What Do Inductors Do in a Circuit? The Core Physics

At the most fundamental level, an inductor resists changes in current. While a capacitor resists changes in voltage by storing energy in an electric field, an inductor stores energy in a magnetic field. The governing equation is V = L(di/dt). If the current through an inductor tries to change rapidly, the inductor generates a voltage spike that opposes that change. If the current is steady DC, the inductor acts as a simple piece of wire (limited only by its DC Resistance, or DCR).

Think of an inductor like a heavy water wheel in a pipe. When you first open the valve, the wheel's inertia prevents water (current) from flowing instantly. Once the wheel is spinning, if you suddenly close the valve, the wheel's momentum forces water to keep moving, potentially bursting the pipe (a voltage spike). This inertia is exactly what makes inductors invaluable for smoothing out choppy, pulsing currents from switching power supplies.

Topology Deep-Dive: The LC Low-Pass Power Filter

To see this physics in action, we will design an LC low-pass filter to clean up a noisy 5V DC rail generated by a buck converter before it reaches a sensitive 12-bit ADC. We use an LC topology rather than an RC (Resistor-Capacitor) filter because an LC filter is largely lossless. An RC filter would drop significant DC voltage and waste power as heat; an LC filter drops almost zero DC voltage and stores/returns energy.

Topology Nodes and Layout

  • Node 1 ($V_{IN}$): The noisy 5V source (e.g., output of an LM2596 buck converter switching at 150kHz).
  • Node 2 ($V_{MID}$): The junction between the inductor and the capacitor.
  • Node 3 ($V_{OUT}$): The clean 5V output connected to the sensitive load.
  • Node 4 ($GND$): The common ground plane.

The inductor (L1) is placed in series between Node 1 and Node 2. The capacitor (C1) is placed in parallel between Node 2 and Node 4. Node 3 is tied directly to Node 2. High-frequency AC ripple from Node 1 is blocked by the inductor's impedance ($X_L = 2\pi fL$), while the capacitor provides a low-impedance shunt path ($X_C = 1 / 2\pi fC$) to ground for any residual noise that makes it through.

Component Behavior & Failure Modes at the Extremes

Understanding how the circuit reacts to component drift or catastrophic failure is critical for robust design. Below is the behavior matrix for our LC topology.

Element Change / Fault Effect on Cutoff Frequency ($f_c$) Effect on DC Output / Load System-Level Consequence
Inductance (L) Increases Decreases (better filtering) No change (ideal) Risk of core saturation if DC current exceeds rating; physical size increases.
Capacitance (C) Increases Decreases (better filtering) No change Higher inrush current at startup; potential for converter instability.
L1 Shorts (0 Ω) N/A (Filter defeated) $V_{OUT}$ = Noisy $V_{IN}$ Load receives full switching ripple; ADC readings become erratic.
L1 Opens (∞ Ω) N/A $V_{OUT}$ = 0V Complete power loss to the load. System halts.
C1 Shorts (0 Ω) N/A $V_{OUT}$ = 0V Massive current draw from $V_{IN}$; upstream fuse blows or buck converter trips OCP.
C1 Opens (∞ Ω) N/A (Filter defeated) $V_{OUT}$ = Noisy $V_{IN}$ Loss of AC shunt path; high-frequency noise passes directly to the load.
Bench Tip: Ceramic capacitors (MLCCs) are highly susceptible to cracking from board flex, which usually results in a short circuit (C1 Shorts). If your upstream supply lacks overcurrent protection, a cracked MLCC can melt your PCB traces. Always use soft-termination MLCCs in high-vibration environments.

Design Walkthrough: Sizing a 5V Rail LC Filter

Let's pick real component values to filter out 500kHz switching noise from a modern DC-DC converter. Our target load is an ESP32-S3 module and an external ADC drawing a combined maximum of 800mA. We want a cutoff frequency ($f_c$) well below the 500kHz switching frequency, targeting roughly 10kHz to 15kHz.

1. Selecting the Inductor

We need an inductor that can handle 800mA of continuous DC current without saturating, and we want a shielded part to prevent magnetic coupling into nearby analog traces.
The Pick: Würth Elektronik 74477420 (WE-PD series).
Specs: 10 µH, shielded, 2.2 A saturation current ($I_{SAT}$), 45 mΩ DCR.
At 800mA, the DC voltage drop is merely $V = I \times R = 0.8A \times 0.045\Omega = 36mV$. Power wasted is just 28mW.

2. Selecting the Capacitor

Using the cutoff frequency formula $f_c = \frac{1}{2\pi\sqrt{LC}}$, we solve for C to hit ~10kHz with our 10µH inductor.
$C = \frac{1}{(2\pi \times 10,000)^2 \times 10\mu H} \approx 25.3\mu F$.
We will use a standard 22 µF value, which yields an actual $f_c$ of 10.7 kHz.
The Pick: Murata GRM21BR61E226ME44 (22 µF, 0805 package, X7R dielectric, 25V rating).
Crucial Detail: We specify X7R over X5R, and a 25V rating for a 5V rail, because MLCCs suffer from DC bias derating. A 6.3V-rated 22µF X5R cap might lose 50% of its capacitance at 5V DC bias. A 25V X7R part retains nearly its full rated value.

Breadboard Testing & Verification Protocol

Do not trust simulation blindly; parasitic elements on a breadboard will alter high-frequency behavior. Follow this exact sequence to verify the filter on the bench.

  1. Prep the Board: Place the inductor and capacitor as physically close together as possible. Use thick, short jumper wires for the power and ground rails to minimize parasitic series inductance.
  2. Inject the Noise: Power Node 1 with your noisy 5V source. Connect an electronic load or a power resistor to Node 3 to draw a realistic 500mA test current.
  3. Probe $V_{IN}$ (Node 1): Use an oscilloscope with AC coupling and a coaxial tip or ground spring. Do not use the standard 6-inch alligator ground clip; its inductance will ring with the circuit capacitance and show fake high-frequency spikes. Measure the peak-to-peak ripple.
  4. Probe $V_{OUT}$ (Node 3): Move the probe to the output side. Keep the ground spring connected to the exact same physical ground point. You should see the 500kHz ripple attenuated by at least -30dB (a factor of ~30x reduction in amplitude).
  5. Load Transient Test: Switch the electronic load between 100mA and 800mA at 1kHz. Observe the voltage dip at $V_{OUT}$. The inductor will limit the slew rate of the current, causing a slight, controlled voltage sag rather than a sharp collapse.

Decision Tree: Inductor, Ferrite Bead, or Resistor?

When designing a power filter, engineers often default to the wrong component. Use this decision matrix to select the correct filtering element for your specific load and noise profile.

Load Current Noise Frequency Profile Recommended Topology Concrete Default Pick (for 5V Rail)
< 50 mA Low Frequency (< 10 kHz) RC Filter (Resistor + Cap) 10Ω 0805 Resistor + 10µF MLCC
> 50 mA Very High Freq EMI (> 50 MHz) Ferrite Bead + Cap Laird 28B0205-000 (Ferrite) + 100nF MLCC
> 50 mA Switching Ripple (10 kHz - 5 MHz) LC Filter (Inductor + Cap) Würth 74477420 (10µH) + 22µF X7R MLCC

The Final Verdict: If you are filtering the output of a DC-DC switching regulator to feed a microcontroller, sensor, or RF module drawing more than 50mA, the LC Filter using the Würth 74477420 and a 25V-rated X7R MLCC is your definitive, lossless default. It provides the necessary energy storage to handle load transients without dropping your DC voltage or burning watts as heat, solving the exact limitations of resistive and ferrite-based alternatives.