How it works
An inductor stores its energy almost entirely in the air gap. The core's job is to guide the flux with little magnetizing effort and without saturating, and the gap's job is to set the inductance and hold the energy ½·L·I². The flux density in the core follows from the inductance and current alone: the flux linkage L·I is shared by N turns over the area A_e, so B_pk = L·I_pk/(N·A_e). For a given inductance and peak current, flux density can only be lowered by more turns or a larger core.
There are two common ways in. If you buy a core already gapped, the datasheet gives its A_L value, the inductance per turn squared. The number of turns follows from N = √(L/A_L), rounded up, and the page then checks whether the resulting flux density is acceptable. A high A_L reaches the inductance with few turns but drives the flux density up, which is the usual reason a first attempt saturates. If you gap the core yourself, it is more natural to start from the flux density: choose the fewest turns that keep B_pk below your limit, then find the gap that brings the inductance to target. With reluctances in series, N²/L = l_e/(μ0·μ_r·A_e) + g/(μ0·A_e), which gives the gap directly.
Turns have a cost in copper. Each turn needs a wire cross-section of I_rms/J, and all of them must fit in the winding window together with insulation and the bobbin. The page reports the bare copper fill N·A_w/W_a. With round wire on a bobbin, 30 to 40 % is a common practical ceiling, lower if safety insulation is needed. The chart shows the trade-off for the target inductance: flux density falls as 1/N while fill rises with N, and a design works only where both are below their limits. If no such region exists, the core is too small.
The peak-to-peak ripple current sets the AC flux swing ΔB, which, together with frequency, drives core loss. The page gives half of it, the amplitude that core-loss curves are usually plotted against.
Design procedure
- 1Take L, the peak current at the worst case, the ripple and the RMS current from the converter design.
- 2Pick a candidate core and enter A_e, l_e and the winding window from its datasheet.
- 3Choose the method: A_L for a pre-gapped core, or B_max for a core you will gap.
- 4Check that B_pk stays below the limit at the highest current and temperature, and that the window fill is realistic.
- 5If both cannot be met, move to a larger core. Then estimate core loss from ΔB/2 and the frequency, and copper loss from the winding resistance.
- 6Wind a prototype, measure L at the operating current, and adjust the gap for fringing.
Formulas
- Turns from A_L
- N = ⌈√(L/A_L)⌉, L_actual = A_L·N²
- Minimum turns from B_max
- N ≥ L·I_pk/(B_max·A_e)
- Air gap (no fringing)
- g = μ0·N²·A_e/L − l_e/μ_r
- Flux density
- B_pk = L·I_pk/(N·A_e), ΔB = L·ΔI/(N·A_e)
- Energy and copper
- E = ½·L·I_pk², A_w = I_rms/J, fill = N·A_w/W_a
Worked example
The defaults describe a 47 µH buck inductor carrying 12 A peak with 3 A of ripple and 10 A rms, on a ferrite E core with an effective area of 97 mm², a 79 mm magnetic path and a 120 mm² winding window. These dimensions are illustrative, not a specific part. With a 300 mT limit the fewest turns are 19.4, so 20 turns. A gap of 0.998 mm then gives 47 µH, and 96 % of the magnetic reluctance sits in that gap; ignoring the core's μ_r of 2000 would give 1.04 mm. The core runs at 291 mT peak with a 73 mT ripple swing, the inductor stores 3.4 mJ, and at 5 A/mm² each turn needs 2 mm² of copper, which fills a third of the window. Using a pre-gapped core with A_L = 250 nH/N² instead reaches the inductance with only 14 turns (49 µH), but the flux density rises to 433 mT, well above the limit. A 100 nH/N² core needs 22 turns and gives 48.4 µH at 272 mT. The gap and flux values were checked against a separate Python calculation that solves the magnetic circuit by bisection.
Common mistakes
- Choosing the highest A_L to minimize turns and ending up in saturation at peak current.
- Using B_sat at 25 °C when the core runs at 100 °C, where ferrite saturates much earlier.
- Forgetting that the gap value is the total in the magnetic path; with spacers under the outer legs of an E core, the effective gap is in two places.
- Sizing copper for average current instead of RMS current.
- Ignoring fringing near a large gap, which raises inductance and heats the turns closest to it.
Frequently asked questions
Why does a higher A_L core saturate more easily?
Because it reaches the same inductance with fewer turns, and B_pk = L·I_pk/(N·A_e) rises as N falls. The stored energy is the same, but it is squeezed through fewer ampere-turns per unit of flux. For an inductor carrying a large DC current, a smaller A_L, meaning a larger gap and more turns, is usually the right choice.
What should I use for B_max?
For ferrite, the saturation flux density falls with temperature, typically to around 350 to 400 mT at 100 °C for power materials. A design limit of 250 to 300 mT at the highest operating current leaves margin. Powder cores do not saturate sharply but lose permeability with DC bias, so they are designed from the manufacturer's permeability-versus-bias curves instead.
How much does fringing change the result?
Flux bulges out around the gap, which effectively widens it and makes the inductance higher than the formula predicts. The correction grows with the gap length relative to the leg width; for gaps of a few percent of the leg width it is small, for larger ones it can be 10 to 30 %. McLyman gives a fringing factor that estimates it. In practice you wind, measure, and adjust the gap or turns.
Is the window fill the copper area or the wire with insulation?
It is the bare copper area I_rms/J times the number of turns, divided by the window area. Enamel, the bobbin, insulation tape and gaps between round wires all take space, which is why practical copper fill factors are much lower than one.
How were the numbers checked?
A separate Python program solves the magnetic circuit independently: it finds the gap by bisection on L(g) = N²/(R_core + R_gap) and computes the flux from N·I divided by the total reluctance. Turns, gap, flux density, ripple swing and fill agree with the calculator to the displayed digits for all three test cases.
References
- Col. W. T. McLyman, "Transformer and Inductor Design Handbook" (gapped-inductor design, the reluctance form of the gap equation, the fringing-flux factor, window utilization).
- R. W. Erickson and D. Maksimović, "Fundamentals of Power Electronics", chapter on inductor design (magnetic circuits, the maximum-flux-density constraint and window fill).
- Ferroxcube, "Soft Ferrites and Accessories" data handbook (definitions of A_L, A_e, l_e and saturation flux density versus temperature for power ferrites).
- The values on this page were recomputed independently in Python by solving the magnetic circuit.
Results are engineering estimates. Verify your design with datasheets, simulation and measurement.