Electrical

Inductor Energy Storage

Why interrupting current makes a coil bite back

An inductor stores energy in its magnetic field, W = ½LI². Push current through it and the field builds; try to interrupt that current and the collapsing field generates voltages high enough to weld contacts, jump gaps, or supply boost converters. The component is the dual of the capacitor — voltage on the cap mirrors current in the inductor — and its limits are set by core saturation, copper loss, and self-resonance.

  • Energy storedW = ½LI²
  • Voltage lawV = L·di/dt
  • L of solenoidL = µ₀µᵣN²A/ℓ
  • Time constant (RL)τ = L/R
  • SaturationB ≥ B_sat → L collapses
  • Quality factorQ = ωL/R

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

How an inductor stores energy

An inductor is a coil of wire — sometimes around an air core, more often around a ferromagnetic core that concentrates the field. When current flows through the wire it produces a magnetic field threading the coil's interior. That field stores energy in the same way a stretched spring stores it: invisibly, but recoverably.

The defining law is Faraday's. A changing current produces a changing magnetic flux, and that changing flux induces an opposing voltage:

V = L · (di/dt)

L (henries) is the proportionality constant — geometry-and-material in one number. For a solenoid of N turns, area A, length ℓ, with core relative permeability µᵣ:

L = µ₀ · µᵣ · N² · A / ℓ

The energy stored is the integral of v·i over time. Substitute v = L·di/dt:

W = ∫₀ᵗ v·i dt = ∫₀ᴵ L·i di = ½·L·I²

That energy lives in the magnetic field — specifically in the volume of space where B is non-zero, with energy density u = B²/(2µ). For a ferrite core operating at B = 0.25 T (just under saturation), u ≈ 25 kJ/m³ — about a thousand times what a high-voltage capacitor manages on a volume basis, but a thousand times worse than a battery.

Worked example: 100 µH carrying 5 A

A typical buck-converter inductor:

L = 100 µH = 100×10⁻⁶ H
I = 5 A
W = ½ · L · I²
  = 0.5 · 100×10⁻⁶ · 25
  = 1.25 × 10⁻³ J = 1.25 mJ

If the converter switches at 500 kHz, that 1.25 mJ is delivered to the load every cycle, equivalent to 1.25 mJ × 500,000 = 625 W of throughput. The inductor's job is not to store energy long-term but to bridge the on-time and off-time of the switch with current continuity.

Now flip the question. The same inductor sits in a flyback. The MOSFET interrupts 5 A in 100 ns:

V_spike = L · di/dt
        = 100×10⁻⁶ · (5 / 100×10⁻⁹)
        = 100×10⁻⁶ · 5×10⁷
        = 5000 V

5 kV across an unprotected MOSFET avalanche-clamps it instantly. The flyback diode or RCD snubber must conduct that current to a sink, dissipating ½LI² = 1.25 mJ as heat each cycle.

RL current curve

An inductor in series with a resistor R and a step voltage V₀ ramps current along an exponential curve, mirroring the capacitor's voltage curve:

i(t) = (V₀/R) · (1 − e^(-t/τ))     where τ = L/R

I/I_max
 1.0 ┤                          ╶─────── 100%
 0.99┤                    ╶─────              ← 5τ
     │              ╶─────
 0.86┤        ╶─────                          ← 2τ
     │     ╶─
 0.63┤   ╶─                                   ← 1τ
     │  ╱
 0.0 ┤─╱
     └──────────────────────────────── t/τ
     0    1    2    3    4    5    6

An RL filter has the dual behavior of an RC filter: voltage swaps for current, capacitance for inductance. The same 63.2% / 5τ engineering shorthand applies.

Inductor families compared

Core typePermeability µᵣQ at 100 kHzB_satLoss profileTypical role
Air core1200–500n/a (no sat)Pure copper lossRF tank circuits, antennas
Ferrite (MnZn)2000–500050–1500.30 THysteresis dominant 100 kHz–1 MHzSMPS chokes, EMI suppression
Ferrite (NiZn)10–150040–1000.20 TLow loss above 1 MHzRF transformers, baluns
Powdered iron5–10030–801.0–1.5 TSoft saturation, distributed gapDC-bias-tolerant chokes
Amorphous (Metglas)15,000–100,000100–3001.5 TVery low core lossCommon-mode chokes, EV inverters
Laminated silicon steel4000–10,0005–301.7–2.0 TEddy-current dominant >1 kHz50/60 Hz transformers, large chokes

The selection driver: pick the core whose saturation flux you can stay below at peak current, whose loss profile sits below your operating frequency, and whose permeability gives the L you need at acceptable copper turns. Ferrite dominates the 50 kHz–1 MHz switching converter band; powdered iron handles DC-bias-heavy chokes; amorphous wins where lowest core loss matters at any cost.

Real-world numbers

  • Transmission-line series reactor: 100s of µH per phase, carrying 1000s of amps to limit fault current. Energy in a 500 µH, 5000 A reactor is 6250 J — equivalent to a kilo of TNT in a single coil.
  • Smartphone buck inductor: 1 µH wirewound shielded chip, 1.5 A saturation, footprint 2 mm × 2 mm × 1 mm. Switches at 6 MHz to deliver 1 W per output rail.
  • EV traction motor stator: tens of mH effective at the controller switching frequency, 200 A peak. Field energy per PWM cycle is hundreds of joules, converted directly into shaft torque.
  • Tokamak central solenoid: 5–10 H, 40 kA. Field energy 8 GJ — a tonne of high explosive in a magnet that costs more than most office buildings.
  • RF coil for a 1 MHz tank: 25 µH air-core, Q ≈ 250. Drives ½LI² = 12.5 µJ at 1 A peak — develops 250× source voltage at the resonant cap.

Variants

  • Iron-core vs gapped-core. Putting an air gap in a ferromagnetic core lowers µ_eff and the inductance per turn, but pushes saturation flux higher relative to ampere-turns — letting you carry far more DC current before B reaches B_sat. Buck inductors and flyback transformers always have a gap.
  • Single-layer vs multi-layer windings. Single-layer (air-core RF) gives lowest interwinding capacitance and highest self-resonant frequency. Multi-layer (low-frequency choke) packs more turns and inductance into the same volume but rings at lower frequencies.
  • Common-mode vs differential-mode. A common-mode choke has both lines wound in the same direction on a single core: differential current produces canceling fields and sees ~zero impedance, while common-mode current sees the full µ_r·N². EMI filters need both flavors.
  • Coupled inductors (transformers). Two windings sharing a core couple energy from one circuit to another. Coupling coefficient k = M/√(L₁L₂), where M is mutual inductance. Tightly coupled (k → 1) is a transformer; loosely coupled (k ≈ 0.5) is a flyback or planar resonant converter.
  • Variable inductors. Slug-tuned (movable ferrite core), saturable reactors (DC-controlled gain), or MEMS variable inductors. Used historically in radio tuning; now mostly displaced by varactor-tuned LC tanks.

Failure modes

  • Saturation when DC bias exceeds rating. Once core flux reaches B_sat, inductance collapses. In a buck converter the slope di/dt = V/L blows up, peak current rockets past the MOSFET's SOA, and you smell magic smoke. Always check the I_sat curve, derate 30% for temperature.
  • Open winding from thermal cycling. Magnet wire fatigues at stress points where the lead enters the core. The inductor goes open, breaking the circuit silently.
  • Unprotected switch interruption. Open a switch in a current-carrying inductor without a freewheel path and V = L·di/dt rings to thousands of volts, arcing or destroying the switch. Every relay coil needs a flyback diode; every MOSFET driving an inductive load needs avalanche rating or an external clamp.
  • Audible coil whine. Magnetostriction vibrates the core at the switching frequency. Drift into the 1–20 kHz audible band under light load and the inductor sings. Fixed-frequency control or potting compound fix it.
  • Self-resonance. Interwinding capacitance forms a parasitic LC that resonates at SRF. Above SRF the part looks capacitive, not inductive. A 1 mH multilayer choke might have SRF of 200 kHz — useless above that.
  • Core loss exceeding heat budget. Hysteresis and eddy losses scale with B²·f^n. High B and f raise temperature, µᵣ falls toward the Curie point (200–250 °C for MnZn), and the cycle runs away thermally.

Where the energy storage matters

  • Switch-mode power supplies — the inductor bridges on/off cycles, transferring ½LI² each pulse.
  • Boost converters — energy is built up at low voltage, dumped at higher voltage when the switch opens.
  • Spark plugs and Tesla coils — abrupt interruption of primary current dumps coupled energy as a kilovolt arc.
  • Pulsed power and railguns — capacitor banks discharge into low-resistance rails forming an inductive loop, briefly storing megajoules in field.
  • Magnetic launch (maglev, mass drivers) — coils build field energy that propels conducting projectiles by Lenz reaction.
  • Wireless charging — primary inductor stores ½LI² each cycle; resonant coupling siphons that energy through the coupled secondary.

Frequently asked questions

Why does opening a switch on an inductor produce a huge voltage spike?

The inductor enforces continuity of current: V_L = L·di/dt. Force di/dt to be very large by opening the switch quickly and V_L scales with it. A 100 mH coil carrying 1 A whose current is interrupted in 1 µs spits out V = 0.1 × 1/1×10⁻⁶ = 100 kV — enough to arc across the contacts. Flyback diodes or snubber RCDs absorb that energy.

What is inductor saturation?

When the magnetic flux density B in the core exceeds the material's saturation value (typically 0.3 T for ferrite, 1.5 T for laminated iron), the relative permeability collapses. The inductance L falls toward the air-core value, sometimes by 10–100×. In a switching converter this means current ramps explode, MOSFETs cook, and you get visible smoke. Datasheets specify I_sat as the current at which L drops 10–30%.

Why is W = ½LI² and not LI²?

Energy is the integral of power over time: W = ∫v·i dt. With v = L·di/dt, that becomes W = ∫L·i di = ½Li². The factor of ½ is the same one that appears in ½CV² for capacitors and ½mv² for kinetic energy — it falls out of integrating a linear relationship.

What's the quality factor Q of an inductor?

Q = ωL/R, the ratio of reactive impedance to series resistance at frequency ω. High Q means low losses, sharp resonance, narrow filter bandwidth. Air-core RF coils reach Q of 200–500. Ferrite chokes drop to Q ≈ 30–100 because of core losses. Q peaks at one frequency and falls off because R rises with skin effect while L stays roughly flat.

Why does a transformer not store energy by design?

An ideal transformer transfers energy from primary to secondary on every cycle without storing it. In practice the magnetizing inductance does store a small amount of energy each half-cycle, which is wasted as core losses. Flyback transformers are the exception — they intentionally store energy in a gapped core during the on-time and dump it through the secondary during off-time, behaving more like coupled inductors than true transformers.

Are inductors interchangeable with capacitors in filters?

Roles are dual but components are not. Inductors block AC and pass DC; capacitors do the opposite. A series inductor in a low-pass filter is replaceable by a shunt capacitor only if you also flip topology (the dual network). In practice inductors are bigger, heavier, more expensive, and noisier than capacitors at low frequencies — which is why digital supply filters use capacitors-plus-resistors, not inductors-plus-capacitors, below ~1 MHz.