurgical units).

High-Frequency Sources

High-frequency EMI sources are particularly problematic because shorter wavelengths couple efficiently into PCB traces and cables that act as unintentional antennas. Examples include:

• Wireless communication devices operating in the GHz range (cell phones, Wi-Fi, Bluetooth)
• Radar systems, which use high-power microwave signals for detection and ranging
• Microwave ovens, which operate near 2.45 GHz and can leak energy through imperfect door seals

Transient Sources

Transient EMI sources generate short-duration, high-energy pulses that can cause severe damage to electronic components. These are distinct from continuous EMI because their peak amplitudes can be orders of magnitude higher than steady-state interference.

• Electrostatic discharge (ESD) events can generate voltage spikes of several kilovolts with sub-nanosecond rise times.
• Lightning strikes can induce large currents and voltages in nearby conductors through both direct coupling and ground potential rise.
• Switching of inductive loads (motors, relays, solenoids) generates voltage spikes due to the sudden collapse of stored magnetic energy. The spike amplitude is governed by $V = -L \frac{dI}{dt}$, so faster current interruption produces larger transients.

Coupling Mechanisms

EMI reaches a victim circuit through specific physical mechanisms. These fall into two broad categories: conducted and radiated coupling. Identifying the dominant coupling path is the first step toward effective mitigation.

Conducted EMI

Conducted EMI travels through physical connections such as power lines, signal cables, or ground planes. Common conducted coupling paths include:

• Power line disturbances: voltage fluctuations, transients, and harmonics riding on the AC mains
• Ground loops: circulating currents caused by multiple ground connections at different potentials, creating voltage differences across the ground network
• Crosstalk between adjacent signal lines: unwanted coupling through shared impedance paths on a PCB or in a cable bundle

Radiated EMI

Radiated EMI propagates through space as electromagnetic fields. Any conductor carrying time-varying current acts as an antenna to some degree. Common radiated EMI sources include intentional antennas, high-frequency digital circuits, and poorly shielded devices.

Radiated coupling is further divided into near-field and far-field regimes, which behave very differently and require different analysis approaches.

Near-Field vs Far-Field Coupling

The boundary between near-field and far-field regions occurs at approximately $r = \frac{\lambda}{2\pi}$, where $\lambda$ is the wavelength of the interference signal.

• Near-field coupling ($r \ll \lambda$): The electric and magnetic fields are not yet coupled into a propagating wave. Depending on the source impedance, either the electric field dominates (capacitive coupling) or the magnetic field dominates (inductive coupling). Mitigation is often straightforward because the fields decay rapidly with distance ($1/r^2$ or $1/r^3$).
• Far-field coupling ($r \gg \lambda$): The electric and magnetic fields are orthogonal and propagate together as plane waves, decaying as $1/r$. Far-field EMI can travel long distances and penetrate obstacles, making it harder to mitigate.

Common-Mode vs Differential-Mode Coupling

This distinction matters because each mode requires a different filtering strategy.

• Common-mode (CM) coupling: The interference signal appears with the same polarity on both conductors relative to ground. CM noise is often caused by ground loops or differences in ground potential. A common-mode choke is the standard countermeasure.
• Differential-mode (DM) coupling: The interference signal appears with opposite polarity on the two conductors. DM noise is typically caused by inductive or capacitive coupling between adjacent signal lines. Standard LC low-pass filters are effective against DM noise.

Misidentifying the mode is a common mistake. If you apply a differential-mode filter to a common-mode problem, you'll get little to no attenuation.

EMI Effects on Circuits

EMI can produce a range of detrimental effects, from subtle performance degradation to permanent hardware damage.

Signal Integrity Issues

EMI introduces unwanted noise and distortion into desired signals. Common signal integrity problems include:

• Jitter: variation in the timing of digital signal edges, which can cause setup/hold violations in synchronous systems
• Bit errors: occur when noise amplitude is large enough to push a signal across the decision threshold, causing incorrect data interpretation
• Analog signal distortion: can manifest as harmonic distortion, intermodulation products, or increased phase noise in oscillators and PLLs

Electromagnetic Compatibility (EMC)

EMC is the ability of a device to function correctly in its intended electromagnetic environment without causing unacceptable interference to other devices. EMC has two complementary aspects:

• Emissions: the electromagnetic energy a device generates that could interfere with other equipment
• Immunity (susceptibility): the device's ability to operate correctly when exposed to external EMI

Regulatory bodies such as the FCC (United States) and standards organizations like the IEC (CISPR series) define emissions limits and immunity requirements. A device must satisfy both to be considered electromagnetically compatible.

Crosstalk in PCBs

Crosstalk is the unwanted coupling of signals between adjacent traces on a PCB. Both capacitive coupling (through the electric field between traces) and inductive coupling (through the mutual inductance of parallel traces) contribute.

Factors that affect crosstalk severity:

• Trace spacing and geometry: closer traces produce stronger coupling
• Signal rise/fall times: faster edges have higher-frequency content that couples more efficiently
• Dielectric properties: the PCB substrate's permittivity affects the capacitive coupling coefficient

Mitigation techniques include increasing trace separation, inserting grounded guard traces, using stripline (embedded between ground planes) rather than microstrip, and controlling trace impedances.

EMI-Induced Noise

EMI can induce unwanted noise voltages and currents in circuits. While thermal noise, shot noise, and flicker noise are intrinsic to components rather than caused by external EMI, they set the noise floor that determines how vulnerable a circuit is to external interference. The lower your intrinsic noise floor, the more headroom you have before EMI causes problems.

• Thermal noise: caused by random electron motion in conductors; power spectral density is $S_v = 4k_BTR$ (where $k_B$ is Boltzmann's constant, $T$ is temperature, $R$ is resistance)
• Shot noise: arises from the discrete nature of charge carriers crossing potential barriers (p-n junctions, Schottky barriers)
• Flicker noise (1/f noise): power spectral density inversely proportional to frequency; dominant at low frequencies in active devices like transistors and op-amps

Practical mitigation includes using low-noise components, minimizing loop areas (to reduce magnetic flux pickup), and implementing proper grounding and shielding.

EMI-Induced Component Damage

In severe cases, EMI can permanently damage components through direct overstress or secondary effects like thermal runaway.

• ESD damage: dielectric breakdown, junction damage, or localized thermal damage in semiconductors. Even sub-threshold ESD events can cause latent damage that degrades reliability over time.
• Electrical overstress (EOS): excessive current or voltage causes melting, vaporization, or thermal decomposition of materials.
• Latchup: in CMOS ICs, a parasitic thyristor (PNPN structure formed by the well/substrate junctions) can be triggered by transient currents, creating a low-impedance path from $V_{DD}$ to ground. Without current limiting, the resulting power dissipation can destroy the device.

Protective measures include transient voltage suppressors (TVS diodes), ESD protection diodes, current-limiting fuses, and proper PCB layout with adequate guard rings around sensitive CMOS inputs.

EMI Measurement Techniques

Accurate EMI measurement is essential for verifying electromagnetic compatibility and ensuring compliance with regulatory standards.

Spectrum Analyzers

Spectrum analyzers display signal amplitude as a function of frequency, making them the workhorse instrument for EMI characterization. Key specifications to understand:

• Frequency range and resolution bandwidth (RBW): RBW determines the minimum frequency separation you can resolve. Narrower RBW gives better resolution but increases sweep time.
• Amplitude accuracy and dynamic range: dynamic range sets the ratio between the largest and smallest signals you can measure simultaneously.
• Sweep time and real-time bandwidth: real-time analyzers can capture transient events that a swept analyzer would miss.

Spectrum analyzers can be paired with various antennas and probes to measure both conducted and radiated EMI.

EMI Receivers

EMI receivers are purpose-built instruments for emissions testing. They combine the functions of a spectrum analyzer, a preamplifier, and standardized detectors (peak, quasi-peak, and average).

EMI receivers conform to measurement standards like CISPR 16-1-1, which specifies measurement bandwidths, detector types, and dwell times. The quasi-peak detector is particularly important: it weights signals based on their repetition rate, so intermittent interference is penalized less than continuous interference. This reflects the subjective annoyance level of different interference types.

EMI receivers typically offer higher sensitivity and dynamic range than general-purpose spectrum analyzers.

Near-Field Probes

Near-field probes measure localized electric and magnetic fields close to the surface of a device under test (DUT). They're invaluable for pinpointing EMI sources during troubleshooting.

• H-field probes: small loops that respond to the magnetic field component. Useful for finding current-carrying traces and loops that radiate.
• E-field probes: monopole or dipole elements that respond to the electric field component. Useful for finding high-voltage nodes and capacitive coupling paths.

These probes are used with a spectrum analyzer or oscilloscope to visualize EMI in both frequency and time domains. Because they measure near-field quantities, the readings don't directly correspond to far-field emissions, but they're excellent for comparative diagnostics.

Antennas for EMI Measurement

Far-field radiated EMI is measured using calibrated antennas. The choice of antenna depends on the frequency range:

Each antenna is calibrated to determine its antenna factor (AF), which relates the received voltage to the incident electric field strength: $E = AF \times V_{received}$ (in linear units), or equivalently $E_{dB\mu V/m} = V_{dB\mu V} + AF_{dB}$.

Conducted EMI Measurement

Conducted EMI is measured using a line impedance stabilization network (LISN). The LISN serves two purposes:

1. It presents a defined, standardized impedance (typically 50 Ω) to the DUT across the measurement frequency range.
2. It isolates the measurement from noise on the external power supply.

The LISN is connected between the power source and the DUT. The EMI voltage appearing across the LISN's measurement port is then read by an EMI receiver or spectrum analyzer. Standard conducted emissions tests typically cover 150 kHz to 30 MHz.

Radiated EMI Measurement

Radiated EMI measurements characterize the electromagnetic fields emitted by a device in the far-field region. The standard procedure:

1. Place the DUT on a non-conductive table in an anechoic chamber or open-area test site (OATS) to minimize reflections and ambient interference.
2. Position the receiving antenna at a specified distance (typically 3 m or 10 m) from the DUT.
3. Rotate the DUT through 360° to find the orientation of maximum emission.
4. Scan the antenna height (1 m to 4 m for OATS) to account for ground reflections.
5. Sweep the frequency range of interest (typically 30 MHz to 1 GHz or higher) and record the maximum emission levels.
6. Compare results against the applicable emissions limits.

EMI Shielding

EMI shielding encloses electronic devices or circuits in conductive materials to reduce both the emission and reception of electromagnetic interference.

Shielding Effectiveness

Shielding effectiveness (SE) quantifies how well a shield attenuates electromagnetic fields. It's defined as:

$SE_{dB} = 20 \log_{10}\left(\frac{E_{incident}}{E_{transmitted}}\right)$

for electric fields (and similarly for magnetic fields or power). SE depends on three loss mechanisms:

• Absorption loss: energy absorbed as the wave travels through the shield material. Increases with material thickness, conductivity, permeability, and frequency.
• Reflection loss: energy reflected at the shield boundaries due to impedance mismatch between free space and the conductor. Dominant at lower frequencies for electric fields.
• Multiple reflection loss: additional reflections between the two surfaces of the shield. Significant only when the shield is thin relative to the skin depth.

The total SE is approximately: $SE = A + R + M$ where $A$ is absorption loss, $R$ is reflection loss, and $M$ is the multiple reflection correction (negative when the shield is electrically thin).

Shielding Materials

The choice of shielding material depends on the application, frequency range, weight constraints, and cost.

• Metals (copper, aluminum, steel): provide excellent SE due to high conductivity. Copper and aluminum are preferred for electric field shielding; steel and mu-metal (high permeability) are better for low-frequency magnetic field shielding.
• Conductive polymers: composites of polymers with conductive fillers (carbon fiber, metal particles, nickel-coated fibers). Lighter and more flexible than solid metals, but generally lower SE.
• Conductive coatings: paints, sprays, or electroless plating applied to non-conductive enclosures. Cost-effective for moderate shielding requirements.

Enclosure Design for EMI Shielding

A shield is only as good as its weakest point. Key design considerations:

• Seam and joint design: every seam is a potential leakage path. Overlapping joints with continuous electrical contact outperform butt joints. Conductive gaskets or spring fingers maintain contact across seams.
• Aperture control: any opening (ventilation holes, display windows, connector cutouts) degrades SE. The critical dimension is the longest linear dimension of the aperture relative to the wavelength. An aperture becomes a significant leakage source when its longest dimension approaches $\lambda/2$.
• Waveguide-below-cutoff: for necessary ventilation, arrays of small holes (honeycomb panels) act as waveguides below cutoff, providing high attenuation while allowing airflow.
• Grounding: the enclosure must be properly bonded to the system ground to provide a low-impedance return path for shield currents.

Cable Shielding Techniques

Cables act as antennas that can both pick up and radiate EMI. Common shielding approaches:

• Braided shields: woven from conductive wire strands. Good flexibility and mechanical durability, but optical coverage is typically 60-95%, leaving small gaps.
• Foil shields: thin aluminum or copper foil bonded to a polyester carrier. Provide 100% optical coverage and high SE, but are fragile and less flexible.
• Combination shields: use both braid and foil layers to combine high coverage with mechanical robustness. Common in high-performance cables.

Proper shield termination is just as important as the shield itself. A shield that's grounded through a long pigtail wire introduces inductance that degrades high-frequency performance. 360° backshell terminations are preferred.

Apertures and Seams in Shielding

Apertures and seams are the dominant leakage paths in most practical shielding enclosures. A single poorly sealed seam can reduce the SE of an otherwise excellent enclosure by 20-40 dB.

Techniques for minimizing leakage:

• Use conductive gaskets (beryllium copper, knitted wire mesh, conductive elastomers) to maintain electrical continuity across seams
• Minimize the maximum linear dimension of any aperture
• Use arrays of small holes instead of a single large opening (many small holes leak less than one large hole of the same total area)
• Apply waveguide-below-cutoff principles: a circular aperture of diameter $d$ attenuates signals below the cutoff frequency $f_c \approx \frac{1.76c}{d}$ (where $c$ is the speed of light)
• Ensure proper bonding and grounding to provide low-impedance paths for EMI currents on the shield surface

EMI Filtering

EMI filtering uses circuit elements to attenuate electromagnetic noise while passing the desired signal. Filter design must account for the noise mode (common-mode vs differential-mode), frequency range, and source/load impedances.

Passive vs Active Filters

Passive EMI filters use resistors, capacitors, and inductors to attenuate unwanted frequencies.

• Simpler, more reliable, and less expensive
• No power supply required
• Effective at high frequencies where active components have limited bandwidth
• Limited attenuation per stage; cascading stages adds insertion loss and size

Active EMI filters incorporate op-amps or transistors alongside passive components.

• Can achieve higher attenuation and sharper rolloff, especially at lower frequencies
• Allow tunable or adaptive filtering
• More complex, more expensive, and require a power supply
• Active component bandwidth and noise floor limit high-frequency performance

For most power-line EMI filtering, passive filters are the standard choice. Active filters find use in specialized applications where passive solutions are too large or insufficiently selective.

Common-Mode Filters

Common-mode filters attenuate noise that appears with the same polarity on both conductors relative to ground. The core component is a common-mode choke: a transformer wound so that differential-mode currents produce opposing magnetic fluxes (and thus see low impedance), while common-mode currents produce additive fluxes (and see high impedance).

Y-capacitors (connected from each line to ground) are often added to provide a low-impedance path for high-frequency common-mode noise to return to ground.

Differential-Mode Filters

Differential-mode filters attenuate noise that appears between the two signal conductors with opposite polarity. These are standard low-pass filter topologies using:

• Series inductors to block high-frequency differential noise
• X-capacitors (connected across the line pair) to shunt high-frequency noise

A typical single-stage differential-mode filter is an LC low-pass section. Multiple stages can be cascaded for steeper rolloff. The filter's cutoff frequency and impedance must be matched to the source and load impedances for effective attenuation; a filter designed for 50 Ω source impedance will perform poorly if the actual source impedance is 0.1 Ω.