Amorphous and Nanocrystalline Power Transformer Cores: C Core, E Core and Block Builds

Two Strip Materials, One Core Platform

Amorphous and nanocrystalline cores start from the same process: melt spinning produces a thin ribbon that is wound into a ring, annealed, impregnated and then cut or moulded into the finished shape. The difference is composition and heat treatment. Fe-based amorphous ribbon in the Fe-Si-B system is used as cast and reaches a saturation flux density of about 1.56 T with a Curie temperature near 410 °C and resistivity around 110–130 µΩ·cm. Nanocrystalline ribbon adds niobium and copper, and a controlled anneal precipitates grains of about 10 nm to 15 nm, which gives permeability an order of magnitude higher and a much lower coercivity, at the cost of a lower saturation flux density of about 1.2 T.

Both materials are described by the same family of measurement standards. Core losses and permeability of ring specimens between 20 Hz and 200 kHz are measured under IEC 60404-6, high-excitation behaviour of soft magnetic cores under IEC 62044-3, and alternating-current properties of toroidal cores by the voltmeter-ammeter-wattmeter method of ASTM A927/A927M. Buyers should ask for the measurement condition, not just the loss figure, because loss quoted at 0.2 T and 20 kHz is not comparable with loss quoted at 0.5 T and 10 kHz.

C Core, E Core and Block Builds

The three builds differ only in how the window is created:

C core - a wound ring cut into two C halves; best for heavy or awkward windings, most common in medium-frequency transformers and filter inductors.

E core - two E-shaped halves from cut and formed ribbon, giving a large winding area and convenient bobbins for multi-winding transformers.

Block core - rectangular stacks of ribbon with an open window; used where a flat, low-profile footprint matters.

Controlling Loss and Temperature Rise

Core loss in either material is the sum of hysteresis and eddy-current components, and both scale with the flux swing raised to a coefficient above one, so the practical rule is to keep the flux swing as low as the volume budget allows and to raise frequency only after recalculating loss. A nanocrystalline core with permeability in the tens of thousands can be much smaller than an amorphous core for the same inductance at low flux, but as soon as a modest DC bias is applied the nanocrystalline core loses more of its initial permeability and needs a larger gap.

Temperature rise is dominated by copper, not by the core, once the current is above a few amperes. Loss in the core still matters because it is generated in the centre of the winding where heat is hardest to remove. Designers should therefore model the core as a distributed heat source and check that the winding hot spot stays inside the insulation class limit, rather than relying on a core-only loss figure.

Gaps, DC Bias and Anti-Saturation Design

An open or distributed gap is standard on power cores for two reasons. First, it linearises the B-H curve so inductance holds up as DC current increases. Second, it lets the designer set the inductance by gap size rather than by material permeability, which reduces lot-to-lot variation. Permeability of a gapped amorphous core is normally specified in the 50 to 300 range, adjustable by gap.

Anti-rust treatment matters as much as the magnetic design. Amorphous and nanocrystalline ribbon cannot be coated with the insulating varnish used on electrical steel because the coating would have to survive the anneal, so finished cores receive a resin dip or a thin epoxy layer. Cores that must pass a salt-spray or humidity test should be ordered with an explicit finish specification.

Applications and Selection Logic

Typical duties are the high-frequency output filter reactor of a switched-mode supply, auxiliary power transformers for rail transit, intermediate-frequency transformers, inductors for electric drive systems, PV inverter and energy storage boost inductors, high-current storage inductors, power transformers and Hall-effect or antenna applications. Selection follows four steps: fix the inductance under worst-case DC bias, cap the flux swing, pick the build that fits the available winding space, then confirm loss and temperature rise with vendor data taken at the actual operating point.

Frequently Asked Questions

Q: When should nanocrystalline be chosen over amorphous?
A: Choose nanocrystalline when high permeability at low flux and a small footprint dominate, as in common-mode chokes or current sensors, and amorphous when the core must carry a large flux swing or significant DC bias, because its 1.56 T saturation is roughly 30 percent higher.

Q: Are these cores suitable for 50 Hz or 60 Hz power transformers?
A: They can be used, but the advantage over grain-oriented electrical steel disappears at line frequency because hysteresis loss per cycle is what matters and steel is cheaper per kilogram, so amorphous and nanocrystalline are normally reserved for 1 kHz and above.

Q: What is the difference between an open gap and a distributed gap?
A: An open gap is a discrete cut or spacer, giving repeatable inductance and strong DC-bias tolerance, while a distributed gap is created by interleaving a non-magnetic material through the ribbon; the distributed form reduces fringing flux and local heating near the gap.

Q: How is core loss specified for a C core at 20 kHz?
A: By a loss per unit mass or per unit volume at a stated flux density and frequency, for example 0.3 T at 20 kHz, measured on the finished core after anneal and impregnation under IEC 62044-3 or an equivalent method.

Q: Can the cores be supplied with a customer-specified window size?
A: Yes. Both C and E builds are produced from wound ribbon, so window width, height and length can be adjusted within the limits of the ribbon width available, and blocks can be stacked to reach a required cross section.

Q: Why must the two halves of a C core stay together?
A: Each pair is ground as a matched set. Mixing halves from different cores opens the joint by a few micrometres, which changes the effective gap and shifts inductance beyond the tolerance quoted for the pair.