Eddy Current Losses — The Magnetic Components That Sizzle When You Don't Pay Attention

"Eddy currents are like tiny lazy rivers inside your core: they go in circles, heat things up, and refuse to do useful work." — Your future, slightly singed power supply

You're already familiar with magnetic core materials and why we pick ferrites, silicon steel or amorphous alloys, and you know how inductors behave in converters. Now let's zoom in on one of the core's worst freeloaders: eddy current losses. These are the invisible, frequency-hungry heat monsters that punish careless designs and laugh at high switching speeds.

---

Why this matters (and yes, it's personal)

• Eddy losses directly steal efficiency, raise temperature, and shorten life. In a power stage with constrained cooling, that extra watt or ten can be the difference between 'works' and 'melts into a sad blob.'
• They scale with frequency squared (f^2) and B^2: so when you crank switching frequency to shrink your magnetics — hello, surface-mount dreamland — eddy losses can explode unless you pick the right materials and geometry.
• Control strategies (like soft-switching or dwell-time management) you learned earlier can dramatically reduce dB/dt, lowering eddy heating — control meets materials in a dance-off. Win both and your converter stays cool and efficient.

---

What are eddy currents? The physics elevator pitch

Induced voltages from a changing magnetic field create circulating currents inside any conductive piece of magnetic material. Those currents flow in loops (like tiny whirlpools) and dissipate energy as heat because the core has finite resistivity. Unlike hysteresis loss (which is about domain flipping and is roughly linear with f), eddy loss grows as f^2. Ouch.

---

The simple formula (and how to use it)

For a thin lamination (thickness t), a common approximation for eddy loss per unit volume is:

P_e ≈ (π^2 * t^2 * f^2 * B_m^2) / (6 * ρ)

where:
  P_e = eddy power per unit volume (W/m^3)
  t   = lamination thickness (m)
  f   = frequency (Hz)
  B_m = peak flux density (T)
  ρ   = resistivity of the material (Ω·m)

Key takeaways from that equation:

• P_e ∝ t^2. Halving thickness cuts eddy loss by 4x.
• P_e ∝ f^2. Double the switching frequency and eddy losses quadruple.
• P_e ∝ B_m^2. Lower flux swing (via larger core or different control) helps.

There’s also the empirical Steinmetz-style form for total core loss:

P_core = k_h * f * B_m^β + k_e * f^2 * B_m^2

The first term is hysteresis-like, the second is eddy-dominated. Datasheets often give Steinmetz parameters — use them.

---

A tiny worked example (no, you won't die of math)

Suppose an EI lamination stack with t = 0.35 mm (0.00035 m), ρ = 5e-7 Ω·m (typical silicon steel), B_m = 1.2 T, f = 50 Hz (mains transformer). Plug in:

• π^2 ≈ 9.87
• t^2 ≈ 1.225e-7
• f^2 = 2500
• B_m^2 = 1.44

Numerator ≈ 9.87 * 1.225e-7 * 2500 * 1.44 ≈ 0.00435
Denominator = 6 * 5e-7 = 3e-6

P_e ≈ 0.00435 / 3e-6 ≈ 1450 W/m^3

So for 1 liter (0.001 m^3) of core you get ~1.45 W of eddy loss at 50 Hz. Reasonable. But now imagine f = 50 kHz (SMPS realm) — that term skyrockets by (1000)^2 = 1e6, so the lamination trick fails and you turn to ferrites (high resistivity) and powdered cores.

---

How designers kill eddy currents (a hit-list)

1. Laminations: Stack thin insulated sheets so the eddy loops are interrupted. Classic for power-frequency transformers.
2. Use high-resistivity materials: Ferrites are ceramic and have huge resistivity, so eddy currents are negligible at kHz and above. Powdered iron cores (with insulating particles) also limit eddy paths.
3. Make the effective lamination thickness < skin depth: skin depth δ = sqrt(2ρ/(ωμ)). If t << δ, currents can’t form big loops.
4. Reduce flux swing B_m: push more core cross-section or change the magnetizing current via control (e.g., limit duty or use flux reset techniques).
5. Control strategies: soft-switching, slew-rate control, and spreading the switching events (interleaving / spread-spectrum PWM) reduce peak dB/dt or average switching energy and hence eddy heating.
6. Shape & slotting: in rotating machines, slotting breaks large circulating paths. In static cores, cut the path geometry so long closed loops are discouraged.
7. Annealing and grain-orientation: for silicon steel, processes that increase resistivity or change magnetic properties can reduce core losses.

---

Interplay with control strategies — the neat part

You learned control strategies earlier to optimize performance. Here's how they directly help with eddy losses:

• Soft-switching (ZVS/ZCS) lowers dv/dt and di/dt during transitions, reducing high-frequency flux spikes and spectral energy that excites eddy currents.
• Spread-spectrum PWM avoids concentrating switching energy at a single frequency (reduces f^2 impact at that frequency), smoothing losses across bandwidth.
• Adaptive frequency control: reduce switching frequency under low-load to cut eddy loss power. Dynamic trade-offs — size vs efficiency.
• Flux reset control ensures the core never saturates wildly; saturation excursions generate larger harmonics and more eddy heating.

In short: good control can be as important as good materials. Think of it as behavioral therapy for your magnetic core.

---

Measurement and specification — how to know what you're buying

• Datasheets give Steinmetz coefficients or loss vs frequency/B plots. Use them.
• Lab: Epstein frame or ring core setups measure core loss per unit mass/volume. Compare with operating B and frequency.
• Thermal testing under worst-case control waveforms is crucial: periodic square-ish flux waveforms produce more high-frequency content than perfect sine assumptions.

---

Final takeaways (so you can solve this on an exam or in the lab between coffee refills)

• Eddy losses are frequency-hungry and geometry-sensitive: they scale with t^2, f^2, and B^2.
• Use thin laminations at mains frequencies, ferrites / powdered cores at switching frequencies above a few kHz.
• Skin depth and resistivity are your friends — make lamination thickness smaller than skin depth or choose high-resistivity materials.
• Control strategies (soft-switching, PWM shaping, frequency modulation) are powerful tools to reduce eddy heating — material + control = best results.

"If your converter's hottest part is the core, you did one of two things wrong: you ignored material physics or you made your controller a bully." — Professor Sarcastic but Useful

Now: go pick a core, check the datasheet for Steinmetz numbers, and if you still want to crank frequency, bring ferrite and a good thermal plan. Your solder joints will thank you.

---

Version note: This builds on our prior discussion of core materials and inductor behavior — remember: we choose material for resistivity, permeability and loss characteristics, and we size inductors to limit flux. Eddy currents are the reason those choices matter so much when frequency goes up.