How to Select a Power Supply Inductor for Buck Converters: Sizing, Saturation, and Losses

To select a power supply inductor for a switching regulator, calculate the target inductance aiming for a peak-to-peak ripple current that is 20% to 40% of your maximum DC load current. The universal baseline formula is L = (Vout × (1 - D)) / (fsw × ΔIL). Crucially, the inductor's saturation current (Isat) rating must exceed your peak inductor current by at least 20% to prevent core saturation, which causes immediate, catastrophic overcurrent events in the switching MOSFETs. For a typical 5V/3A output, you are looking for a 4.7µH to 10µH shielded ferrite inductor with an Isat rating above 5A.

Topology Context: Linear vs. Switching for High-Current Loads

Before sizing magnetics, we must justify the switching topology over a linear regulator (LDO). When stepping down a 12V nominal rail to 5V at 3A, the choice between linear and switching dictates your entire thermal and PCB layout strategy.

An LDO operates by burning excess voltage as heat. The dropout and headroom math for a linear regulator in this scenario is brutal: Power Dissipation = (Vin - Vout) × Iout. At 12V in and 5V out at 3A, an LDO will dissipate 21W of heat. This requires massive heatsinks and is practically impossible on a standard 2oz copper PCB without thermal shutdown. Conversely, a synchronous buck converter operates at roughly 90% efficiency. Total power loss is scattered across the high-side MOSFET, low-side MOSFET, and the power supply inductor, totaling roughly 1.6W of manageable heat.

Because of the thermal reality, high-current DC-DC conversion demands a switching topology, making the power supply inductor the most critical passive component in your BOM.

Design Example: Sizing a Power Supply Inductor for a 12V to 5V/3A Buck

Let us walk through a concrete design example. We are building a solar charge controller auxiliary rail that takes a 12V nominal battery input and steps it down to 5V at 3A to power an ESP32 and a relay driver.

Input Range and Protection Requirements

A 12V lead-acid or LiFePO4 battery is rarely exactly 12V. The input range will swing from 9.0V (deep discharge/cranking) to 14.4V (absorption charge). Furthermore, automotive and solar environments suffer from inductive kickback and load-dump transients. Your input protection must include a bidirectional TVS diode (like the SMAJ24A) clamping at 24V, and input ceramic capacitors rated for at least 35V (never use 16V or 25V caps on a 24V clamped rail, as DC bias derating will crush their effective capacitance).

Inductance and Saturation Math

We will use a modern synchronous buck IC like the Texas Instruments TPS54308, switching at fsw = 500 kHz.

• Duty Cycle (D): Vout / Vin(nom) = 5 / 12 = 0.416
• Target Ripple (ΔIL): 30% of 3A = 0.9A peak-to-peak
• Inductance (L): (5V × (1 - 0.416)) / (500,000 Hz × 0.9A) = 6.48 µH

We select the nearest standard value: 6.8 µH.

Next, we calculate the peak current the inductor will experience to determine the saturation current (Isat) requirement:

• Peak Current (Ipeak): Iout(max) + (ΔIL / 2) = 3A + 0.45A = 3.45A
• Minimum Isat: 3.45A × 1.20 (20% safety margin) = 4.14A

Ripple and Noise Expectations

Output voltage ripple is primarily dictated by the inductor's ripple current interacting with the output capacitor's Equivalent Series Resistance (ESR). If we use a low-ESR 22µF X7R ceramic capacitor with an ESR of 3mΩ, the resistive ripple is ΔVout = 0.9A × 0.003Ω = 2.7mV. Adding the capacitive ripple component yields a total expected output noise of roughly 8mV to 12mV peak-to-peak, which is well within the tolerance of the ESP32 and logic circuits.

Thermal Derating and Core Losses in High-Ambient Environments

A common mistake on the bench is reading the Isat value on the inductor datasheet at 20°C and assuming it holds true inside a sealed solar enclosure at 70°C ambient. Ferrite core materials lose permeability as temperature rises. By the time the core reaches 100°C, the saturation current of many standard ferrite blends drops by 20% to 30%. This is exactly why we specified the Coilcraft XEL6060-682 with an 8.4A Isat for a 3.45A peak load; that massive margin ensures the inductor will not saturate even when baking in a high-ambient environment.

Furthermore, inductor losses are split into two categories:

1. Copper Loss (I²R): The DC resistance (DCR) of the copper windings. Copper has a positive temperature coefficient; as the inductor heats up, DCR increases, generating even more heat. At 100°C, DCR is roughly 30% higher than at 20°C.
2. Core Loss: Hysteresis and eddy current losses in the ferrite material, governed by the Steinmetz equation. Core losses scale non-linearly with switching frequency and AC ripple current. If you push a 500 kHz optimized inductor at 2 MHz, core losses will spike, potentially causing thermal runaway even if the DC load is light.

For deeper analysis of magnetics behavior and core loss modeling, reference the Analog Devices inductor selection guidelines and the Coilcraft power inductor learning center, which provide excellent empirical data on temperature derating curves.

Power Supply Inductor FAQ