The moment you open a switch on an inductive load, the magnetic field collapses and demands a path. If you don't provide one, the voltage will spike until it finds a path through your silicon, your insulation, or the air itself. Understanding the exact energy stored in an inductor is the difference between a reliable switching power supply and a bench covered in shattered MOSFETs and melted diodes.
The governing equation is deceptively simple: E = ½ × L × I². Energy (E) in Joules equals half the inductance (L) in Henries multiplied by the square of the current (I) in Amps. Notice that current is squared. Doubling your inductance doubles the stored energy, but doubling your current quadruples it. This non-linear relationship is why high-current, low-inductance chokes in modern switching regulators can still pack enough punch to destroy undersized clamp diodes.
Inductor Types: Which Core for Which Job?
Not all inductors are created equal. The core material dictates the saturation current, the temperature coefficient, and the electromagnetic interference (EMI) profile. Here is how the four main physical constructions stack up on the bench.
| Type | Construction | Typical Tolerance | Tempco (ppm/°C) | Typical Use Case |
|---|---|---|---|---|
| Molded / Shielded | Ferrite powder suspended in resin, fully enclosed | ±10% to ±20% | ~100 | DC-DC buck/boost converters, high-density SMD boards where EMI must be contained. |
| Unshielded Drum | Ferrite bobbin with exposed copper windings | ±10% to ±30% | Varies widely | Low-cost input filtering, non-critical power rails, hobbyist breadboarding. |
| Toroidal | Tape-wound or ferrite ring, hand or machine wound | ±10% to ±20% | Low (core dependent) | High-current EMI filtering, audio crossovers, linear power supply chokes. |
| Air Core | Copper coil with no magnetic core material | ±1% to ±5% | ~3900 (copper expansion) | RF circuits, high-frequency resonant tanks, applications where core saturation is unacceptable. |
Selection Rule of Thumb: Use shielded molded inductors (like the Würth WE-PD or Coilcraft MSS series) for any switching node within 2 inches of a sensitive analog or RF circuit. Use air-core only when operating above 10 MHz or when you absolutely cannot tolerate the non-linear inductance drop caused by magnetic saturation.
Decoding the Markings: What the Codes on the Physical Part Mean
Unlike resistors, which have largely standardized on the EIA-96 or standard 3-digit codes, inductor markings can be a frustrating mix of legacy and modern conventions. Here is how to read the SMD and through-hole parts in your bin.
The 3-Digit SMD Code
The first two digits are the significant figures, and the third digit is the multiplier (number of zeros), with the base unit being microhenries (µH).
- 100 = 10 × 10⁰ = 10 µH (Not 100 µH!)
- 101 = 10 × 10¹ = 100 µH
- 472 = 47 × 10² = 4700 µH (or 4.7 mH)
The 'R' Decimal Notation
For values under 10 µH, the letter 'R' replaces the decimal point.
- 4R7 = 4.7 µH
- R10 = 0.10 µH
Color Bands (Axial RF Chokes)
Older or specialized axial inductors use the standard resistor color code, but the base unit is microhenries. A brown-black-brown-silver band translates to 10 × 10¹ µH = 100 µH with a 10% tolerance. Always verify with an LCR meter; relying on faded paint on a 30-year-old surplus part is a fast track to a failed prototype.
Bench War Story: When 14 Millijoules Fries a Flyback Diode
Theory is clean; the bench is messy. Let’s look at a real-world scenario where miscalculating the energy stored in an inductor led to a catastrophic component failure.
The Setup
A hobbyist was building an Arduino-controlled irrigation system. The load was a standard 12V DC solenoid valve. The control circuit used an N-channel MOSFET to switch the low side of the solenoid, with a flyback diode placed in reverse-bias across the solenoid coil to clamp the inductive spike when the MOSFET turned off. For the diode, they grabbed a 1N4148 small-signal diode from their bench bin.
The Numbers
The solenoid coil had a DC resistance of 50Ω and a measured inductance of 500 mH (0.5 H).
Steady-state current: I = V / R = 12V / 50Ω = 0.24 A.
Let's calculate the energy stored in the inductor at the moment the MOSFET switches off:
E = ½ × L × I²
E = 0.5 × 0.5 H × (0.24 A)²
E = 0.0144 Joules (14.4 mJ)
The Outcome
The first time the Arduino commanded the valve to close, there was a sharp crack. The 1N4148 diode shattered, leaving a black scorch mark on the FR4 board, and the MOSFET shorted drain-to-source.
What Went Wrong
14.4 mJ sounds like a tiny amount of energy—barely enough to heat a drop of water. But look at the component datasheets. The physics of inductive decay dictates that the current must freewheel through the diode until the energy is dissipated as heat in the diode's forward voltage drop and the coil's resistance. The decay time constant (τ = L/R) here is 10 milliseconds.
The 1N4148 is a small-signal diode with a tiny silicon die. It is rated for 200mA continuous current and can handle brief nanosecond surges, but it cannot absorb 14.4 mJ of thermal energy over a 10ms decay window without exceeding its junction temperature limit. The die experienced localized thermal runaway, shorted, and subsequently exposed the MOSFET to the full unclamped flyback spike.
The Fix: Swap the 1N4148 for a 1N4004. The 1N4004 has a massive silicon die designed to handle 30A half-cycle surges and easily absorbs the 14.4 mJ thermal load. For even faster turn-off times in high-speed PWM circuits, a bidirectional TVS diode or a Zener-clamp network is required.
Failure Modes and Visual Symptoms
Inductors are generally robust, but they do fail. Recognizing the symptoms saves hours of oscilloscope debugging. According to magnetics design guidelines, these are the primary failure vectors:
- Core Saturation (Invisible but deadly): If the current exceeds the saturation current (Isat), the core's permeability drops to near that of air. The inductance plummets, and the current spikes exponentially. Symptom: No visual damage to the inductor, but your switching MOSFET runs hot or explodes. Visible only on a current probe as a sharp upward 'kink' in the current ramp.
- Thermal Overload (I²R Heating): Exceeding the RMS current rating (Irms) causes the copper windings to overheat. Symptom: A distinct acrid smell of burning enamel. The heat-shrink sleeve (if present) will melt or discolor, and the copper wire may turn black or snap at the termination pad.
- Mechanical Cracking: Ferrite is essentially ceramic. Dropping a toroidal or drum-core inductor on a hard bench can cause micro-fractures in the core. Symptom: An audible rattle when shaken. Electrically, the air gap introduced by the crack drastically lowers the inductance and increases EMI leakage.
- Insulation Breakdown: In high-voltage or high-dV/dt circuits, the thin enamel coating on adjacent windings can arc over. Symptom: Pitting or tiny black craters visible under magnification on the outer layer of the windings, often accompanied by a drop in DC resistance (DCR) due to shorted turns.
Safe Substitution: What to Do When the Exact Part is Missing
You're prototyping on a Friday night, and the exact 4.7µH shielded inductor specified in the TI reference design is out of stock. Can you substitute? Follow these strict rules to avoid blowing up your prototype:
1. Never Substitute a Lower Saturation Current (Isat)
You can always use an inductor with a higher Isat rating. If the original calls for 2.5A saturation, a 4A part is fine (though physically larger). If you drop to 1.5A, your regulator will likely hit overcurrent protection or destroy the switch node under heavy load transients.
2. Shielded Can Replace Unshielded (But Not Vice Versa)
If the BOM calls for an unshielded drum core and you only have a shielded molded part of the same value and current rating, use it. The magnetic flux is contained, which is strictly better for EMI. However, replacing a shielded part with an unshielded one near a sensitive feedback trace will inject switching noise directly into your control loop, causing jitter or instability.
3. Mind the DCR (DC Resistance) Trade-off
Substituting a physically larger inductor usually means thicker wire and lower DCR. This is generally good for efficiency. However, in some specific control loops (like certain peak-current-mode controllers), a minimum ESR/DCR is required to provide the leading-edge ramp signal. If your substitute has near-zero DCR, you may need to add a tiny series resistor (e.g., 0.05Ω) or rely on the controller's internal slope compensation.
4. Inductance Value Tolerance
In power supply filtering and DC-DC conversion, going up 10-20% in inductance (e.g., using 5.6µH instead of 4.7µH) is usually safe and will slightly reduce output ripple current, though it may degrade transient response. In RF matching networks or resonant converters, a 5% shift will detune the circuit entirely. Stick to tight-tolerance (±1% or ±2%) air-core or ceramic multilayer parts for RF.
