Amorphous Alloy C Core for High Power Conversion: Dimensions, Loss and Selection
Sep 22, 2025
Leave a message
Why Amorphous C Cores Are Used in High-Power Designs
An amorphous C core is wound from Fe-based amorphous ribbon in the Fe-Si-B system, impregnated, then cut into two C-shaped halves so copper wire can be wound onto a bobbin and slipped into the window. The alloy is solidified at a cooling rate near 106 K/s, which leaves the atoms without a long-range crystal lattice. No grain boundaries and no crystalline anisotropy mean the domain walls move easily, so hysteresis loss per kilogram is far below that of grain-oriented silicon steel at the same flux swing.
The ribbon used for these cores is 25 µm thick. Finished cores reach a stacking factor above 0.8, saturation flux density of 1.56 T, Curie temperature of 410 °C, electrical resistivity of about 110 µΩ·cm and saturation magnetostriction below 30×10-6. Resistivity in the 110–130 µΩ·cm range is what keeps eddy-current loss manageable from 10 kHz to 50 kHz, where a 0.23 mm electrical steel lamination of comparable mass would already run hot. Frequency-domain measurement methods for ring and toroidal specimens in this band are described in IEC 60404-6, and high-excitation measurement of soft magnetic cores in IEC 62044-3.
Loss and Saturation Behaviour Under Load
Two numbers decide whether a C core survives a high-power duty cycle: the peak flux the core actually sees, and the resulting temperature rise. Because Bs is 1.56 T, a designer can run a 0.5 T to 0.7 T peak flux swing in a PFC choke or 0.2 T to 0.4 T in a medium-frequency transformer and keep volume modest. Cut cores with a small distributed gap tolerate DC bias far better than gap-free wound cores, because the gap linearises the B-H curve and stores energy without driving the material into saturation.
Loss of a finished core is normally verified on the component itself rather than on a strip sample, because cutting, resin impregnation and clamping all add mechanical stress. Alternating-current magnetic properties of toroidal and C-type specimens are commonly verified with the voltmeter-ammeter-wattmeter method of ASTM A927/A927M. Procured lots should carry loss data at the specified flux density and frequency, measured after the final anneal and impregnation, not from a strip coupon taken before cutting.
Standard C Core Dimension Series
The table below lists a typical high-power C core series. Dimensions A to F are the finished cut-core geometry in millimetres; AFe is the effective cross section, lFe the mean magnetic path length and mFe the core mass.
| A (mm) | B (mm) | C (mm) | D (mm) | E (mm) | F (mm) | AFe (cm²) | lFe (cm) | mFe (g) |
|---|---|---|---|---|---|---|---|---|
| 11 | 13 | 50 | 30 | 35 | 72 | 15.7 | 2.94 | 331 |
| 13 | 15 | 56 | 30 | 41 | 82 | 17.9 | 3.47 | 447 |
| 16 | 20 | 70 | 30 | 52 | 102 | 22.7 | 4.27 | 696 |
| 19 | 25 | 83 | 40 | 63 | 121 | 27.2 | 6.76 | 1322 |
| 22 | 35 | 85 | 50 | 79 | 129 | 30.6 | 9.79 | 2148 |
| 25 | 40 | 85 | 70 | 90 | 135 | 32.5 | 15.58 | 3635 |
| 30 | 40 | 95 | 85 | 100 | 155 | 36.1 | 22.70 | 5879 |
| 33 | 40 | 105 | 85 | 106 | 171 | 39.0 | 24.96 | 6994 |
Note how the mass grows faster than the cross section. Moving from the 19/25/83 stack to the 25/40/85 stack multiplies AFe by only 1.2 while mass rises by 2.8 times, because lFe more than doubles. For a fixed number of ampere-turns, magnetising current scales with lFe/AFe, so oversizing the window to fit more copper usually buys less than a designer expects.
Winding, Gapping and Assembly
C cores are supplied as matched pairs, and the two halves of one pair must not be swapped with another pair. The ground faces are lapped flat; a single particle of swarf in the joint adds an unintended gap, drops inductance and raises leakage. Recommended practice:
Keep the mating surfaces clean and assemble dry unless a controlled gap is specified.
Clamp with uniform pressure, normally by stainless banding or a metal bracket, not by the bobbin alone.
Wind to fill the window evenly; a single-layer winding on the inner leg keeps leakage low in filter chokes.
After assembly, re-check inductance at the operating flux level, not at a small signal level.
For a given core, gap size sets the achievable permeability and the DC-bias capability. Increasing the gap reduces inductance but pushes saturation to a higher current, which is normally the desired trade for a PFC choke or an output inductor carrying a large DC component. Where the gap is critical, specify it as a magnetic quantity (A·turns per tesla) plus a mechanical tolerance, since a specified mechanical gap and the resulting AL are not perfectly interchangeable across production lots.
Typical Applications
Solar inverter filter inductors, power factor correction chokes, medium-frequency transformers, differential-mode chokes and output smoothing filters are the usual duties. In each case the selection logic is the same: set the peak flux from the ripple current plus DC bias, choose AFe so the core stays below roughly 0.7 T, then choose the window so the required turns fit with acceptable copper loss. Where the ambient is above 60 °C, derate the allowed temperature rise rather than the flux, because loss rises with temperature in amorphous ribbon more slowly than in ferrite but the winding insulation class sets the real limit.
Frequently Asked Questions
Q: What is the saturation flux density of an amorphous C core?
A: The Fe-based amorphous ribbon used in these cores has a saturation flux density of about 1.56 T at room temperature, so it carries roughly 20 percent more flux than a silicon steel lamination grade rated near 1.3 T without saturating.
Q: Can an amorphous C core replace a ferrite core in a high-power inductor?
A: Yes in most 10 kHz to 50 kHz designs, because the amorphous C core offers higher saturation flux density and better DC-bias tolerance, though the insulated ferrite core remains cheaper when flux swing is small and the window is generous.
Q: Why are C cores cut into two halves instead of being supplied as a closed ring?
A: Cutting allows a pre-wound bobbin with heavy copper or litz wire to be dropped into the window and the two halves re-closed, which is not possible with a closed toroid when the winding has many turns of thick conductor.
Q: How much does the joint affect measured inductance?
A: A clean lapped joint adds only a small effective gap, but contamination or an uneven clamp load can cut inductance noticeably, so inductance is normally specified and inspected after assembly at the specified flux level.
Q: Does the core need to be re-annealed after winding?
A: No. Amorphous ribbon is already in its final magnetic state, so the core is annealed and impregnated before delivery. Winding stress is small compared with the stress imparted during cutting, so the supplied loss data remains valid if clamping practice is followed.
Q: What frequency range suits a C-type amorphous core?
A: Roughly 1 kHz to 50 kHz is the comfortable band. Below 1 kHz consider electrical steel, and above 50 kHz winding and eddy currents usually dominate, so the core advantage shrinks.