Nanocrystalline common-mode chokes for mains EMI filters: tape-wound VITROPERM and Nanoperm cores compared with ferrite ring-core double chokes

01

Permeability 40,000 that rolls off above 20 kHz: why a nanocrystalline choke's |Z| is larger than ωL

VACUUMSCHMELZE (VAC) gives VITROPERM 500 F a permeability of 40,000 and VITROPERM 250 F 5,000, against 5,000 for the MnZn ferrite it compares them with, a saturation flux density of 1.2 T and an electrical resistivity of 1.1–1.2 µΩm. The ferrite's permeability stays flat to about 1 MHz (to about 200 kHz for a µ = 10,000 ferrite), so inductance sets its impedance and |Z| ≈ ωL; the flat range of VITROPERM 500 F ends at tens of kHz, depending on the initial permeability (20 kHz in VAC's example), so in the EMC range above 150 kHz its impedance is dominated by the resistive part and |Z| ≫ ωL — the noise is absorbed rather than reflected. [1]

VAC's catalogue data show this on a standard part. Its 10 A low-profile choke T60405-R6123-X210 lists 2 × 11.4 mH at 10 kHz falling to 2 × 2.6 mH at 100 kHz, an impedance of 3,200 Ω at 100 kHz, a resonance at 0.7 MHz and 12 mΩ winding resistance per winding. [1] Editorial calculation: ωL at 100 kHz for 2.6 mH is 2π × 100,000 × 0.0026 ≈ 1,630 Ω, about half the 3,200 Ω listed, so an inductance-only comparison at 10 kHz would overstate the gap between 11.4 and 2.6 mH and misjudge the attenuation. [1]

VAC adds that winding layout matters as much as the core at high frequency: two chokes on the same 25 × 20 × 10 mm VITROPERM 500 F core with 2 × 28 turns of 0.71 mm wire differed by up to 16 dB of insertion loss at 4 MHz when the winding capacitance was cut from 21 pF to 4 pF. [1] Magnetec, which winds its chokes on Nanoperm cores, states that against ferrite versions they reach higher attenuation up to the MHz range at the same core size. [2] Both statements come from nanocrystalline core makers; the comparison has to be checked with an insertion-loss or impedance curve for the actual candidate parts.

Ask for impedance |Z| (or insertion loss in a 50 Ω system) against frequency from 150 kHz to 30 MHz, the inductance at 10 kHz and 100 kHz, the self-resonance frequency and the winding capacitance, not inductance at one frequency alone.

02

How many milliamps of common-mode current saturate the core: 13 to 485 mA on one maker's 2–40 A range

Because the line currents cancel, the core only sees the net common-mode current: leakage through the filter's Y capacitors and the drive's stray capacitances, unbalance between lines, and the noise itself. Magnetec lists this as a quasi-saturation current Isat in mA for each 2-fold choke (defined by Magnetec at B = 1.0 T, with the Nanoperm saturation current taken as the peak excitation current at which initial inductance has dropped to 10 %), and VAC lists a common-mode current Icm in mA beside each standard choke and asks for the maximum common-mode current at low frequency (<20 kHz) on its design checklist. [2] [1]

The two Magnetec 10 A parts show the trade: the 12 mH choke saturates at 40 mA of common-mode current, the 0.6 mH choke at 450 mA. Across the 2–40 A range Isat runs from 13 mA (MB-690) to 485 mA (MB-740). [2] Editorial note: a filter that will carry more common-mode current, for example from larger Y capacitors, points toward the low-inductance, high-Isat end; a filter with little leakage current can use the high-inductance part.

Stray inductance gives a common-mode choke its differential-mode attenuation. EPCOS describes its ferrite ring-core double chokes as having approximately 1 % stray inductance for symmetrical interference suppression; editorial calculation from its tables: 760 µH on the 47 mH B82722 part is 1.6 %, and 20 µH on the 1.8 mH B82725-A part 1.1 %. [3] The Magnetec MB-603 lists about 10 µH against 2 × 12 mH, a much smaller share (editorial calculation: about 0.08 % of one winding's inductance), so a nanocrystalline filter may need separate differential-mode parts. [2]

Give the supplier the expected common-mode current at low frequency (leakage and unbalance, in mA), the switching frequency of the equipment behind the filter, the Y-capacitor values and the differential-mode attenuation the filter needs.

03

Less than 10 % permeability change from −40 to 120 °C, and the temperature rise and ratings behind the nominal current

VAC gives the permeability change of VITROPERM between −40 °C and 120 °C as less than 10 % (typ.), while that of MnZn ferrites can drift ±40–60 % around the room-temperature value; the saturation flux density of VITROPERM changes by a few percent up to 150 °C, while that of MnZn ferrites falls by up to 40 % above 100 °C. The Curie temperature of VITROPERM alloys is above 600 °C. [1] Magnetec quotes a Curie temperature of approximately 600 °C and operation up to 130 °C for its Nanoperm chokes. [2]

VAC states that its standard chokes need data derating at other ambient temperatures or currents, and Magnetec gives its nominal currents at 70 °C ambient with a derating method for other temperatures and assumes double the thermal resistance figure for forced cooling. [1] [2] The 14 A forced-cooling rating of the Magnetec MB-603 against 10 A in convection shows how much the nominal current depends on how the filter is cooled. [2]

State the ambient and maximum operating temperature, cooling (convection or forced), overload current and duration, overvoltage category and the insulation and approval standards the filter must meet.

04

Choke data to send

Paste it into your enquiry below or into an email to a maker, then replace every bracket.

Lines: [1-phase or 3-phase; with or without neutral]
Current: [nominal; overload and duration; ambient; cooling]
Voltage: [rated; overvoltage category; standard EN 50178 / IEC 62109 / IEC 61800]
Noise: [switching frequency; target attenuation or |Z| at 150 kHz and above]
Common-mode current: [leakage and unbalance in mA at low frequency]
Differential mode: [stray inductance needed or separate DM parts]
Mounting: [upright or low profile; pins; maximum W × D × H]

See manufacturer records Send an enquiry through CNFX