Warning - MHCOZY Zigbee relays

I have previously recommended the use of this MHCOZY relay that is available in the following form-factors:

  1. Single channel relay
  2. Dual channel relay
  3. Quad channel relay

I have been using a few of these, starting with a single channel relay about 20 months ago.

Two or three months ago, the single-channel relay started misbehaving in momentary mode. When flipped to on, they relay would turn itself off within a second (as anticipated), but then flip rapidly between on/off about 4-10 times (variable number of on/off repeats). This behavior didn't change when I reset and re-paired the relay, so I stopped using it.

Today, the dual channel relay started behaving in exactly the same way. I'm not using that relay with Hubitat, so whatever the issue is, it is unlikely to have been caused by a Hubitat platform or zigbee update.

Bottom line - I don't recommend them anymore.


I think anything using the "Songle" relays are going to have reliability issues. I frequent the Arduino forum and have read many issues due to failure of this relay. (i.e. the blue 5V relay.)

Relays being mechanical devices are really quite complex and some of the important areas are prime for cost reduction. Contact material and plating thickness is maybe the most important.
Its difficult to imagine but on each current carrying contact closure, very small areas of the plating (where the metal actually is in contact) melts, softens and "mushes" down to allow more contact area.


I had 2 Zooz power strips that started doing this, rapidly activating/deactivating relays and I wonder if the relays are from the same company? Sounded like a machine gun.

1 Like

That's what it sounds like up close. Anywhere from 4-10+ seconds of rapid relay clicks.

I literally just bought one but it has yet to arrive. Could you recommend something else?

It will be used for a momentary push on a driveway gate. The device also needs to be powered via low voltage.

I have and use Sonoff ZBminis with success but believe they need line voltage.

EDit: I have a Zen 32 downstairs I think. That should work.
2nd EDit: its a Zen 17 but i forgot its Z-Wave. My Z-Wave network isnt large enough. The gate is 500' from the C7 hub. Surprisingly my zigbee network reaches 2/3rds of the way. Was going to just add a Sonoff wall wart to an outlet I have way down there on pole to reach the rest of the way.

Hopefully these cheap relays I got on AliExpress dont suffer the same fate. Any way to find out what relay they are using without bosting one open? Need recommendations for unfinished basement lighting - #18 by jtp10181

Those are all line voltage relays. I use sonoff ZB minis in that application and they work great. Neutral wire is required though.

Here we are talking about low voltage. Usually 5V to 24v.

Curious if from your reading what the "favorite" relay manufacturer is. Those Songle's are everywhere.

Kinda bummed seeing these MHCOZYs falling into failure.

Ironically I'm here re-reading relevant threads because a couple of long time reliable Zigbee devices disconnected from the mesh (as if their batteries were dead) within a couple days of just putting a 4ch MHCOZY online nearby.

Batteries weren't dead....got one device to come back (placed it well away from the MHCOZY just in case). Will play with the other one tomorrow.

This is the second time I've brought up an MHCOZY and had other things go wonky in the mesh. Yet I have a single channel MHCOZY that works flawlessly in another area.

I wish I could quickly see (conclusively from Zigbee radio data) if there is any correlation without a lot of trial & error.

Finding some good discussion on this relay topic thanks to your reference. Good to find this out now. When similar functionality costs 3-5x more from a familiar brand manufacturer sometimes there's a good reason.

You can find ZooZ Zen17's regularly on sale for $20. They somewhat fill this void but they are Z-wave and only have 2 channels. They do however have sensing inputs as well.

They are pretty rock solid when they work. I have had about two out of six fail on me though in one way or another.

It would be interesting to replace a Songle with an Omron G5LE-1 DC5. From a quick look, the pins have a similar layout, however I'm not sure of the actual dimensions.
Omron is a commercial grade relay with a good reputation.

1 Like

Zooz has said they are working on a 3-relay version of the Z17 -- perhaps as a successor to the Z16? Maybe 800-series? Dunno...

But it would be cool to have available!


I recently installed the 4 channel flavor and haven't had any problems with my ZigBee network but I'm still on a C7.

Does anyone know of a way to be able to have more than one of the relays on at the same time? On mine if you turn any relay on it turns whatever else is on to off.

You using the Generic ZigBee Multi Endpoint Switch driver? That's what I'm using and mine do not have that problem.


Thanks, and yes I'm using that driver.

I've setup this device. It works fine for a few hours, but them stops responding to commands send from the hub (device page). If i physically switch the relay button on the device itself, the status does update on the device page. Any ideas why this turned into just one way? when using the device page, hubitat shows the on/off in log, but the device doesn't seem to get the signal

just for giggles. i just turned off my prod c-8 hub and restarted the device and it started working again on my dev hub c-7. Do you think that bad things are happening with both hubs running?

I have no idea. I do know that the Songle relays on these seem to have a high failure rate (described by @JohnRob above).

The MHCOZY relays I used worked reliably when they were new, but every one of them failed within 2 years with the symptoms described above.

I wonder if a capacitor is going bad on these, and rapidly rebooting the processor? That would be relatively easy to test by replacing those 2-3 caps.

My other theory would be the filtering is just poorly designed in the first place and the relay is spiking the processor. I have scoped many automotive relays, and there are some without adequate snubber circuits on the magnetic coil side of the relay that can produce over 50V (on an automotive 12v relay!) when the magnetic field collapses upon turning off the relay. That spike, while brief, can cause the microprocessor's brains to become scrambled. I have seen numerous processors damaged on vehicles due to this phenomenon.


Don't think it is the capacitor. While I've thrown my relays away, I know the caps were not bulging or leaking. OTOH, I didn't de-solder them and test them.

I'm guessing your other explanation is probably correct.

1 Like

The failure due to inductive kickback should be clamped by the relay driver, a ULN2003 with built in freewheeling diodes on the outputs.
Here is the datasheet:

I am going to scope the relay coils to see how bad the spike is.
Hard to believe that the diodes aren't clamping the spike, but only a scope will tell!

A thought about a possible problem is the board layout....
The connection to pin 9 of the ULN2003 (freewheeling diode common) may be spiking the processor's 5 volt rail because of trace impedance and/or routing.

A possible "fix" might be to add a piece of wire from the pin 9 directly back to the 5 volt supply and NOT through the trace to the processor's 5 volt rail.

Needs further investigation!