I was thinking more about the circuit in Will a FDP3682 MOSFET work in this delay circuit?.

In my current design, I have a TVS (SA18CA, Vc ≈ 30V) across the relay coils. However, I also have a bunch of other components upstream (i.e. across the +/- of V1) that are potentially exposed to the coil surge.

This can be represented approximately as follows:


(The RC delay between SW1 and M1 has been omitted; see linked question for details. SW1 here corresponds to V1 of that circuit. The LED is standing in for "other components upstream".)

Does the presence of M1 stop the surge voltage from being seen upstream? If not, what would be an appropriate way of protecting upstream components?

p.s. As a reminder, yes, a flyback diode (instead of a TVS) would eliminate the problem, but they create other problems that I'd rather avoid.

  • \$\begingroup\$ Thanks for the paper! A zener+diode also may be considered. Have you, and rejected it? Even with that, I think the rest of your question still stands, even with the paper's subject addressed. There are also snubbers that can be applied to mitigate higher frequencies in de-energizing. But I'm no expert on this -- certainly the paper taught me a few things which make me wonder more about the interaction of mechanical and electrical behaviors on each other. A fuller analysis would include the entire mutually interacting mechanical and electrical circuit as a system. I haven't seen that done, yet. \$\endgroup\$ Commented Mar 3, 2023 at 20:25
  • \$\begingroup\$ @periblepsis, thanks! My understanding is that a Z+D is basically the same as a (directional) TVS, at least for this application? (Alas, electronics.stackexchange.com/questions/649822 never got any answers.) At any rate, my understanding is that you can't avoid a voltage spike without slowing the magnetic collapse. I think an RC snubber would work, by absorbing the spike, but I'm not very familiar with how to build those, whereas a TVS is a drop-in component. \$\endgroup\$
    – Matthew
    Commented Mar 3, 2023 at 20:57
  • \$\begingroup\$ An RC snubber design isn't complicated. Normally, you look at the relay datasheet to find the engage and disengage times they specify. (Most of the datasheets I see do specify times, though not always both times.) From the coil resistance and this time, you can readily work out the number of Henries in the coil. This new information, plus once again the time value, helps design the RC snubber. That said, there's a lot more hinted at in that paper that tells me that even what I just said to you has more lurking under the hood and that a truly smart design would apply still more theory. \$\endgroup\$ Commented Mar 3, 2023 at 21:20

1 Answer 1


Coil flyback is not some magic surge; it's simply the response to the switch's relatively fast turn-off (in comparison to the L/R time constant of the coil).

The current can never be higher than the current at the instant of turn-off, and the voltage is controlled at the switch node first and foremost. Simply putting a diode there, clamps the voltage seen by the switch.

The coil reaction is with respect to the supply. No transient is seen at the supply, aside from the change in load current, and whatever effect that has upon it.

The supply itself may be inductive and show some peak voltage — but that's characteristic of the supply, not of putting a switched coil load on it.

With this in mind, perhaps you may find this anecdote amusing. Or perhaps a little enlightening still:


simulate this circuit – Schematic created using CircuitLab

I was once tasked with repairing an electromagnet. This was a heavy benchtop unit, with a pair of thick steel pole pieces on top to hold the workpiece against. The switch was a momentary rocker type, so you hold the piece, tap the switch, BUZZZ, and your part is magnetized. Simple enough. I opened up the housing and observed this circuit. (Component values are vague guesses.)

I might be misremembering whether the MOV was connected across the AC or DC side of the full-wave rectifier (FWB), but in either case, clearly it was an attempt to limit surge voltage across the coil.

If one were switching such a heavy coil on AC mains directly, the switch would experience considerable arcing and wear, and a MOV would be beneficial to suppress the arc energy. It seems this thought process was applied here. The designer/builder did not appreciate that, in fact, the FWB already serves this purpose.

Consider when the switch is pressed: AC is rectified to DC, and current flows through the coil, charging it. [Conventional] current flows in a loop, upward through the diodes, downward through the coil. AC power pushes more current into this loop, through alternate pairs of diodes; near the zero crossings, current decays, dropping through the coil's resistance and the diodes' drop.

When the switch is opened, the coil's current continues, discharging through its resistance and the diodes, in precisely the same loop. Nothing has changed, dI/dt is small, and peak voltage is irrelevant.

I forget what the fault was anymore (maybe just the switch?), but I do recall the MOV was not the culprit.

  • \$\begingroup\$ Alas, this has quite a few dissimilarities with my circuit. Critically, the switch (a MOSFET in this instance) is on the DC side, so it interrupts the discharge loop you described. \$\endgroup\$
    – Matthew
    Commented Mar 3, 2023 at 21:00
  • \$\begingroup\$ It's not an analog of your circuit, merely an anecdote, with related behavior. \$\endgroup\$ Commented Mar 3, 2023 at 21:02
  • \$\begingroup\$ Is there something about "upstream" that you're not telling us? ...Has this whole series of questions been a huge X-Y problem? \$\endgroup\$ Commented Mar 3, 2023 at 21:31
  • \$\begingroup\$ Let us continue this discussion in chat. \$\endgroup\$
    – Matthew
    Commented Mar 3, 2023 at 21:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.