How Does an Inductor Work in Power and RF Circuits? A Bench-Side Deep Dive

An inductor stores energy in a magnetic field when electrical current flows through it, fundamentally resisting any sudden changes in that current. While capacitors fight voltage changes, inductors fight current changes. Whether you are designing a high-frequency RF filter or a heavy-current DC-DC buck converter, understanding the physical limits of your magnetics is the difference between a stable power rail and a melted switching IC.

The Core Physics: How Does an Inductor Work?

At the bench level, the governing equation for an inductor is V = L(di/dt). This means the voltage across the component is directly proportional to its inductance (L) and the rate at which current changes through it (di/dt). If you try to stop current flow instantly (dt approaches zero), the inductor will generate a massive voltage spike to keep the current moving, often resulting in an arc or a blown semiconductor.

Think of an inductor like a heavy, mechanical water wheel in a pipe. When you open the valve, the water (current) doesn't flow at full speed immediately; the heavy wheel (magnetic field) takes time to spin up, resisting the initial surge. Once the wheel is spinning at full speed, it maintains a steady flow. If you suddenly slam the valve shut, the momentum of the heavy wheel forces water past the closed valve, creating a massive pressure spike (voltage spike) that can rupture the pipe.

Let's look at a concrete numeric example. Suppose you have a 10µH inductor in a switching circuit, and the MOSFET turns off, forcing the current to drop by 2 Amps in just 1 microsecond (1µs).

• di = 2A
• dt = 0.000001s
• V = 10 × 10^-6 H × (2 / 0.000001) = 20 Volts

That 20V spike adds directly to your supply rail. If your MOSFET is only rated for 20V V_DS and you are running a 12V rail, that 32V total spike will punch through the silicon die. This is exactly why we use flyback diodes or snubber networks in inductive circuits.

Decoding the Datasheet: Types, Markings, and Selection

Not all inductors are created equal. A part that works perfectly in an audio crossover will saturate and fail instantly in a 2MHz switching regulator. Here is how to select the right construction for your specific job.

How to Read the Markings

Surface mount (SMD) power inductors typically use a three-digit code, similar to resistors, but the base unit is microhenries (µH), not ohms.

• 100 = 10 × 10⁰ = 10µH
• 471 = 47 × 10¹ = 470µH
• 152 = 15 × 10² = 1500µH (1.5mH)

For values under 10µH, manufacturers use an 'R' to denote the decimal point. 4R7 means 4.7µH, and R47 means 0.47µH. Molded axial inductors use standard 4-band resistor color codes, but the multiplier yields µH, and the fourth band is usually tolerance (Gold = ±5%, Silver = ±10%).

Bench Scenario: When a 33µH Power Inductor Fails

To understand why datasheet parameters matter, let's walk through a real-world bench failure involving a classic LM2596 buck converter.

The Setup: We needed to step down a 12V automotive battery rail to a 5V rail capable of delivering 3 Amps to power a Raspberry Pi cluster and some peripheral sensors. The LM2596-5.0 was chosen. The datasheet calls for a 33µH inductor.

The Numbers: I grabbed a cheap, unshielded radial-leaded 33µH inductor from a bulk bin. It had the right inductance value, verified on an LCR meter. However, its saturation current (I_sat) was only rated for 1.2A, and its RMS current (I_rms)—the thermal limit—was 1.5A. The circuit required an inductor with an I_sat of at least 4.5A (to handle the 3A load plus the ~30% ripple current without saturating).

The Outcome: With a 100mA dummy load, the 5V rail was perfectly clean. When I switched to the actual 2.8A cluster load, the output voltage immediately sagged to 3.1V. Within 15 seconds, the LM2596 case grew too hot to touch, and the chip's internal thermal shutdown began hiccuping the output. Eventually, the internal switch MOSFET shorted out, killing the chip.

What Went Wrong: Core saturation. When the current exceeded 1.2A, the magnetic domains in the cheap ferrite core aligned completely. The inductor effectively turned into a straight piece of wire (inductance dropped from 33µH to < 0.5µH). With almost zero inductance to limit the di/dt, the peak current spiked to over 10 Amps per switching cycle. The LM2596's internal current limit couldn't react fast enough, leading to thermal runaway and silicon death. Always check I_sat and I_rms, not just the µH value.

Failure Modes and Visual Autopsy

When an inductor fails, it rarely does so silently. Here is how to diagnose them on the bench based on visual and electrical symptoms.

1. Open Circuit (Thermal Burnout):
   • Visual Symptom: Blackened, blistered epoxy coating. The part smells distinctly of burnt phenolic resin or ozone.
   • Cause: Exceeding the I_rms (heating) rating. The copper wire's I²R losses generated more heat than the core could dissipate, melting the internal winding until it snapped.
2. Shorted Turns (Insulation Breakdown):
   • Visual Symptom: Looks physically perfect. However, measuring the DC Resistance (DCR) with a milliohm meter shows a value 20-50% lower than the datasheet spec.
   • Cause: High voltage spikes (exceeding the dielectric strength of the thin enamel coating on the copper wire) caused micro-arcs between adjacent windings, bypassing sections of the coil. The inductance drops, and DCR plummets.
3. Core Fracture (Mechanical/Thermal Stress):
   • Visual Symptom: A hairline crack visible on the ferrite core under a 10x loupe. Under load, the part emits an audible, high-pitched clicking or buzzing.
   • Cause: Ferrite is essentially ceramic and highly brittle. Rapid thermal cycling or physical board flexion cracks the core. This introduces an unintended air gap, drastically altering the inductance and causing magnetostriction noise.

The Substitution Matrix: Swapping Parts Safely

Supply chain shortages often force engineers to substitute magnetics. According to magnetics design guidelines from sources like Coilcraft and general component theory from All About Circuits, you cannot simply swap parts based on the µH stamp. Follow this numbered decision path to substitute safely:

1. Match Inductance (L): For power converters, a ±20% variance is usually acceptable because the control loop will adjust the duty cycle to compensate. For RF filters and oscillators, you must match within ±2% to ±5%, or your resonant frequency will shift out of band.
2. Verify Saturation Current (I_sat): The substitute's I_sat must be greater than or equal to the original part. Calculate your peak current: I_peak = I_load + (ΔI_ripple / 2). If the substitute's I_sat is lower than I_peak, reject it immediately.
3. Verify Thermal Current (I_rms): This dictates the temperature rise (usually specified as a 40°C rise). The substitute's I_rms must exceed your maximum continuous DC load. If it is lower, the part will overheat and desolder itself from the pad.
4. Check DC Resistance (DCR): Lower DCR is always better for efficiency. If the substitute has a higher DCR, calculate the new I²R loss to ensure your thermal budget isn't exceeded.
5. Confirm Shielding and Footprint: Do not swap a shielded SMD part for an unshielded one if the inductor sits within 10mm of a sensitive feedback trace, an ADC, or an audio codec. The radiated EMI will corrupt your signals.

By treating inductors as complex, non-ideal components rather than simple wire coils, you eliminate the most common points of failure in both power delivery and signal conditioning networks. Always read the saturation curves in the datasheet before you place the part on your BOM.