Inductor Design Guide: Irms, Isat, Wire Diameter, Core Material, SRF & Impedance

Choosing an inductor is never a single-number decision. The rated inductance and current on a datasheet are only the visible tip of a design that must balance heat, saturation, high-frequency loss, parasitic resonance and cost.

In this guide we break down the six parameters engineers argue about most — Irms, Isat, wire diameter, core material, self-resonant frequency (SRF) and impedance — show how they constrain one another, and work through a demanding real-world example: a 4.2 mH / 25 A inductor operating at 1 MHz .

Whether you are specifying a power choke for a resonant converter, a filter inductor for an inverter, or a custom magnetic component, the same trade-offs apply. Read to the end for a practical design checklist and a FAQ section written for search visibility.

The Six Parameters at a Glance

1. Irms — The Heating Limit, and How Wire Diameter Follows

Irms is the continuous RMS current the inductor can carry while staying within its allowable temperature rise — typically 40 °C above ambient. It is fundamentally a thermal specification, not a magnetic one. The copper loss is:

To keep Irms high you need copper area. The required conductor cross-section comes straight from your chosen current density J (a design rule of thumb, not a physical constant):

So wire diameter is not picked for looks — it is the direct consequence of Irms and your cooling budget. A 25 A inductor needs roughly 4–8 mm² of total copper area. But at high frequency that area is only useful if the current can actually reach it (see skin effect below).

2. Isat — The Magnetic Limit, Set by Core Material

Isat is the DC (or low-frequency) current at which the inductance has fallen to a specified fraction of its initial value — usually 80% or 90%. Beyond this point the core can no longer support additional flux and the part loses its inductance. The peak flux density in the core is:

At Isat, B_peak approaches the material’s saturation flux density B_sat. Two design levers follow:

• To survive a given current without saturating: raise N or A_e, or pick a core with more cross-sectional area — but more turns also raise DCR and lower SRF.
• To get more current headroom: a higher B_sat material (iron powder, sendust, MPP, high-flux) tolerates more current in the same volume than ferrite.

The stored energy tells the same story from the energy side. The energy an inductor must hold at its rated current is:

That energy must fit inside the core before it saturates, which is why a 4.2 mH / 25 A part is a large component: even at a conservative Isat of 30 A it must store about 1.9 J. Powder cores (distributed air gap) are favoured here because their soft, gradual saturation curve rides through current spikes without a sudden inductance collapse.

3. Wire Diameter vs Frequency — Why 1 MHz Demands Litz Wire

At low frequency, wire diameter is set only by Irms and current density. At high frequency, the skin effect forces current into a thin surface shell. The skin depth in copper is:

Rule of thumb: each individual conductor should have a diameter no larger than about 2δ (≈ 0.13 mm at 1 MHz) or its centre stops carrying current and your effective copper area collapses. A solid 2.5 mm wire is therefore almost useless at 1 MHz — you would pay for copper that does nothing.

The standard fix is Filo Litz: many individually insulated fine strands (each ≤ 2δ) braided so that, on average, every strand occupies every radial position. For our 25 A example with J ≈ 4–5 A/mm² we need ≈ 5–6 mm² of copper area, delivered as, for instance, hundreds of 0.071–0.10 mm strands grouped into 2–3 parallel Litz bundles. This keeps AC resistance close to DCR and is the only practical way to hit both 25 A and 1 MHz.

4. Core Material Selection — μ, B_sat, Losses and Frequency

The material choice sits at the centre of every trade-off. The usual suspects:

For a 4.2 mH / 25 A / 1 MHz choke the priorities are: (a) enough effective permeability to reach 4.2 mH without an absurd turns count, (b) a high B_sat or soft-saturation curve to absorb 25 A, and (c) low core loss at 1 MHz. That points to a distributed-gap powder core — MPP or carbonyl iron for the lowest loss, high-flux or sendust where B_sat headroom matters most. Ferrite would need very high μ to keep turns down but its low B_sat makes 25 A saturation risky; nano-crystalline is the standout choice for this 1 MHz spec: with very high permeability (μ_r in the tens of thousands) it reaches 4.2 mH with a fraction of the turns a powder core would need — the single biggest lever for keeping SRF above 1 MHz — while its ~1.2 T saturation and low high-frequency loss easily absorb 25 A. TrafoPSU’s O-26092 planar transformer is a real production example of 4.2 mH / 25 A at 1 MHz built on a nano-crystalline core. Distributed-gap powder cores (MPP, sendust, high-flux) remain a strong alternative where cost or soft-saturation behaviour is the priority.

5. Self-Resonant Frequency (SRF) — The Hidden Capacitor

Every winding has parasitic capacitance between turns and layers. Together with the inductance it forms a parallel resonant tank. The self-resonant frequency is:

Below f_srf the part is inductive; at f_srf its impedance peaks; above f_srf it behaves like a capacitor and is useless as an inductor. Two rules:

• Margin: keep f_srf at least 2–3× (ideally 5×+) above your operating frequency.
• Lever: fewer turns, tighter/sectioned winding, and lower inter-winding capacitance all raise f_srf — but fewer turns also lower Isat for a given core and raise the μ needed.

This is where our example bites. To reach 4.2 mH you need many turns, and many turns mean more C_p and a lower f_srf. Holding f_srf ≥ 2 MHz with L = 4.2 mH forces C_p ≤ ~1.5 pF — a tiny budget that demands a high-μ material (to minimise N) and a carefully sectioned winding. There is the core tension of the whole design in one sentence: SRF wants few turns, Isat wants many turns.

6. Impedance — What the Circuit Actually Sees

The inductive reactance at operating frequency is the headline number for a choke:

For 4.2 mH at 1 MHz: X_L = 2π × 1×10⁶ × 4.2×10⁻³ ≈ 26.4 kΩ. That is the impedance the inductor presents to AC — and it must dominate the series resistance (DCR + AC resistance) for the part to behave like an inductor rather than a resistor. The quality factor makes this explicit:

Near, but below, SRF, the full impedance is |Z| = √( R² + (ωL − 1/(ωC))² ). As frequency approaches f_srf the capacitive term cancels the inductive term and Z collapses. In practice you optimise for a high, flat impedance across your band of interest while keeping R_ser (hence DCR and AC loss) small enough that the 25 A load does not cook the part.

7. How the Parameters Interlock — The Trade-off Map

No parameter moves alone. The dominant couplings:

• Current vs turns: Irms ↑ ⇒ more copper area ⇒ bigger wire/bundles ⇒ less window space for turns ⇒ fewer N ⇒ Isat ↓ and SRF ↑.
• Material vs SRF/Isat: higher μ_r ⇒ fewer turns for same L ⇒ SRF ↑ and DCR ↓, but usually lower B_sat ⇒ Isat ↓ (use powder/soft-sat cores to recover).
• Frequency vs wire: f ↑ ⇒ skin depth ↓ ⇒ need Litz (more strands, bigger bundle) ⇒ window fill pressure ⇒ fewer turns possible.
• Turns vs everything: more turns ⇒ L ↑ and Isat ↑, but C_p ↑ ⇒ SRF ↓ and DCR ↑.

Designing an inductor is the act of walking these couplings to a feasible point. For 4.2 mH / 25 A / 1 MHz the feasible point is a large distributed-gap powder core, high-μ to keep turns manageable, wound with parallel Litz bundles, sectioned to protect SRF, and cooled so the 25 A copper loss stays within budget.

8. Worked Example — 4.2 mH / 25 A at 1 MHz

Step 1 — Skin depth and wire

Step 2 — Saturation energy and core size

This 1.9 J must fit before saturation → we select a nano-crystalline core (B_sat ≈ 1.2 T) large enough in volume, with distributed-gap powder cores (MPP / high-flux / carbonyl iron) as the main alternative where cost or soft-saturation is preferred.

Step 3 — Turns from required A_L

This is the key reason the 1 MHz target is reachable: very high μ_r lets the choke hit 4.2 mH with few turns, which keeps inter-winding capacitance tiny and pushes SRF well above the operating frequency. Fewer turns also mean lower DCR and easier window fill for the Litz bundle.

Step 4 — SRF budget

Easily achievable here because nano-crystalline’s high μ already cut N to a few tens of turns;  sectioned winding keeps inter-winding capacitance in the sub-pF range, pushing f_srf comfortably above 1 MHz. This is the direct payoff of the high-μ choice made in Step 3.

Step 5 — Impedance and loss check

Confirm Q = X_L / R_ser is high enough for the circuit and that core loss at 1 MHz (from the material’s loss curve at your B_peak) stays within the temperature rise target. Iterate core size / μ_r / bundle count until SRF, Isat, Irms and temperature all pass. This nano-crystalline core’s low loss at 1 MHz keeps core heating small, so the 25 A copper loss dominates the thermal budget — another reason the design closes.

9. Inductor Design Checklist

1. Specify: Define L, I_rms, I_peak, f, and allowed ΔT.
2. Thermal: Pick J → compute A_cu → choose Litz strand count for δ at f.
3. Magnetic: Compute E = ½ L I_sat² → size core for B_sat headroom.
4. Material: Select material by μ, B_sat, loss at f, and cost.
5. Turns: N = √(L/A_L); check window fill with Litz bundle.
6. SRF: Budget C_p so f_srf ≥ 2–3× f; use sectioned winding.
7. Impedance: Verify X_L and Q; confirm DCR/ACR loss vs ΔT.
8. Validate: Prototype, measure L(I), SRF, DCR, and temperature rise.

Frequently Asked Questions

What is the difference between Irms and Isat in an inductor?

Irms is the continuous current limited by heating (I²R copper loss and temperature rise), while Isat is the current at which the core saturates and inductance collapses. A part can be limited by either — always verify both ratings.

How do I choose inductor wire diameter?

Start from current density J (≈ 3–6 A/mm² natural cooling) to get the copper area for Irms, then cap each conductor at about 2× skin depth. At 1 MHz that means Litz wire with strands ≤ ~0.13 mm.

Which core material is best for 1 MHz inductors?

For 1 MHz power magnetics, distributed-gap powder cores — MPP, sendust, carbonyl iron or high-flux — combine enough permeability with high B_sat and low loss. Ferrite suits lower frequency or RF; nano-crystalline is the standout for this 1 MHz spec — its very high μ reaches 4.2 mH in few turns, keeping SRF above 1 MHz while its ~1.2 T B_sat and low loss absorb 25 A (see TrafoPSU’s O-26092).

Why does self-resonant frequency (SRF) matter?

Above SRF the winding’s parasitic capacitance dominates and the part behaves like a capacitor. Keep f_srf at least 2–3× above your operating frequency, which usually means minimising turns and inter-winding capacitance.

How is inductor impedance calculated at high frequency?

Below SRF, |Z| ≈ 2πfL. Near resonance use |Z| = √(R² + (ωL − 1/ωC)²). The impedance must stay well above the series resistance so the part acts as an inductor, not a resistor.

Related: Inductors & Chokes - Trasformatori planari - Trasformatori di corrente - Nano-Crystalline Cores

Website: www.trafopsu.com