Thyrectors (TVS Diode)

Indeed, that's what happens. And the higher voltage drop during that breakdown is key: the fundamental inductor equation is V = L * dI/dt, so more V gives faster dI/dt. :)

One doesn't need a brand name component to do it; a bidirectional TVS is simply two zeners end to end, or you can use a rectifier and zener to get a one-sided avalanche characteristic.

Another note: the diode actually belongs at the switch, not the coil. The coil generates /most/ of the EMF, but it's not the cause. It's just doing what it's told, so to speak. The switch is what's doing the work. This distinction is not very important for relays, but becomes paramount as you go up in frequency. For SMPSs, the stray inductance of interconnects, and transformer leakage, is comparable to the winding inductance, and cannot be ignored.

Tim

Agreed.

The problem I see with that, is that if you put the diode across the switch (and not the coil), you then have the coil dumping into the supply circuit. In those early days, spikes in the supply were a bit of a problem for the 8008 as built. :)

I can see that SMPSs have all sorts of challenges.

Warren ve3wwg

Well, not the logic supply. The relay supply. Also, you need a bypass at the switch, to close the loop -- otherwise it's who knows how long = how much stray inductance!

Again: the relay is just doing what it does, but the switch is the cause. Close the loop. :)

Tim