Conducted Emissions

Modern electronics are defined by their speed and efficiency, but these advancements come at a hidden cost: the generation of unwanted electrical noise. This noise, known as conducted emissions, travels along power cords, capable of disrupting other devices and compromising system reliability. Understanding and controlling this invisible chaos is one of the fundamental challenges in electrical engineering, marking the difference between a product that works in harmony with its environment and one that creates electronic pollution. This article addresses the critical knowledge gap between ideal circuit theory and the complex physical reality of high-frequency noise in power systems.

This exploration will guide you through the intricate world of conducted emissions, providing a comprehensive understanding of both its causes and cures. First, in "Principles and Mechanisms," we will delve into the fundamental physics, dissecting noise into its two primary forms—differential-mode and common-mode. We will trace their origins back to the violent switching events within power converters and explore the basic tools used to tame them. Subsequently, in "Applications and Interdisciplinary Connections," we will bridge theory and practice, examining the real-world engineering trade-offs between efficiency and noise, advanced mitigation strategies, and the surprising ways this field intersects with system safety and predictive diagnostics.

Imagine you are a doctor with superhuman hearing, able to listen to the flow of electricity inside a power cord. To a normal observer, the cord supplies a steady, predictable stream of energy to your laptop or phone charger. But to you, it's a cacophony. Beneath the powerful, low-frequency hum of the mains power, you hear a chaotic symphony of high-frequency whistles, clicks, and buzzes. This is the world of conducted emissions—the unwanted electrical noise generated by modern electronics that travels, or is conducted, along power lines.

This noise isn't just an academic curiosity; it's the reason your AM radio crackles when a dimmer switch is on, and why engineers spend countless hours ensuring that one piece of medical equipment doesn't interfere with another. To understand this phenomenon, we must first learn to distinguish the different "voices" in this choir of noise. Then, we must become detectives, tracing these voices back to their source deep within the heart of our devices: the humble electronic switch.

If we could watch the high-frequency noise currents flowing on the two wires of a power cable—the line and the neutral—we would see that they conspire to create two fundamentally different kinds of disturbances. By applying a beautifully simple piece of mathematical insight, akin to separating a musical chord into its individual notes, we can decompose the complex, messy reality into two pure forms: ​​differential-mode (DM)​​ and ​​common-mode (CM)​​ noise.

​​Differential-mode (DM) noise​​ is the more intuitive of the two. It behaves like a proper, well-mannered electrical current. A noise current flows out on the line conductor and an equal and opposite current flows back on the neutral conductor. It's a closed loop, a private conversation confined to the two wires. Think of it as a ripple on the surface of the main river of power, flowing within the river's existing banks.

​​Common-mode (CM) noise​​, on the other hand, is a rogue agent. It is characterized by noise currents that flow in the same direction on both the line and neutral wires. But if the current is flowing out on both wires, Kirchhoff's Current Law—one of physics' most steadfast accounting principles—demands that the current must return somewhere. Where does it go? It returns through an unintended, invisible path: the "ground." This "ground" might be the third protective earth wire in your plug, the metal chassis of the device, or even just the capacitance to the table the device is sitting on. This is not a private conversation; it is a public broadcast, with current leaking out into the wider environment before finding its way home. This makes it particularly troublesome, as it's far more likely to radiate and interfere with other systems.

So, where does all this noise come from? The villain of our story is the very thing that makes modern electronics so efficient: the fast-acting electronic switch. In a typical power converter, like the one charging your phone, transistors switch on and off hundreds of thousands, or even millions, of times per second. Each switch is an incredibly violent event on a microscopic scale, governed by two fundamental consequences of Maxwell's equations.

The Sinner: Fast Changing Currents ($di/dt$)

First, consider the current. When a switch closes, the current can ramp up from zero to several amperes in a few nanoseconds. This rapid change in current, or high ​​slew rate ($di/dt$)​​, creates a rapidly changing magnetic field. As Faraday's Law of Induction teaches us, a changing magnetic flux through a loop of wire induces a voltage in that loop.

Inside a power converter, the switching current flows in a tight path called the "hot loop," typically comprising the input capacitor and the switching transistors. This loop, no matter how well-designed, has some physical area and therefore some small amount of inductance, $L$. A high $di/dt$ flowing through this inductance creates a voltage disturbance, $v = L \frac{di}{dt}$. This rapidly changing magnetic field acts like a tiny, powerful radio transmitter. If the input power cable happens to form a loop nearby, this magnetic field will induce a voltage in it, driving a ripple of differential-mode noise out onto the power line.

The beauty here is that the solution comes directly from the physics. To reduce the magnetic field, we must reduce the area of the "hot loop". By routing the forward and return paths of the current as close together as possible, their opposing magnetic fields cancel each other out. This elegant principle of minimizing loop area is a cornerstone of good electronic design, turning a shout into a whisper by applying fundamental physics.

The Ghost: Fast Changing Voltages ($dv/dt$)

The second, and often more problematic, source of noise comes from rapidly changing voltages. When a switch opens, the voltage across it can swing by hundreds of volts in mere nanoseconds. This is a high voltage slew rate, or $dv/dt$. Here we encounter one of the most profound ideas in physics: Maxwell's ​​displacement current​​. A changing electric field in the vacuum or an insulator acts, in a way, just like a real current.

Think of any two conductive surfaces separated by an insulator—this forms a capacitor. Even if there's no physical connection, a high $dv/dt$ across them will drive a current "through" the capacitor, given by the simple and powerful relation $i = C \frac{dv}{dt}$. The problem is that our electronic circuits are filled with unintentional, "parasitic" capacitances. The metal pad of a switching transistor has a tiny capacitance to the metal chassis it's mounted on. The primary and secondary windings of a transformer, though galvanically isolated, have capacitance between them.

This is the primary origin of common-mode noise. A switching node swinging by hundreds of volts drives a significant displacement current through these parasitic capacitances into the chassis or secondary circuit. Let's consider a modern power converter using wide-bandgap transistors. It might have a switch-node slew rate of $160 \text{ V/ns}$—that's 160 billion volts per second! If this node has a mere $30 \text{ pF}$ of parasitic capacitance to the chassis (roughly the capacitance between two coins held a few centimeters apart), the peak displacement current is a staggering $i = (30 \times 10^{-12} \text{ F}) \times (160 \times 10^9 \text{ V/s}) = 4.8 \text{ A}$!. This enormous spike of current, injected directly into the chassis, is the prime suspect for our common-mode noise problem. This current then embarks on a journey through the system, seeking a path back to its source, and in doing so, pollutes the power lines.

To understand how to control common-mode noise, we must understand the paths it takes. This requires clearly defining our terms for "ground".

• The ​​Chassis​​ is the conductive metal box housing the electronics. It's our local reference plane.
• The ​​Protective Earth (PE)​​ is the safety wire connected to the physical Earth, designed to protect users from electric shock.
• The ​​Functional Ground​​ is the internal $0 \text{ V}$ reference for the delicate control logic.

The common-mode current, born from a high $dv/dt$ node, is first injected into the chassis. From there, it needs to get back to the main power circuit. Its primary intended path back to the global reference (Earth) is through the Protective Earth strap. If this strap is a short, wide piece of metal, it offers a very low-impedance path, effectively tying the chassis to ground and keeping it quiet. However, if the strap is long and thin, it has significant inductance. At high frequencies, this inductance can present a huge impedance, causing the "grounded" chassis to develop a significant high-frequency voltage, turning the entire box into an unwanted antenna that radiates noise.

Knowing the sources and paths, we can finally devise strategies to tame the beast. Our tools are filters, shields, and careful design.

For ​​differential-mode noise​​, the solution is conceptually straightforward. Since the noise is confined to the two power wires, we can place inductors in series with the wires to block the high-frequency current and a capacitor (an ​​X-capacitor​​) across the wires to provide a short-circuit path for it.

For ​​common-mode noise​​, we need a more sophisticated approach.

• We use a ​​common-mode choke​​. This brilliant device has two windings on a single core, one for the line and one for the neutral. It's wound such that for normal DM current, the magnetic fields cancel out, and it looks like a simple wire. But for CM current flowing in the same direction, the magnetic fields add up, creating a huge impedance that blocks the noise. It is a filter that is "smart" to the type of current passing through it.
• We add ​​Y-capacitors​​, which are small capacitors connected from each power line to the chassis. These provide an easy, local return path for the common-mode current, coaxing it to circulate harmlessly within the device instead of venturing out onto the power cord.
• In isolated systems, we can use an ​​electrostatic shield​​ (or Faraday shield) inside the transformer. This is a thin layer of copper placed between the primary and secondary windings and connected to a quiet reference point. It intercepts the displacement current from the primary's $dv/dt$ and returns it locally, preventing it from ever crossing the isolation barrier and creating a common-mode problem.

However, there is a final, humbling twist. Our components are not ideal. A capacitor, our go-to tool for shunting noise, has some small but unavoidable parasitic inductance from its physical construction, known as ​​Equivalent Series Inductance (ESL)​​. The capacitor and its own ESL form a series resonant circuit. Below resonance, it acts as a capacitor. At a specific ​​self-resonant frequency (SRF)​​, $f_{res} = \frac{1}{2\pi\sqrt{LC}}$, its impedance drops to a minimum (ideally zero). But above the SRF, the inductive nature dominates, and its impedance increases with frequency. A $10 \text{ nF}$ capacitor, which has an impedance of just a fraction of an ohm at a few megahertz, might look like a high-impedance inductor at a few tens of megahertz, rendering it useless or even harmful for filtering. This is a profound lesson: at high frequencies, everything matters, and a circuit diagram is only a pale shadow of the physical reality.

To verify our designs, we need a standardized measurement. This is the role of the ​​Line Impedance Stabilization Network (LISN)​​. It's a special box that sits between the power source and our device. It does two things: it isolates the device from the unpredictable noise already on the power grid, and it presents a precise, stable $50\,\Omega$ impedance for the noise currents to flow into. This ensures that a measurement made in a lab in Tokyo is comparable to one made in Munich, providing a common language for the global effort to keep our electronic world peacefully coexisting. This standardized measurement is then compared against regulatory limits, such as those from CISPR, which are carefully crafted to protect our radio spectrum from this unintentional pollution.

The study of conducted emissions, then, is a journey from simple circuit laws to the depths of Maxwell's equations, and from idealized components to the messy, beautiful complexity of the real world. It is a field where careful observation, physical intuition, and an appreciation for fundamental principles are the keys to taming the unseen chaos.

Our exploration of conducted emissions has, until now, focused on the fundamental principles—the why and what of this invisible electronic chatter. But to truly appreciate the subject, we must venture beyond theory and see how these principles come to life. This is where the real fun begins. It turns out that understanding conducted emissions is not merely an academic exercise in satisfying regulations; it is a gateway to designing more efficient, reliable, and safer technology. It is a field where physics, engineering, safety, and even diagnostics converge in a beautiful and often surprising dance.

At the heart of modern power electronics lies a fundamental tension. To make devices smaller and more efficient, we must switch currents and voltages at incredible speeds. A modern Silicon Carbide (SiC) or Gallium Nitride (GaN) transistor can switch hundreds of volts in a few nanoseconds, a feat of engineering that minimizes the time the device spends in a resistive state, thereby saving energy. But this very speed is the villain in our story. The sharp edges of these switching waveforms, the large rates of change of voltage ($dv/dt$) and current ($di/dt$), are incredibly rich in high-frequency harmonics. This is the origin of conducted emissions.

So, the engineer faces a classic trade-off. We can slow down the switching to make the converter "quieter," but this increases switching losses, generating more heat and reducing efficiency. Consider the gate resistor of a MOSFET, a tiny component that controls how quickly the switch turns on and off. Increasing this resistance from a few ohms to ten ohms might slow the voltage slew rate by a factor of five. This deliberate slowdown dramatically quiets the converter, potentially reducing the high-frequency noise amplitude by as much as 14 dB—a huge improvement. However, this comes at a cost. The longer switching transition means the device spends more time in a state of simultaneous high voltage and high current, causing the energy lost as heat in each switching cycle to increase fivefold. At a switching frequency of $500\,\mathrm{kHz}$, this could mean going from a manageable $1.4\,\mathrm{W}$ of switching loss to a much more problematic $7.2\,\mathrm{W}$. This single example beautifully encapsulates the central design conflict: the quest for efficiency is a quest for speed, and the quest for speed is an invitation to noise. To have both, we need more clever solutions than simply "slowing down."

Before we can tame this electromagnetic beast, we must first learn to see and measure it. This is done using a specialized piece of equipment called a Line Impedance Stabilization Network, or LISN. The LISN is our standardized "ear," providing a consistent impedance against which we can listen to the noise currents a device is feeding back onto the power lines.

With a way to listen, we need a language to describe what we hear. When we add a filter to a noisy device, how do we quantify its effectiveness? The universal metric is ​​Insertion Loss​​ ($IL$). On the surface, it's a simple idea: the ratio, in decibels, of the noise voltage without the filter to the noise voltage with the filter. But here lies a subtlety that traps many an unwary engineer. One might naively think that insertion loss is an intrinsic property of the filter itself, like its mass or color. It is not.

The effectiveness of an EMI filter is profoundly dependent on the impedances of the noise source and the load it's connected to. The simple, elegant formula relating insertion loss to the filter's transmission characteristic, $IL(f) = -20\log_{10}|S_{21}(f)|$, holds true only in a perfectly "matched" system where the source and load impedances are equal to the filter's characteristic impedance (typically $50\,\Omega$). In the real world of power electronics, source and load impedances are wildly complex, non-linear, and vary dramatically with frequency. The true expression for insertion loss is a more complicated affair involving the filter's network parameters and both the source and load impedances. This tells us something crucial: you cannot design a filter in isolation. You must design it for the specific system it will live in.

Furthermore, even a perfectly designed filter on paper can be rendered useless by poor physical layout. A tiny, almost invisible parasitic capacitance between the filter's input and output—perhaps just a few picofarads—can create a "sneak path" for high-frequency noise to bypass the filter entirely. Similarly, long, thin traces connecting a filter capacitor can add parasitic inductance, making the capacitor useless at the very high frequencies it's meant to suppress. EMI is not just a circuit diagram problem; it is a three-dimensional physics problem.

Armed with the tools of measurement, we can now devise strategies to quiet our circuits. The approaches are as varied as they are ingenious, ranging from brute-force filtering to elegant architectural choices that prevent noise from being generated in the first place.

Attacking the Noise at its Source

The most elegant solutions tackle the problem at its root. Instead of letting the beast roar and then trying to muffle it, why not persuade it to be quiet?

One direct method is to shape the switching waveforms. While aggressive slowing of the entire transition is inefficient, we can use a ​​snubber circuit​​—typically a simple resistor and capacitor—to selectively slow down just the voltage transition ($dv/dt$). By modeling the switching voltage's Fourier spectrum, we can see that a slower edge rate rolls off the high-frequency content more steeply. Halving the $dv/dt$ of a switching waveform, for instance, results in a clean 6 dB reduction in the conducted noise voltage it generates. Implementing a snubber that turns a dangerously fast $100\\,\\mathrm{V/ns}$ transition into a more manageable $20\\,\\mathrm{V/ns}$ can reduce the EMI at $30\\,\\mathrm{MHz}$ by a remarkable 14 dB.

Even more profound is the choice of the converter's fundamental architecture, or ​​topology​​. Consider two ways to build a converter that can both step-up and step-down voltage: the SEPIC and the Cuk topologies. While they perform similar DC functions, their EMI "personalities" are vastly different. The Cuk converter is a thing of beauty from an EMI perspective. It naturally places inductors—which act like flywheels for current, smoothing its flow—in series with both the input source and the output load. The result is continuous, smooth current at both ends. The pulsating, noisy currents associated with the fast-switching action are cleverly confined to an internal loop. The SEPIC, in contrast, delivers current to the output in sharp, pulsating bursts. This makes its output inherently noisier. This shows that good EMI design begins with the first lines drawn on the whiteboard.

For an even more elegant source-suppression technique, consider ​​interleaving​​. If one converter creates a ripple current, why not add a second, identical converter and run it perfectly out of phase? By running two converter phases with a $180$-degree phase shift, their individual ripple currents cancel each other out at the input. The superposition is exquisite: the fundamental component of the switching ripple can, in theory, be completely eliminated. This is the same principle behind noise-cancelling headphones, applied to the flow of power. The result is a much cleaner input current, which dramatically reduces the size, cost, and complexity of the required input filter.

Finally, even the ​​control strategy​​ can be an EMI mitigation tool. In a Power Factor Correction (PFC) circuit operating in Critical Conduction Mode (CrCM), the switching frequency is not constant. It naturally varies over the AC line cycle, sweeping from a high frequency near the zero-crossings to a lower frequency at the peak of the line voltage. This has a wonderful side effect: the noise energy, instead of being concentrated in sharp peaks at fixed frequencies, is "smeared" out over a wider band. To a regulatory receiver with a narrow bandwidth, the noise appears less intense, a phenomenon known as spread-spectrum modulation. This can be the difference between passing and failing an EMI test, achieved not with a filter, but with a clever control algorithm.

As we've hinted, conducted emissions are deeply tied to the physical world of layout, grounding, and safety. Nowhere is this clearer than in the study of common-mode (CM) noise. This is the noise current that flows in unison down both power lines and returns through an external path—often, the earth ground.

The driver for this noise is a high $dv/dt$ on a switching node coupled through a tiny parasitic capacitance to a grounded chassis or heatsink. The relationship is simple and powerful: $i_{cm}(t) = C_{parasitic} \frac{dv}{dt}$. With the blistering $dv/dt$ of modern WBG devices (e.g., $200\\,\\mathrm{V/ns}$), even a minuscule capacitance of $100\\,\\mathrm{pF}$ can generate a shocking peak noise current of $20\\,\\mathrm{A}$!

This leads to a fascinating and somewhat counter-intuitive insight when comparing earthed (three-wire) and unearthed (two-wire) equipment. One might guess the earthed appliance is inherently "quieter." The opposite is often true. That very earth connection provides a convenient, low-impedance return path for common-mode currents to flow. An unearthed, "floating" device, encased in plastic, offers a much higher impedance path, starving the noise current and making the system inherently less prone to generating large common-mode emissions.

The presence of an earth connection also brings us to a crucial intersection with safety engineering. To filter common-mode noise, designers use "Y-capacitors" to shunt the noise from the power lines directly to the earthed chassis. These capacitors are vital for good EMI performance. However, they also create a path for a small amount of current to flow from the mains line to the chassis at the line frequency (50 or 60 Hz). This is called ​​leakage current​​. To prevent electric shock hazards, safety standards like those from the IEC strictly limit this current, often to less than a milliampere. This creates another fundamental trade-off: the desire for larger Y-capacitors for better EMI filtering is in direct conflict with the need for smaller Y-capacitors for user safety. The final design is always a carefully judged compromise between these two opposing constraints.

Perhaps the most forward-looking application of our understanding of conducted emissions lies in an entirely different field: diagnostics and condition monitoring. The conducted EMI spectrum of a power converter is, in a very real sense, its unique fingerprint. A healthy converter, operating as designed, produces a predictable and stable noise signature.

What happens when a component begins to fail? The fingerprint changes. An aging aluminum electrolytic capacitor in the input filter might see its internal resistance (ESR) increase. This fault has a dual effect on the EMI spectrum: the filter's resonance peak becomes more damped (lower in amplitude), but its overall filtering performance in the mid-band gets worse, raising the noise floor. In another scenario, if the main power inductor starts to saturate due to overload or degradation, its inductance becomes non-linear. This distorts the current waveform, dramatically increasing the magnitude of the low-order switching harmonics. A failing gate driver might cause faster, uncontrolled switching edges, leading to a broadband rise in the high-frequency noise floor.

Each of these failure modes produces a distinct, identifiable change in the conducted EMI spectrum. By monitoring a device's EMI signature over its lifetime, we can perform non-invasive diagnostics. We can "listen to the heartbeat" of the converter and detect the subtle signs of component degradation long before a catastrophic failure occurs. This opens up exciting new possibilities for building self-aware, highly reliable systems, connecting the world of EMI to predictive maintenance and the Internet of Things (IoT).

In the end, conducted emissions are far more than a regulatory nuisance. They are a window into the very soul of an electronic device. To understand them is to understand the deep interplay of efficiency and noise, the tension between ideal schematics and physical reality, the compromise between performance and safety, and the powerful connection between a system's electrical noise and its physical health. The quest to design "quiet" electronics is a profound journey that reveals the beautiful unity of science and engineering.