Magnetics & Transformer Design Optimizer
NSGA-II optimizer finds Pareto-optimal core and winding designs across 40 cores (E, ETD, PQ, RM and toroids) and 104 core/material combinations, with TDK and Ferroxcube ferrites and Micrometals iron powder. Minimizes total loss and volume simultaneously — shows the full efficiency-vs-size trade-off frontier.
How It Works
Designing a transformer or inductor by hand means picking one core from hundreds and hoping the efficiency–size trade-off is acceptable. This tool runs NSGA-II across 40 cores in 104 core/material combinations simultaneously, evaluating every candidate at 10 operating points (5 load levels × 2 input voltage extremes) for robust worst-case design.
Two objectives are minimised simultaneously: total power loss (core loss + Dowell AC winding loss) and core volume. These objectives conflict — a smaller core runs at higher flux density and higher loss. The result is a Pareto front: every design where no efficiency improvement is possible without increasing size.
Core loss is fitted to the manufacturers' published loss data — Steinmetz fits to TDK and Ferroxcube data sheet points, and Micrometals' own formula for iron powder — and evaluated at the AC flux amplitude, not the peak: half the flux swing for a flyback, Vin·D/(2·N·Ae·f) for a forward converter, and the ripple, not the DC bias, for an inductor. A material is used only within 1.5× the frequency span of its data. Materials whose constants could not be traced to a data sheet are still offered, but every design that uses one is marked unverified. Thermal resistance depends on the core's size, from the TDK data book, not on its material.
Constraints enforced on every candidate: Bpeak < 0.8 × Bsat, fill factor < 0.40, and junction temperature ≤ Tmax.
Related Calculators
Read the Guide
FAQ
What does the objective weight slider do?+
At 0, the tool selects the minimum-loss design from the Pareto front. At 1, it selects minimum volume. At 0.3 (default), it selects the balanced point closest to the knee of the Pareto curve.
Why does the optimizer sometimes choose powdered iron over ferrite?+
Powdered iron toroids have much higher saturation flux density (1.4–1.5 T vs 0.35–0.43 T for ferrites). For high-current inductors with significant DC bias, iron powder can achieve higher inductance density despite higher core loss coefficients.
What is the Dowell model?+
The Dowell model computes AC resistance factor F_R = R_ac/R_dc as a function of wire diameter, skin depth, and number of winding layers. At 100 kHz, a 4-layer winding of AWG22 wire can have F_R ≈ 5, meaning AC copper loss is 5× the DC prediction.
How does interleaving reduce winding loss?+
P-S-P interleaving splits the primary into two halves on either side of the secondary, halving the effective layer count. This cuts the proximity effect term (which scales as n²−1) by up to 5×, dramatically reducing AC copper loss above 50 kHz.