Bay note 04 — Common-mode choke from the noise floor up
Size Lc from the dB you must kill, not from a catalogue guess — with a worked 150 kHz LISN example and the ferrite vs nanocrystalline fork.
A common-mode choke is not "an inductor with two windings". It is a filter element sized against a spectrum. Start from the line that fails, then wind.
The fork we care about
| Core | Best at | Watch-outs |
|---|---|---|
| MnZn ferrite | Cost, mid HF | Saturates if differential current is high |
| Nanocrystalline | High µ, lower turns | Cost, mechanical care |
| NiZn | Higher MHz | Lower µi — more turns |
Worked strip — 150 kHz, need about 20 dB
Order-of-magnitude CM estimate with line impedance Z ≈ 100 Ω (LISN-ish):
Attenuation (dB) ≈ 20 log10(ωLc / Z) when ωLc ≫ Z.
Want 20 dB at 150 kHz:
ωL / Z ≈ 10 → L ≈ 10 × Z / ω = 1000 / (2π × 150e3) ≈ 1.1 mH
So the bay starts near 1 mH common-mode inductance — then we check:
- Differential leakage (too much leakage bucks the load current)
- Iload heating in copper
- Core saturation from CM surge / Y-cap imbalance
Turns from AL
If the chosen toroid AL = 4000 nH/turn² and L = 1.1 mH = 1.1×10⁶ nH:
N = √(L / AL) = √(1.1e6 / 4000) ≈ √275 ≈ 17 turns bifilar.
What to put on the print
- Mains or DC bus voltage, Iload rms
- Frequency band that fails (or a LISN screenshot)
- Target attenuation or remaining margin
- Max height / footprint
- Safety: reinforced isolation between windings if required
See common mode chokes for wound stock; custom AL and turns are normal.