Custom Magnetics Design: Practical Engineering, Thermal, and Sourcing Guide

A custom magnetic program almost always starts with a constraint that no catalogue part satisfies. A 200 kHz LLC stage needs 22 uH at 6 A of DC bias, the board leaves 4 mm of height above the surface, and the shutdown threshold sits at 118 C. Stock inductors inside that envelope run hot, saturate early, or fail the creepage requirement. Custom magnetics design is the work of turning those numbers into a wound component that holds its inductance, stays inside its thermal budget, and can be built again next quarter with the same result.

What Custom Magnetics Design Actually Means

Custom magnetics design is the process of translating a customer-specific electrical, thermal, and mechanical requirement into a magnetic component built for that requirement alone, rather than selecting the nearest catalogue part and accepting its compromises.

The scope reaches further than most buyers expect. A single custom program touches several independent decisions at once:

• Core material grade, including initial permeability, saturation flux density, and core loss at the operating frequency
• Core shape and air gap, which set the inductance value, the DC bias behaviour, and the leakage inductance
• Winding window usage, wire gauge, and interleaving, which drive copper loss and proximity effects
• Termination style, pin out, and mounting geometry, which decide whether the part fits the board at all
• Insulation system, creepage, and clearance, which determine whether the design clears safety review

Each of those choices trades against the others. A wider gap raises the saturation margin but increases fringing loss near the winding. Finer wire lowers skin effect loss but fills the bobbin faster and pushes the build height up. The design task is finding the combination that clears every constraint, not maximising any single figure of merit.

When Off-the-Shelf Parts Stop Working

Off-the-shelf magnetics stop working when several constraints tighten at the same time, not when one of them does.

A single tight specification is usually solvable from a catalogue. Two are awkward. Three or more, and the selection table runs out of rows. The failures tend to fall into a handful of recognisable patterns:

• Inductance collapses under DC bias because the core grade has limited superposition performance at the rated current
• Winding temperature climbs past the insulation class rating once core loss and copper loss are added together
• The available footprint forces a core size that cannot physically hold the required turns of the required wire
• Creepage and clearance fail the safety standard after the pin out is fixed by the board layout
• Radiated noise from leakage inductance cannot be filtered without adding components the enclosure has no room for

Core Material Choice Drives the Whole Design

Core material is the first decision in custom magnetics design, because it fixes the loss budget, the bias behaviour, and the usable frequency window before any winding geometry is considered.

Manganese zinc ferrite remains the default for power conversion between roughly 20 kHz and a few hundred kilohertz. Within that family, grades diverge sharply. A low-loss power grade keeps core loss down at elevated flux density, which suits flyback and forward transformers running hard. A high-conductivity grade pushes resistivity up so eddy currents stay small, which matters as switching frequency climbs and AC resistance starts to dominate total loss.

The loss minimum matters more than the datasheet headline. A design that runs the core at 40 C pays a penalty that a design running at 90 C does not, even with identical material. If the enclosure cannot hold the core near its loss minimum, the material choice has to compensate, often by moving to a lower-loss grade or accepting a larger core.

Nickel zinc ferrite enters the picture above a few megahertz, where MnZn resistivity is no longer sufficient. A practical comparison of MnZn and NiZn core types helps, but the short version is that NiZn trades saturation flux density for frequency headroom. Below 1 MHz it usually loses on both loss and cost.

The Design Loop: From Requirement to Wound Component

A custom magnetics program moves through five stages, and each stage can send the design back to the one before it.

Designs that stop after stage three tend to reappear as field returns, because the bias sweep and the thermal soak reveal problems that a single-point inductance measurement cannot.

What to Put in a Custom Magnetics RFQ

A complete RFQ states the operating conditions, not just the target inductance, because the same inductance value can be built in several ways with very different behaviour.

Sourcing Risk and Quality Signals

The quality risk in a custom magnetics program sits mostly in process control, not in the design itself.

Two suppliers can build the same drawing and deliver very different parts, because winding tension, gap grinding tolerance, and sintering profile control all shift the final inductance distribution. When a program is going into volume, the questions worth asking are about the process, not the datasheet:

• Which management system certifications are actually held, and for which product lines
• Whether the supplier manufactures the powder, the core, and the wound component, or buys one of those stages in
• How inductance is verified in production, by sampling or by full test
• What the change notification process looks like when a core grade or a wire supplier changes

A vertically integrated supplier that produces ferrite powder, sintered cores, and finished transformers under one roof has fewer handover points where a material substitution can go unnoticed. That matters most for programs in automotive or industrial power, where an out-of-tolerance inductor is not a cosmetic issue.

For programs moving from prototype to volume, it is worth requesting a design review before the drawing is frozen, because a gap or wire change is cheap at prototype stage and expensive after qualification.

Frequently Asked Questions

How long does a custom magnetics design typically take?

From a complete requirement set, a first prototype usually takes a few weeks, with the bulk of that time going into material selection and winding trials rather than the drawing itself. Programs that stall are almost always missing a bias current figure or a temperature limit, which forces the design loop to restart.

Can a custom magnetic always beat a catalogue part on cost?

Not at low volume. Custom tooling, gap grinding setup, and qualification effort carry fixed costs that a catalogue part spreads across thousands of customers. The economics shift once the annual volume is high enough, or once a catalogue part forces extra filtering, extra heatsinking, or a larger enclosure.

Is a gapped core always better for DC bias performance?

Usually, but not unconditionally. A gap raises the saturation margin and flattens the inductance versus current curve, which is what a switching converter wants. It also introduces fringing flux near the gap, which adds loss in the adjacent winding turns unless the winding is spaced away or the gap is distributed across the core.

What single measurement best predicts whether a design will survive?

Temperature rise at full load with the final enclosure in place. Inductance, saturation, and loss all influence it, and it is the one number that reflects how the magnetic behaves in the real thermal environment rather than on an open bench.