A message rides on the difference between two wires. Noise lands on both at once. One toroid, wound a particular way, can tell those two apart by feel alone — it lets the difference sail through and chokes whatever the wires hold in common. No frequencies measured, no logic. Just flux that either cancels or piles up.
Run both wires of a signal pair through the same toroid, wound the same direction. Now there are two kinds of current that can flow, and the core treats them as opposites:
The differential current is the signal — it goes out on one wire and back on the other, a complete loop. Through the core, those two flow in opposite senses, so their fluxes cancel. The core barely notices: low impedance, the signal passes untouched.
The common-mode current is the noise — picked up from the world, it pushes the same direction on both wires at once. Through the core, those fluxes add. The core slams down a large impedance and chokes it. Same toroid, same wires — opposite fate, decided entirely by whether the current is a difference or a sameness.
It never measures the noise. It just offers an easy path to the signal and a wall to anything the two wires are doing together.
Watch the core decide. In differential mode the two windings' fluxes oppose and the ring goes quiet — current strolls through. Flip to common-mode and the fluxes stack, the ring lights up, and the impedance shoots high — the current hits a wall. The toroid is doing subtraction with magnetism.
Now the communication payoff, the reason any of this matters. Send a differential PWM message down a pair, and let the world dump noise onto the cable — the way a motor, a radio, or a nearby switching supply actually would, hitting both wires equally. The choke strangles the common-mode noise current; the receiver reads the difference of the two wires, where the shared noise subtracts to nothing. Out the far end comes the message, intact, as if the channel were clean.
The message was never in either wire — it was in the gap between them. Noise can climb onto both wires, but it cannot climb into the difference. That is why two wires beat one, and why the choke is the last piece of the channel.
The toroid turned out to be a complete communications device, and it never did anything but respond to a change in flux. Across three papers it learned to carry, to cross, and to discriminate —
v = N·dΦ/dt. a held level dies in the core, so the message must be edges — and PWM is the message-in-edges.Send on the edges, cross the gap isolated, reject what the wires share — that's a transmitter, an isolator, and a noise filter, all in one ring of wound iron. And like its electric twin the capacitor, it does it all with a single stubborn refusal: it cannot feel what holds still.
The two series are duals — the same idea in opposite fields. Worth seeing them side by side:
i = C·dv/dtv = N·dΦ/dtTwo parts, two fields, one law of communication: notice only the change, and let the change cross the space between. The capacitor watches the voltage; the toroid watches the flux. Neither can hear a thing that holds still — which is exactly what makes them listen.
A closing note for the archive: this magnetic series is the deliberate mirror of the capacitor series — electric and magnetic duals, the two halves of how passive parts carry a signal. Together they bracket the whole field. If it ever extends, the next ring is the current transformer (the toroid that reads a wire without touching it — sensing, the V·dC/dt of the magnetic world), or up into real cores where saturation and hysteresis finally make the ideal pay its debts.