Smart plug relay sticking for A/V devices

I have a couple A/V devices powered via a smart plug. After a few months those smart plugs start to fail: relay stays in on state (sticked) and can’t be turned off. Kicking it gently a few times resolves the problem for a while but eventually it comes back pretty soon.
I’ve tried various brands of devices and it always results in the same problem (Tuya WiFi, Tuya Zwave, Blitzwolf Zigbee). I guess this is linked to the load type of those A/V devices (maybe the big transformer in the Yamaha amplifier?).
Is there any other device that would be proven to work reliably for a long time with such load?


Have you taken any of the failed devices to pieces to check for a switching relay and ideally the internal contacts? If you hear a when switching it’s a mechanical relay, which can indeed weld shut from arcing contacts from a high current inductive load like an amp transformer. “Percussive maintenance” releasing a contact also sounds like arcing.

Two options come to mind:

  • Move from a cheap relay likely rated for resistive loads to something bigger, ideally solid-state, rated for inductive loads
  • Suppress or snub the back-EMF from the inductive transformer and reduce the arcing on relay contacts

Low-cost “stupid” plugs typically have low-cost mechanical relays rated for resistive loads, not inductive. More expensive kit likely has better parts, and you might have options for a higher current device. For instance, in the UK a typical 13A plug switch might be actually rated (and fused) for 10A or less, but there are devices that are rated at 16A and above (for water heaters).

If you have the skills, a replacement up-rated relay from the likes of Omron might also work well (guess which manufacturer gets cloned? At least it makes replacement easier…).

As for a snubber or suppressor, it’s hard make a recommendation without knowing any details of mains voltage, device characteristics, etc., but a consumer grade “mains filter” is probably just a network of MOV (metal oxide varistor). Sadly these can only take so many “hits” (which is why better filters have a “working” LED) so a suppressor module might be a better choice (N.B. not a power factor compensation capacitor - that’s something else). Look for “snubber network” - typically a resistor and capacitor in a sealed box.

If this is a regular issue, there are higher power solid-state relay modules available (like a big relay / contactor but electronic) which could be switched from an ESP or in series with another standard “stupid” switch.

#include standard electrical safety disclaimer here - especially if the load has a back-EMF kick.

If this helps, :heart: this post!

Thanks for quick response. No, I have not dismantled the smart plug. Actually after kicking it (or maybe even without it) it works ok on other loads (e.g. phone charger).
The load is rather low, so it is anywhere near the rated max load. It is between 20 and 150W on 230V AC. The two biggest devices are Yamaha Amplituner (it takes ~70W when on) and LED TV (another ~60W when on) plus a few other, smaller loads (e.g. SAT TV set top box, Blu-ray player etc.)
Investing too much into this does not make much sense as energy cost savings would be too low. I might try with the RC Snubber (like the one from Shelly) to see if it helps - I guess that’s an example of “snubber network”?

It’s not about the steady-state current - back-EMF even from a small inductor can generate a spark so you can imagine the impact on contacts. If you’ve heard an amp “thump” when switched, it’s a similar thing and inrush currents can be huge. The first radio transmitters were basically just a relay used like this.

Inductive loads can easily de-rate switch current by half or even more compared to resistive.

The Shelly part looks like a standard Zobel RC network (0.1uF + 100R) with leads soldered on for easier consumer use. It should be used inside an insulated case but does look the right sort of device to reduce arcing.

1 Like

Cool. I’ll give it a try as it is still pretty cheap solution. Will report back how it went.

Sorry to necropost, but I have now had three of the Hue Smart Plugs that have failed in the on state with sticky contacts. They typically get between 1 and 10 on/off cycles/week. They’re all there in an effort to reduce the standby power: it’s not far off 100W 24/7 otherwise, particularly the lab server rack which has some beefy servers with OOB management that need powering when “off”, a SAN, 10G switches etc. No need to have these slurping on standby as it’s lab kit, not 24/7 ops.

I use the smart plugs to control:

o Desktop PC setup
o Electronics LAB (I’m an EE)
o Lab server rack
o AV equipment
o Gaming PC setup
o EV charger.

The first three have all failed in the past six months with the same failure mode (see OP) having been in place for up to about a year.

Sometimes they might hobble through after half a dozen on/off cycles for a few days, but in general it’s time to replace.

Unfortunately, these units are, as far as I can tell, completely sealed, and nothing short of getting medieval with a Dremel is going to get inside them. See below, I managed to get in non-destructively.

I’m gradually switching over the Innr smart plugs as a result of the Hue failures: these still work with Hue as well as through HA. I haven’t had any failures yet with the Innr units, but it’s a bit too early to say.

I wish they’d use solid state relays: certainly if I could take them apart, that would be the mod I’d make. Edit: solid state relays for this kind of current and voltage switching are prohibitively expensive for a consumer device at around 10x the price of a mechanical relay.

I can see a possible mod is to take the innards out and install them in a power strip for example, but the original enclosure will certainly be toast.

So a word of warning is not to be dependent on these smart plugs, at least not the Philips ones.

Hue Smart Plug Teardown and Repair - sticky relay

Video here:

I have managed to take one apart, and I think non-destructively.

There is a seam all around the edge on the plug side. You can get a reasonably sharp metal prying tool in there and gradually work your way along each of the four edges, introducing slightly thicker metal prying tools as necessary.

Once you’ve relieved the straight edges, concentrate on one of the corners, using prying tools to gradually lever the corner out from the straight edges. Once one corner’s out, the rest come easy.

That took me about 20 minutes to figure out. I think I could do another in 10 minutes now I know what to do.

Prying with metal tools will leave tell-tale marks, and you will need to use glue to put it back together., as this is not a snap-on joint.

Once you’ve got inside, to remove the relay, first you’ll have to undo one screw on the PCB and unsolder the two lugs going to the live and neutral of the plug: these will need a soldering iron with a larger bit and a decent amount of power: it took a short while with my 80W Weller and some desolder braid. In the end I gradually prised it out, applying heat alternately to each lug.

The relay has four contacts, being SPST: one pair for the 5V coil, and another pair for the switching contacts. The model is American Zettler AZ9481-1AET-5D. I don’t have anything in stock with the same pin layout, so I’ll have to order some in.

1 Like

The American Zettler AZ9481-1AET-5D aren’t widely distributed, but the OMRON G5CA-1A-DC5 are, and have compatible dimensions, so I’ll try those. One negative is that they aren’t spec’d to 15A 250VAC, only 10A 250VAC, but that will be fine for my applications.

Update 24 October 2023:
The Omron G5CA-1A-DC5 relays arrived today, and I’ve replaced the relay in one of the broken modules. Fits like a glove, and works very well.

Cleaned off the flux residue and superglued with one dab in the middle of each edge, using clamps while the glue dried: you don’t want this pulling off when you unplug it, but equally in case you need to get in again you don’t want unnecessary hassle.

Time will tell…

Thank you, very helpful. Would be great to see the internals of the welded relais contacts, especially to see what kind of material is used; nickel with something I presume