Shelly 2.5 getting hot to touch (63°C external case) - should I be worried?

Hardware
4

I repeated the temperature testing of the Shelly 2.5. It was cooler in my garage today so the figures are a bit lower than yesterday.

No load, Ambient 25°C
Measured external case temp = 47°C = 22°C above ambient
Internal temp sensor reports 58°C = 33°C above ambient.

Load of 2x 8W LED down lights, Ambient 25°C
Shelly 2.5 has two outputs, connected one 8W downlight to each output
Temperature stabilised after about 1.5 hours (see screen grab)
Measured external case temp 57°C = 32°C above ambient (well below the 45°C allowed by IEC61010)
Internal temp sensor reports 72°C = 47°C above ambient.

Screen Shot 2020-03-01 at 3.32.10 pm
Screen Shot 2020-03-01 at 3.32.10 pm728×936 61.3 KB

Obviously this is limited testing using a thermometer with ±2°C accuracy. And it should really be done at full rated load of 10Amps (maybe it would fail the temp rise limit then), but I’m more interested in LED downlights.

So with an LED downlight, if the inside of my walls could reach 40°C (unlikely but possible), then the external case would reach about 40+32=72°C. And the Shelly internal temperature would be 40+47 = 87°C, still below the 90°C at which point the Shelly will automatically turn off.

I also measured the downlight temperatures too, with the following results:

Brilliant smart WiFi LED downlight (the one on the left hand side in photo above), at full brightness after 1.5 hours, measured 61°C, temperature rise above ambient of 36°C.

Arlec dumb LED downlights (the two I used to load the Shelly, on the RHS), full brightness after 1.5 hours, measured 55°C, temperature rise above ambient of 30°C.

Conclusion
Not sure what to make of all this, it seems that the Shelly 2.5 easily meets the IEC61010 standard, and the LED downlights themselves are putting out similar levels of heat. Maybe I just need to accept that we regularly put “hot to the touch” stuff in our houses?

5

It’s a pleasure to read someone who a) understands the concerns b) has a testing plan c) carries out that plan in a logical consistent manner d) has read up on the relevant standards

I can’t fault anything you have done.

My first thought was that you should ‘wrap’ these devices to protect from accidental contact BUT that would just keep the heat in so DON’T do that.

My second thought was “what do my dimmers run at” I use fibaro z wave dimmers and I took the switch face plate off, moved the switch to get at the dimmer, it was cold (ambient) which is what I expected for off. On full power (I only left mine 40 mins) it was moderately warm but easily not warm enough to cause any concern (touch is my only test method at the moment but the central heating runs at 65 C and it was nowhere near that, exit pipework) so then I ran it (40 mins) at 70% (as that should be about wosrt case) it felt a bit warmer (subjective or piling on from the last test ??? ) but still not hot enough for concern.

Two things strike me, 1) the spiders in your service spaces are going to have a toasty place to live. 2) How much energy are you throwing away in heat ?

Maybe these are acting within specification but I would not buy these or use them if I had any choice.
You will have to assess each usage case and installation scenario to determine if you are happy, your approach and methodology tells me you will make the right choice.

Comments:

  1. Bulbs have always been major heat sources gu10 types especially.
  2. LED ones run a lot cooler than their predecessors so a move in the right direction (but can still be surprisingly “warm” ).
  3. I couldn’t work out why the gorilla pod was there, then I saw it was a metal filing cabinet. I usually steal a glass shelf off the cooker hob (Ikea halogen cover) or a wooden chopping board. 4. Accidental cable movement can cause unfortunate stains in your underwear or worse destroy the kit you are testing
  4. I notice the devices bear both CE and RoHS markings and your sockets indicate you are in Aus or NZ so it “should” be compliant. (I know you know this).

There is a thread here on “electrical kit certified for use in Aus/NZ” I suggest you search for that and have a read, though it is a LOOONNNGG read :rofl:

Good Luck, Happy Automating

1 Like
6

Situation: The temperature of a lightly-loaded (just 3 LED lights) Wi-Fi controlled relay rises more than 35°C above ambient temperature.

First-order analysis: it’s either defective or there’s a design flaw.

I have several wall-mounted Wi-Fi switches, all lightly-loaded, and they get barely warm to the touch. I have many more wall-mounted UPB switches (not RF-based) and they too get barely warm. In contrast, a case temperature of 65°C is substantially more than barely warm.

FWIW, the device may ‘fall within specs’ but compared to other choices, it runs hot.

7

maybe someone can compare Shelly1 to Shelly2.5?
Temp grow of Shelly1 working without load is not recognizable with naked finger. Shelly2.5 is significantly warmer. I don’t know which part of shelly circuit is responsible for this. Maybe it’s related to current/power measurement (?)

8

I have a few Shellies in use. This is what they report as their internal temp:

Shower Fan and Light (combined) SW0 and Nightlight SW1
Ambient: 69 F / 20.56 C
Shelly 2.5 no load in 3-gang box with plenty of spare room: 110.8 F / 43.78 C

Toilet Exhaust Fan and Light (combined) SW0 and Nightlight SW1
Ambient: 69 F / 20.56 C
Shelly 2.5 no load in 2-gang box somewhat cramped: 123.1 F / 50.61 C

Dishwasher
Ambient: 60 F / 15.56 C
Shelly 1PM no load: 99.5 F / 37.5 C

Hope that helps.

1 Like
9

ok my post is long, but it was actually a separate downlight (Brilliantsmart) that rose 36 above ambient.

I also tested the Brilliantsmart smart dimmer mech, and some other Wi-Fi switches, and same as you, they barely get warm.

But to your other comments, I tend to agree, the bottom line is it runs hot on a light load (2x LED downlights) and even if it complies with a standard, I’m not comfortable with it.

10

exactly, this is why I started investigating Shelly 2.5 as it was noticeably warmer to the touch. I believe it is a combination of the CPU and currnt/power measurement.

11

Thanks @Mutt appreciate that. I also have some Fibaro z-wave dimmers / switches which I should test.

Maybe these are acting within specification but I would not buy these or use them if I had any choice.

I agree, this is basically my conclusion. Problem is I don’t want to battle with z-wave any more, so trying to find a Wi-Fi equivalent.

I’m also aware Shelly are not certified for Australia, but there aren’t many options. There are some Zigby devices new to the Australian market, I might get some and test them. I just really like the responsiveness of Wi-Fi. I wonder if Zigbee is as good once you have 20+ devices.

The gorilla pod was there to lift the Shellys up off the cabinet for airflow, like they were hanging in my wall space. BTW the cabinet is Ikea and made of wood, not metal! I wouldn’t put this rig on a metal cabinet :slight_smile:

Yes I read the Aus/NZ certified thread very often!

12

I’ve just installed a Shelly 1PM to control a 3.1kW immersion heater. overnight run when the Octopus Go rate is only 5p/kWh (cheeky referral link). It’s sitting in a large single-gang back box behind a single UK socket.

I’ve only just set up the internal temperature sensor as a Home Assistant sensor, so we’ll see how hot it gets after the overnight run. From a short test, it looks like it’ll approach 75°C at ~13A. It’s sitting at 47°C switched off in the ambient of a warm airing cupboard, probably about 28°C.

13

That’s hot.

To serve as a point of reference, for over ten years I’ve used an X10 Heavy Duty Relay Module to control my 220VAC 1.5 HP pool pump motor. It’s CSA approved, rated to handle 20A, and has run reliably for over ten summers. It barely gets warm to the touch.

If a device meant to just turn on a load runs so damned hot (75°C) then it’s the wrong tool for the job.

14

When I get home from work, the engine is burning hot, yet I can touch the hood without any pain. The shelly sensor measures temp of the PCB, not the outside of the case. 70C isn’t too much for a PCB in general, but depending on the thermodynamics at play inside the case, there may be nearby components that have lower temp ratings. What are those ratings? Even cheapo electrolytic caps usually are ok to 85C. YMMV, but bottom line is if you flash a even a UL listed shelly, they’re not going to be liable if your house burns down lol.

1 Like
15

If you’re going to use a car as an analogy then let’s do it right. Use a stock passenger vehicle pull a large house-trailer from Denver up into the Front Range while maintaining at least 60 mph. Then return to Denver. Let’s see how its transmission, cooling system, and brakes handle that.

The Sonoff behaves just like that passenger vehicle. It can do the deed but not without suffering. For heavy-loads, a purpose-built vehicle is the better choice (or, in this situation, a device with a robust relay).

My “barely gets warm to the touch” reference was a shortcut to saying the internal operating temperature of this sealed device (X10 relay) is so low that it’s case barely gets above ambient (and that’s not due to any insulation like on the underside of a car’s hood or assisted convection-cooling like in a moving car).

16

That passenger car trailer analogy would be closer to running a stick welder off a shelly, not some new led bulbs. Anyhow, I agree that the tool has to fit the job. The shelly’s are not rated for big amps though, and they do fine within their rated specs. If we needed to switch a 20A load, yeah the shelly probably isn’t a great idea, but that would be no different than running 14awg to a 50A breaker… no electrician would install it like that, and hopefully no DIY’er would either.

[edit: Note that shelly 2.5’s are only rated to 10A (per output)… lol when calibrating mine I went way past that. However installed, the biggest draw I’ve got on a shelly 2.5 is 700W. Even then they do heat up a little bit… definitely not the same hunk of meat as an industrial relay, with the power monitoring resistor and whatnot inside, and a super insanely compact PCB that can’t have much copper area.]

17

User Troon is running a continuous 13A through a self-certified device rated to 15A. That’s 87% of its maximum load for a device that self-destructs at or near its rated maximum (see image in first post and in related topics about melting Sonoffs). The car hauling a heavy load isn’t too far off the mark, with the exception of turning itself into a molten blob.

If Sonoff wired a house, they’d do it without fuses or breakers, use 16 gauge wire, and claim everything is (self) certified to 15A and 90°C. Not to worry if the house wiring runs hot because, hey, it’s rated to 90°C.

/s

18

Trug, read the top of this thread where patfelst did tests measuring the external temperature. So there is no ‘hood’ to protect your wittle fingers from the incandescent dwiving thingy.
The devices are pretty tiny anyway so actual insulation is negligible.

20

17 minutes running my 1PM at 12–13A (depending on line voltage) last night. Looks like “terminal internal temperature” would be around 82–85°C if run for long enough, just going by the shape of the graph.

Worth noting that the Shelly’s relay remains on until 04:25 but the immersion 'stat has opened at roughly 04:02, cutting the current. You can see the cool-down temperature gradient steepen after 04:25 when the Shelly relay also shuts off.

image

1 Like
21

By self destruction, you mean an internal temperature of 85C? Also, I can’t seem to find that image of a melted Sonoff in this thread or related topics. I don’t doubt that there are pictures of burned up sonoffs, or pictures of burning fire extinguishers on the internet. I’m not trying to an arse here… if it comes off that way, I’m just pushing for the truth.

1 Like
22

Yep, if I had house wiring getting near 90C I’d be quite worried. That would take quite a few amps though, and the breaker should trip before that. In the case of a PCB inside a shelly reaching 80Cish… I think that’s pretty safe don’t you? Yeah 50C would be even better, but price/size would go up for sure. I bet there’s a market for something bigger/pricier though, to be honest I’d be game for a better ‘in j-box’ device.

Regarding sonoff wiring a house… lol yeah I literally wired my own house from the areal to every single outlet/switch/light… all the low voltage too. So yeah I don’t need the help of sonoff’s engineers. I also modified some sonoff R2 devices… they aren’t all that bad, but common sense says those tiny terminal blocks on all these things (shelly’s included) aren’t going to stay very cool at 15A. That is where the industry needs to really change… come up with a better screw terminal for cripes sakes!

1 Like
23

Self-destruction as in: Self. Destruction.

The “hack” that was implemented (see topic) wasn’t the cause, it was the fact the device lacks any means to fail gracefully in the event of overcurrent and/or excessive heating.

The truth is these cheap, self-certified devices are probably fine for switching on LED lighting but are woefully underdesigned for handing high-current loads. Nevertheless, that’s how many people use them and then rationalize it by quoting the device’s specifications (which amounts to relying on materials-failure as opposed to a proactive, graceful shutdown).

EDIT
Google ‘burned sonoff’ and check the images. They don’t all fail as catastrophically as in the image above; other examples involve the localized failure (re: thermal destruction) of traces and/or individual components. Basically, in lieu of any self-protection it relies on its components to also serve as fuses.

24

I would actually argue that running a heater off a 1n4007 was the cause, no?

I agree that the truth is they’re cheap “self-certified” devices if you will, and that there will never be a shortage of folks who misuse them. I just don’t want to spread fear that sonoffs are completely unsafe handling jobs that lie within their specs (under 10A). Yeah, I’d like to see some protective devices in there too (varistor, fuse, etc), but I know myself how to keep 'em safe as is. It’s implied that when you cut into a mains wire and connect it somewhere, you better know what the heck you’re doing!!! The author of that post you linked most certainly does not know enough to be doing what he did safely. Sonoffs are definitely included in this category, so having a UL stamp etc is almost irrelevant (I can wire a UL stamped outlet in a way that catches fire too).

I also have some DIY esp devices that control similarly ‘poor quality’ overseas imported relay boards (I’m all aboard with doing something about the quality of our supplies… dreaming of what a made in usa sonoff might look like). Those boards are spec’d at 15A, and have 15A relays, but no way I’d trust the tiny copper trace to handle that for more than a split second… so they got ‘self-certified’ at 8A based on my measurements of the traces and online ampacity tables. :wink:

1 Like