Designing Output Inductors for Bidirectional DC-DC Converters

An output inductor for a bidirectional DC-DC converter must store and release energy predictably in both power-flow directions. The essential inputs are inductance under DC bias, ripple current, peak current, switching frequency, core loss, copper loss, temperature rise, acoustic behavior, insulation, and fault exposure. A nominal inductance measured at zero current is not enough.

Bidirectional converters are used in battery storage, EV charging, regenerative drives, DC microgrids, and backup power. The hardware may operate as a buck converter in one direction and a boost converter in the other. Those modes often produce different duty cycles and ripple even when average power is similar.

Map both power-flow directions

For each direction, calculate minimum and maximum bus voltage, switching-frequency range, duty cycle, average current, peak current, and ripple current. Include startup, current limiting, pre-charge, light-load operation, and phase shedding. The magnetic worst case may occur at low bus voltage, while the thermal worst case appears at another operating point.

In an interleaved converter, ideal cancellation reduces terminal ripple, but each phase inductor still carries its own DC current and ripple. Control tolerance and inductance variation can create current imbalance, so the component requirement should not assume perfect sharing.

Specify inductance under DC bias

Powder cores, discretely gapped ferrites, and distributed-gap materials have different permeability-versus-bias behavior. State the minimum inductance at a defined DC current and measurement condition. A gradual reduction may be acceptable if the control loop and ripple limit allow it. A sharp drop near saturation is dangerous because current can accelerate before protection reacts.

Separate copper and core losses

DC copper loss depends on winding resistance at operating temperature. AC copper loss depends on ripple frequency, conductor size, layer arrangement, proximity effect, and fringing field. Core loss depends on flux swing, waveform, frequency, temperature, material, and gap. Catalog sinusoidal data can misrepresent converter waveforms, so a carefully instrumented temperature-rise test is often the most practical system-level check.

Watch gap fringing and acoustics

A concentrated gap produces fringing flux that can heat nearby turns, brackets, shields, or PCB copper. Winding placement, gap distribution, insulation, and mounting hardware should be reviewed together. Ripple forces can also move windings or produce audible tones. Varnish, core clamping, adhesive, and mounting influence noise, but the specification should define operating modes and measurement conditions rather than simply request a “quiet” part.

Prototype checks

  • Inductance at zero bias and at specified DC current
  • Incremental inductance near maximum credible current
  • Winding resistance corrected to the agreed temperature
  • Temperature rise in both power-flow directions
  • Current waveform during reversal, startup, and fault limiting
  • Core, winding, terminal, and nearby PCB hot spots
  • Audible noise across switching and modulation modes

When discussing a custom inductor with BaoHui Tech, provide operating tables or waveforms for both directions. As a transformer manufacturer and magnetic component supplier, BaoHui Tech can translate those points into a controlled core, winding, gap, terminal, and test specification.

Frequently asked questions

Can the same inductance value be used in both directions?

The physical component is non-polar, but each mode can impose different current, ripple, duty cycle, and temperature. Both modes need independent verification.

What saturation value is useful?

Use inductance behavior at and beyond the maximum credible current under a stated condition. A saturation-current number without its inductance-drop criterion is ambiguous.

Why can the inductor overheat below rated current?

Ripple, AC winding loss, core loss, gap fringing, restricted airflow, or a hot enclosure can dominate temperature rise. Average current alone does not describe these losses.