I have run into the same problem described here and here, when I added diode to electromagnet. Basically, the magnet still holds with about 20kg force several minutes after removing power. The mechanical solution of leaving air gap or adding a "peel-off" spring is not plausible due to design constraints. So, I need a snubber circuit that will protect relay contacts why still de-energize the magnet rapidly.

While looking for the solution I've stumbled upon this answer, listing many possible methods. Not having any experience with powerful electromagnets I can't decide which one would be the best for this particular setup (24V 14W magnet, 30V 10A SRD-05VDC-SL-C relay).

At this moment I am leaning towards either rectifier + zener or RC + MOV combination. Any advice, please?


I've confirmed that the problem is residual magnetization, not back EMF, just as @DKNguyen and others suggested. After disconnecting power I also disconnected diode and the magnet was still holding. Apparently the initial experiment without diode was done with heavy (~200kg) load that was enough to hide the magnetization.

So, I am going with RC snubber, hoping that oscillations will provide sufficient demagnetizing effect. Maybe combined with MOV if initial amplitude is too high.

  • \$\begingroup\$ The Amazon links don't really work. Please link to manufacturer datasheets directly. \$\endgroup\$
    – Justme
    Commented Apr 30, 2022 at 15:45
  • 7
    \$\begingroup\$ No, the problem is problem is residual magnetism, nor snubbing. The collapse of the current in the electromagnet actually happens fastest with NO snubber. You need to temporarily reverse the current through the electromagnet to a lesser amount than what you used to magnetize. Not too much though) to reset the magnetic domains. \$\endgroup\$
    – DKNguyen
    Commented Apr 30, 2022 at 15:46
  • \$\begingroup\$ The SRD-05VDC is a small contact relay, not an electromagnet. Something I didnt catch here. Are you trying to drive a big electromagnet with an SRD relay? is that the idea? \$\endgroup\$ Commented Apr 30, 2022 at 15:47
  • 1
    \$\begingroup\$ Unless the problem you really mean to say is that your relay is sticking when you try to open it so the electromagnet remains energized which is a different problem altogether from residual magnetism. But if that was the problem I would have expected you to word your question differently. \$\endgroup\$
    – DKNguyen
    Commented Apr 30, 2022 at 15:49
  • 1
    \$\begingroup\$ I don't know how to word the question differently. I did say that I added diode to electromagnet to protect relay contacts. So, there are just three components in discussion - relay switching the power to electromagnet and a diode to protect relay contacts from arcing. \$\endgroup\$
    – Maple
    Commented Apr 30, 2022 at 15:57

2 Answers 2


Snubbing is not a problem at all.

The core is magnetized. To turn such an electromagnet fully off, you need to apply a demagnetizing waveform.

Typically, you’d attach a capacitor in parallel with the coil, and then disconnect the power source, eg. using a small relay, vs. a mosfet or a transistor that would go into reverse conduction.

The current will decay as a sinusoidal waveform with exponential envelope. It will demagnetize the electromagnet. That’s how it’s typically done, and it’s very simple.

  • 2
    \$\begingroup\$ exactly. But please expand the cap selection because you can easily blow up the cap. It has to store the full inductive energy. and it shouldn't be too large in C in order to have reasonable high Q (actually oscillate a.few times) I.e. a smallish cap with a high voltage rating \$\endgroup\$
    – tobalt
    Commented May 1, 2022 at 5:48
  • \$\begingroup\$ I have mentioned RC + MOV combination as one of the options I was considering. I hoped that LC will oscillate and demagnetize itself, while being much simpler than externally supplying reverse power as @Tony suggested in his answer. Will the RC+MOV combination behave similar to the capacitor alone? My understanding is that it will have a) limited voltage swing due to MOV and b) faster decay due to R, but otherwise it should oscillate just the same as capacitor in your suggestion. \$\endgroup\$
    – Maple
    Commented May 1, 2022 at 16:57
  • \$\begingroup\$ I'm not sure what sort of decay times you had in mind, but we're talking seconds, and the MOV is not necessary at all. You may have a resistance in series with the capacitor to lower the quality of the tank and make it decay faster. Too fast will not demagnetize as well as it could. Design for oscillation frequency of a kHz or so, and then the decay time constant can be 0.1 to 1s. That's plenty fast for this application I'm sure. Another option is to use an electromagnet with a core that has no remanence. \$\endgroup\$ Commented May 2, 2022 at 11:35

This is not a simple Inductor with 100 Henries and 40 Ohms DCR to yield 24V^2/40ohm =Pd=~14 W because that (assuming huge L ) has a time constant of L/R= 2.5 seconds which does not account for your time measurements of several minutes after removing power.

The diode resistance is negligible compared to the DCR which limits power in the coil during Power On or time constant switch release.

So what causes the time delay?

Core Remanence will be huge leaving it almost permanently magnetized until the charge decays. i.e a semi-permanent lossy electromagnet.

How can you cancel Remanence?

Apply and opposite energy or equal power x time of 14W for stored energy after L/R seconds stored in E=0.5LI^2 ( L is unknown ).

What voltage and duration is that? (Hypothetical answer)

Without L value and remanence values, it will be trial and error but could be -24V for a few seconds on the order of L/R =Tau seconds, or perhaps a bit less time than it takes to activate the electromagnet to full current.

Tip: Measuring the magnetic force with some sensor helps to determine this and expect thermal variations in results.

How can you realize this?

With a timed "full-bridge" and bipolar diode snubbers.


simulate this circuit – Schematic created using CircuitLab

How do you design this more elegantly?

That's up to you. A one-shot timer switch springs to mind. Small RC snubbers can assist if the switch is slower than the ionization-time of an arc in microseconds)


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.