Steinmetz Equation

In the world of modern electronics, managing energy efficiency is paramount, and a significant challenge lies in predicting and controlling the waste heat generated within magnetic components. This energy dissipation, known as core loss, is a complex phenomenon occurring inside the inductors and transformers that are fundamental to countless devices. The Steinmetz equation emerges as a powerful and elegant tool that provides engineers with a practical method to quantify this loss, bridging the gap between abstract material physics and concrete engineering design. This article explores the depth and utility of this crucial formula. First, it will delve into the underlying physical principles and mechanisms that govern the equation, explaining how it approximates the combined effects of hysteresis and eddy currents. Following this, it will examine the equation's diverse applications and interdisciplinary connections, demonstrating its role in designing efficient power supplies and managing thermal constraints in the real world. Our exploration begins by dissecting the core principles that give the Steinmetz equation its predictive power.

An equation as simple and elegant as the one Charles Proteus Steinmetz proposed is both a delight and a puzzle. A delight because it tames a complex, invisible process—energy loss in a magnetic material—and packages it into a tidy power-law relationship. And a puzzle because we are compelled to ask: Why? Why this particular form? Is it a lucky guess, a mere curve-fit, or does it hint at a deeper, more beautiful physical story? The journey to answer this question takes us from simple empirical rules to the heart of electromagnetism.

The classical ​​Steinmetz equation​​ states that the power lost as heat, per unit volume of a magnetic core material, follows a simple rule:

$P_v = k f^{\alpha} B_{\text{pk}}^{\beta}$

Here, $P_v$ is the volumetric power loss, $f$ is the frequency of the alternating magnetization, and $B_{\text{pk}}$ is the peak magnetic flux density reached in each cycle. The terms $k$, $\alpha$, and $\beta$ are the famous Steinmetz parameters, numbers determined by experiment for each specific material.

At first glance, this might seem like just another engineering formula. But it has a rather beautiful geometric interpretation. If you were to take measurements of loss at different frequencies (while keeping $B_{\text{pk}}$ constant) and plot them on a special type of graph paper called a log-log plot, you would find the data points form a nearly straight line. The slope of that line is the exponent $\alpha$. Similarly, a log-log plot of loss versus peak flux density would give a straight line with a slope of $\beta$. The equation is simply a statement about these straight lines. But this still doesn't tell us why the lines are straight. To find out, we must look inside the material.

The total energy loss isn't due to a single cause. It's the sum of several distinct physical processes happening simultaneously. For most materials, two characters play the leading roles: the first is a kind of internal friction, and the second is a swarm of unwanted electrical whirlpools.

Hysteresis: The Cost of Indecision

Imagine the magnetic material is filled with tiny magnetic compass needles, called ​​magnetic domains​​. When we apply an external magnetic field, these domains try to align with it. As the field reverses, they have to flip around. This flipping isn't perfectly smooth; the domains snag and rub against imperfections in the material's crystal structure. This process creates friction, dissipating energy as heat. This loss is called ​​hysteresis loss​​.

The energy lost in one complete back-and-forth cycle is equal to the area of the material's magnetic ​​hysteresis loop​​ (the B-H loop). To find the power loss—the energy per second—we simply multiply this per-cycle energy by the number of cycles per second, which is the frequency $f$. This immediately tells us something profound: the hysteresis power loss, $P_h$, must be directly proportional to frequency.

$P_h \propto f^1$

The dependence on flux density is more complicated, as it relates to how "wide" the hysteresis loop gets as we push the material harder. Empirically, this energy scales roughly as $B_{\text{pk}}^n$, where $n$ is often between 1.6 and 2.5 for magnetic steels. Putting it together, the hysteresis component of the loss looks something like $P_h \approx k_h f^1 B_{\text{pk}}^n$.

Eddy Currents: Unwanted Electrical Whirlpools

The second character in our story comes directly from one of the pillars of physics: ​​Faraday's Law of Induction​​. Faraday discovered that a changing magnetic field creates an electric field. Now, if the magnetic core material is an electrical conductor (even a poor one, like ferrite), this induced electric field will drive currents within the core itself. These are not the useful currents in the windings of our transformer or inductor; they are parasitic currents that swirl around inside the core material like little whirlpools, or eddies. And just like any current flowing through a resistor, they dissipate energy as heat ($I^2R$ loss). This is ​​eddy current loss​​.

We can reason out how this loss depends on frequency and flux density. For a sinusoidal magnetic field, the rate of change of the field, $\frac{dB}{dt}$, is proportional to both the frequency $f$ and the peak flux density $B_{\text{pk}}$. From Faraday's Law, the induced electric field $E$ is proportional to this rate of change, so $E \propto f B_{\text{pk}}$. The power dissipated is proportional to the square of the electric field ($E^2$), so the eddy current power loss, $P_e$, must scale as:

$P_e \propto (f B_{\text{pk}})^2 = f^2 B_{\text{pk}}^2$

This is a beautiful and direct consequence of fundamental physics.

Now the puzzle of the Steinmetz exponents $\alpha$ and $\beta$ begins to resolve itself. The total loss we measure is the sum of these two effects (plus a third, more subtle "excess loss" related to domain wall dynamics:

$P_v = P_h + P_e + \dots \approx k_h f^1 B_{\text{pk}}^n + k_e f^2 B_{\text{pk}}^2$

The Steinmetz equation is simply an empirical approximation—a convenient single-term formula—that tries to mimic this more complex sum. The exponents $\alpha$ and $\beta$ are therefore not fundamental constants but effective values that depend on which loss mechanism is the star of the show in a given operating range.

• ​​At low frequencies​​, the $f^1$ hysteresis term dominates the sum. If we fit a power law to data in this region, we will find an effective frequency exponent $\alpha$ close to 1.
• ​​At high frequencies​​, the $f^2$ eddy current term grows much faster and eventually dominates. A fit in this region will yield an exponent $\alpha$ that approaches 2.

This is why, for most materials over a practical frequency range, the measured value of $\alpha$ lies somewhere between 1 and 2. The value of $\beta$ is likewise a blend, typically falling between the hysteresis exponent $n$ (often $n > 2$ for ferrites) and the eddy current exponent of 2. There is a "crossover" frequency where the two loss types are equal; below this frequency, hysteresis reigns, and above it, eddy currents take over. This crossover point is not fixed; it can be shifted. For example, by using very thin laminations in a steel core, we dramatically reduce the size of the eddy current whirlpools, lowering the eddy current coefficient $k_e$ and pushing the crossover to much higher frequencies. This is precisely why transformers for power grids use laminated steel.

The classical Steinmetz equation was born in the age of the electrical grid, a world of smooth, clean, sinusoidal waves. Modern power electronics, however, speak a different language: the sharp, abrupt language of digital switches. The voltage and current waveforms in a computer power supply or an electric vehicle charger are not sines; they are triangles, squares, and trapezoids.

Here, the simple equation begins to fail, and for a very deep reason. The true physical driver for dynamic losses like eddy currents is not "frequency" as a cycle-averaged concept, but the instantaneous rate of change of the magnetic flux, $\frac{dB}{dt}$.

Consider a thought experiment. Imagine a trapezoidal flux waveform that ramps up, holds steady for a "dwell time," ramps down, and holds steady again. Let's keep the period and the peak flux the same, but we make the dwell times longer. To get from the bottom to the top in less time, the ramps must become steeper—meaning $\frac{dB}{dt}$ during the ramps is much larger.

The classical Steinmetz equation, seeing only a constant frequency ($1/T$) and a constant peak flux, would predict the exact same loss. But this can't be right! The much faster-changing flux during the ramps must induce larger electric fields and thus greater eddy current losses. The experiment proves our intuition correct: the loss increases significantly. The classical model is blind to the shape of the wave.

To solve this, physicists and engineers developed more sophisticated models, like the ​​Improved Generalized Steinmetz Equation (iGSE)​​. The beauty of this new approach is its shift in perspective. Instead of thinking about average power over a cycle, it proposes a formula for the instantaneous power loss, $p_v(t)$. This instantaneous loss is a direct function of the instantaneous rate of change of flux:

$p_v(t) \propto \left| \frac{dB}{dt} \right|^\alpha$

To find the total average power, one simply adds up (integrates) this instantaneous loss over one full cycle. This "waveform-aware" model correctly predicts what happens in our thought experiment. During the dwell times, $\frac{dB}{dt}=0$, and the instantaneous loss is zero. During the steep ramps, $\frac{dB}{dt}$ is large, and the instantaneous loss is large. The model correctly captures the physics that the classical equation missed.

The journey doesn't end there. The real world is always richer and more complex than our models. For instance, in many modern inductors, a physical air gap is cut into the core. The magnetic flux has to "jump" this gap, and in doing so, it bulges outwards in what is called ​​fringing flux​​. To feed this bulge, the flux inside the core material must crowd together at the edges of the gap. This crowding creates local ​​hot spots​​ where the flux density $B$ and its rate of change $\frac{dB}{dt}$ are much higher than the average value in the core. A simple model assuming uniform flux would completely miss these hot spots and dangerously underpredict the total loss.

Furthermore, the "constants" in the Steinmetz equation are not truly constant. They all depend on ​​temperature​​. As a component heats up, its magnetic properties change. For many common ferrite materials, for example, their electrical resistivity actually decreases as they get hotter, making them better conductors and worsening the eddy current problem. An accurate model for a real-world device must account for this by treating the Steinmetz parameters themselves as functions of temperature: $k(T)$, $\alpha(T)$, and $\beta(T)$.

From a simple empirical rule, we have journeyed through the fundamental physics of hysteresis and eddy currents, uncovered the limitations of our models, and developed more powerful ones to take their place. The Steinmetz equation, in its classical and modern forms, is a perfect example of science in action: a continuous cycle of observation, explanation, prediction, and refinement, always seeking a truer, more beautiful description of the world.

Having acquainted ourselves with the principles behind the Steinmetz equation, we might be tempted to view it as a mere empirical curiosity, a convenient but somewhat arbitrary formula. Nothing could be further from the truth. This simple-looking relation is, in fact, a powerful bridge connecting the abstract world of material science to the concrete, demanding reality of engineering. It is the lens through which we can understand, predict, and ultimately control one of the most critical and ubiquitous challenges in modern technology: the generation of waste heat in magnetic components. Let us embark on a journey to see how this equation unfolds its power across various disciplines and applications, transforming from a simple rule of thumb into a sophisticated design tool.

At the core of nearly every electronic device you own—from the charger for your phone to the power supply in your computer to the vast systems that manage our electrical grid—lies a family of components called switch-mode power supplies. These devices are the unsung heroes of energy efficiency, and their operation hinges on the rapid switching of energy into and out of magnetic components like inductors and transformers.

When an engineer sets out to design, say, the inductor for a power factor correction (PFC) circuit that ensures your appliances draw power cleanly from the wall, they face a delicate balancing act. The circuit's voltage and switching frequency dictate how rapidly the magnetic flux ($B$) inside the inductor's core must change. Faraday's law gives us this connection directly. Once we know the peak flux density swing, $B_{\text{peak}}$, and the operating frequency, $f$, the Steinmetz equation steps in and gives us a number: the power that will be lost as heat per unit volume of the core material. This isn't an academic exercise; this number tells us how hot the component will get, how much energy it will waste, and ultimately, how small and efficient we can make it.

This process reveals a fascinating and rather counter-intuitive design principle. Consider designing the inductor for a common buck converter, which steps down voltage efficiently. We can derive the flux density ripple, $\Delta B$, directly from the converter's voltages and switching frequency. When we plug this into the Steinmetz equation, we find that the core loss density often depends on frequency as $P_v \propto f_s^{\alpha - \beta}$. For most ferrite materials, the flux density exponent $\beta$ (typically around 2.5) is larger than the frequency exponent $\alpha$ (typically around 1.5). This means the core loss is proportional to $f_s$ raised to a negative power!

What does this imply? It means that, all else being equal, increasing the switching frequency can actually decrease the core loss. This happy circumstance arises because a higher frequency demands a smaller flux swing per cycle to transfer the same amount of energy, and the loss is far more sensitive to the magnitude of the flux swing (the $B^\beta$ term) than it is to the rate of switching (the $f^\alpha$ term). This insight is a cornerstone of modern power electronics, driving the push toward higher and higher frequencies to achieve smaller, lighter, and more efficient power converters.

The power loss calculated by the Steinmetz equation does not simply vanish. It manifests as heat. This realization brings us to a beautiful interdisciplinary connection: a "thermal handshake" between the laws of magnetism and the principles of thermodynamics. A magnetic component is not just an electrical device; it is a thermal one.

Imagine you are designing a high-frequency transformer. Your design is constrained not only by electrical requirements but also by a strict thermal budget: the core cannot exceed a certain temperature, lest it damage itself or surrounding components. How do you ensure your design is safe? Here, the Steinmetz equation joins forces with one of the simplest laws of heat transfer.

The steady-state temperature rise of the core, $\Delta T$, is simply the total power it dissipates, $P_{\text{core}}$, multiplied by its thermal resistance, $\theta_{\text{th}}$, a measure of how easily it sheds heat to the environment. This gives us the elegant relation $\Delta T = P_{\text{core}} \cdot \theta_{\text{th}}$. We can now work backward. Knowing the maximum allowable temperature rise and the thermal resistance of our chosen core and cooling setup, we can calculate the maximum allowable core loss, $P_{\text{core,allow}}$. This value becomes our power budget. We then turn to the Steinmetz equation, $P_{\text{core}} = V_e \cdot k f^{\alpha} B_{\text{pk}}^{\beta}$, and solve it for the one parameter the designer has the most control over: the peak flux density, $B_{\text{pk}}$. This tells us the absolute magnetic limit we can push the core to without causing it to overheat. This elegant interplay between magnetic loss generation and thermal dissipation is fundamental to the design of all high-power magnetic components.

Our discussion so far has a hidden assumption. The classical Steinmetz equation, with its simple exponents, was developed and characterized using pure sinusoidal waveforms. However, the world of switch-mode power electronics is dominated by square-wave voltages and triangular-wave currents. The flux waveforms inside these components are not smooth sinusoids but sharp, linear ramps. Does this difference in shape matter?

Immensely. The physical mechanisms of core loss are sensitive not just to the amplitude of the flux swing, but to its rate of change, $\frac{dB}{dt}$. A triangular wave has a constant (and high) $\frac{dB}{dt}$ on its rising and falling edges, whereas a sine wave's slope is constantly changing, passing through zero at its peaks. To handle these non-sinusoidal waveforms, the Steinmetz equation had to evolve.

Engineers, in their pragmatism, first developed clever approximations. For a component like a gate-drive transformer excited by a square voltage wave, which produces a triangular flux wave, one could calculate the root-mean-square (RMS) value of this triangular flux and plug it into a Steinmetz formula characterized for sinusoidal RMS values. This is a workaround, a way to use the available data, but it papers over the underlying physics.

A more profound approach is found in the ​​Generalized Steinmetz Equation (GSE)​​ and its relatives, like the iGSE. These advanced models are built on the insight that the instantaneous loss at any moment is proportional to $\left|\frac{dB}{dt}\right|^\alpha$. By integrating this instantaneous loss over the true, non-sinusoidal cycle, we get a far more accurate prediction. This method allows us to precisely quantify the difference in loss between, for example, a triangular wave and a sine wave of the same peak amplitude and frequency. The result is a correction factor, a function of the exponent $\alpha$, that cleanly separates the material physics from the waveform geometry.

This generalization is not just an academic refinement. It is essential for accurately modeling real-world converters. In a flyback converter operating in discontinuous mode, the flux waveform is an asymmetric ramp-up and ramp-down, followed by a "dwell" period where the flux is constant. The GSE framework handles this with ease, calculating the loss during the up-ramp and down-ramp separately and correctly assigning zero loss to the dwell period where $\frac{dB}{dt}=0$. This accuracy is crucial when analyzing phenomena like the transition from continuous to discontinuous conduction mode (CCM to DCM), where the very shape and duration of the flux ramps change, altering the core loss in ways a simple sinusoidal model could never predict.

Another layer of complexity—and another opportunity for interdisciplinary connection—arises when we acknowledge that the magnetic field within a core is rarely uniform. A particularly important example occurs in inductors, which require a physical air gap in the ferrite core to store energy properly. This gap forces the magnetic field lines to "fringe" or bulge out into the surrounding space.

Right at the edges of this gap, the flux lines are crowded together, creating local regions where the flux density, $B$, is significantly higher than in the bulk of the core. At first glance, this might seem like a small effect. But the Steinmetz equation, with its strong dependence on flux density ($P_v \propto B^\beta$ where $\beta$ can be 2.5 or more), tells us otherwise. A small local increase in $B$ can lead to a very large local increase in power loss. If we model the core as two regions—a large bulk volume with flux $B_{\text{bulk}}$ and a small edge volume with a higher flux $B_{\text{edge}}$—we can calculate the total loss. The result is striking: even if the high-flux edge region constitutes only a few percent of the total volume, it can be responsible for a disproportionately large fraction of the total core loss, significantly increasing the component's temperature.

This connects the empirical Steinmetz model to the domain of computational electromagnetics. Engineers use Finite Element Analysis (FEA) software to simulate these complex, non-uniform field patterns. By applying the Steinmetz equation on an element-by-element basis to the simulated field distribution, they can obtain a highly accurate map of power loss density, identifying potential thermal "hotspots" that would be invisible to a simpler, uniform-field analysis.

We arrive at our final and perhaps most profound question: what physics is the Steinmetz equation actually capturing? The loss it describes is not a single phenomenon but a combination of effects. The two most prominent are ​​hysteresis loss​​, the energy required to repeatedly re-align the magnetic domains in the material, and ​​eddy current loss​​, which arises from microscopic circular currents induced in the conductive ferrite by the changing magnetic field.

Here, we can perform a beautiful thought experiment, brought to life by numerical simulation. Using Maxwell's equations, we can calculate the classical eddy current loss from first principles. For a given time-varying magnetic field, we can find the induced electric field, $\mathbf{E}$, and from the material's conductivity, $\sigma$, the resulting eddy current density, $\mathbf{J}$. The total power dissipated by these currents is the volume integral of $\mathbf{E} \cdot \mathbf{J}$.

Let's imagine we have an FEA simulation that gives us the fields throughout a ferrite core. We can compute this first-principles eddy current loss. We can also take the peak flux density in each element from the same simulation and compute the total loss using the empirical Steinmetz equation. When we compare the two numbers, we find something remarkable: the Steinmetz loss is significantly larger than the classical eddy current loss.

This "discrepancy" is not an error. It is physics. It reveals that the Steinmetz equation is not just a formula for eddy currents. It is an empirical model that brilliantly lumps together the classical eddy current loss, the far more complex hysteresis loss, and other "anomalous" loss mechanisms into one compact, practical expression. It acknowledges that a full first-principles calculation of all these intertwined, nonlinear phenomena is extraordinarily difficult. The Steinmetz equation is the triumph of engineering phenomenology—a testament to our ability to create beautifully effective models that capture the essence of complex physical reality, enabling us to design and build the technologies that shape our world.