Sonoff TH16 and Sonoff POW are inexpensive and useful wireless switches based on ESP8266 WiSoC. I’m using Sonoff TH16 to control a water pump, and Sonoff POW to monitor my office power consumption. I received mine in early November 2016, so they were probably manufactured sometimes in October. Does manufacturing month matter? Yes, it does, as ITEAD Studio has just issued a recall notice for both devices for a batch manufactured between December 2016 and January 2017.
The problem is that while the switches come with a 16A relay, the trace were not thick enough, and if you connect a device that draws enough power, they would heat enough to make the case slowly melt. The problem is that ITEAD Studio asked the factory to add “sufficient tin to the wires to ensure low enough impedances”, but they did not quantify it clearly, and that’s why the product is not safe to use.
If you are affected, the good thing is the company owed their mistake, and you can contact the company for a refund or replacement by providing a photo or video, as well as your order number. What they did not say exactly is how to identify a product with the defect, that is before it melts…
In order to avoid the issue in the future, they’ve “optimized their workflow”, and “are seeking for better solution to improve the product during the manufacturing process”.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
and thats why you never use chinese ac stuff.
@TC
WE STAND BEHIND OUR PRODUCT.
Think about BOOM7 from Samsung… even such huge company makes mistakes, and Samsung even do not willing to voluntarily admit their mistakes and correct them in time.
But, WE DO.
BR
Jerry
Member of ITEAD
A huge problem with chinese ac-mains product was always, that most companies only test for the US. this means 110V @ 10A
there was an article from a dutch guy who tested usb chargers and showed how traces and spaces were designed spot on to pass 110V safety but no tad more, and that was usually the cause for accidents on european lines
if you really care about the EU market, do the tests by ‘TüV’ also. Germany is most strict on this (customs will destoy any high voltage stuff without ce mark + certificate)
Do you think this is only related to Dec16/Jan17 stuff, or also the previous?
Exactly why these devices need to do UL or ETL testing. Trace size and spacing is a specific item that is checked in the UL/ETF tests. Screw this up like they did and these devices will start fires and burn people’s houses down.
Another common failure that is checked for by UL/ETL – AC traces too close to each other. In humid environments (say India or CA during monsoon season) humidity will condense onto the PCB. If the traces are too close together this will start fires and burn people’s houses down.
Don’t buy AC mains stuff that does not have UL/ETL testing. ETL has a nearby Shenzhen office, I suggest ITEAD make use of it. ETL is much less expensive than UL.
PS – adding tinning by hand is an awful way to try and fix this since hand work has a lot of variability. Instead increase the oz of copper on the PCB. But of course if you do that you’ll have to rework the antenna design which I don’t know if ITEAD has the training to do.
A simple way to fix this would be to increase the copper oz on the base PCB. Then mount an ESP-12 module onto the base PCB. The ESP-12 modules already have FCC. You still need to do FCC conducted emissions but that is a cheap test. Or you could skip the module and rework the antenna to work with the thicker copper, but that takes more skill.
There are many PCB calculators on the web that will tell you how thick and wide the traces need to be to carry 15A.
You will still need to do the ETF testing.
Good to know, I marked them as “10A” only, this is fine for me. Otherwise I’ll add more solder to the trace.
@Gaetano
I have some pictures of mine with the traces @ http://www.cnx-software.com/2016/11/04/sonoff-pow-and-sonoff-th16-wifi-relays-review-part-1-the-hardware/
It does not look that different. Hard to say.
There’s an easily replaceable fuse inside the device. It might be a good idea to replace the 16A fuse with a 10A fuse.
@Jon Smirl
We have the right engineer, our CTO, the designer of first RFID-SIM (mobile Shenzhen Tong), to rework the antenna. It’s not difficult.
https://en.wikipedia.org/wiki/Shenzhen_Tong
But sadly, simply add the thickness of copper from 1oz to 2oz will do no help. think about that, for a 16a current, 0.17u or 0.34u thickness copper makes no difference.
Surely we will obsolete the “adding tinning by hand” method… i will publish the report when we come out with a solution….still need more test.
Regarding the UL/ETL test, we will do that step by step. You know we even do not have CE before, but we do the testing for new products.
We can only take one small bite of the elephant at a time.
One option is to use big external contactor (or relay) in such cases.
Calculator to help figure out your PCB traces.
http://circuitcalculator.com/wordpress/2006/01/31/pcb-trace-width-calculator/
CE “china export” ?
You can double the trace with a soldered plain copper cable suited for 16A.
But using any relay board at its relay component limit (16A) is always a bit risky and reckless.
This is valid for any product and not only for sub-10$ product like sonoff.
Full support to ITEAD on this. Just retag the product limit to 10A. Nobody cares.
sonoff are great but you need to be a bit aware of what you are doing when playing with mains.
Well at least when the house burns down there is no need to worry about the cost of the light bulb being on 🙂
So you didn’t have CE certification but that didn’t stop you slapping the CE symbol on the devices.
@Jerry Shi
Jerry,
Thanks for your follow-ups here.
Is there a date sticker on the units, or some other method to help identify the affected modules before they actually cause a problem?
I plugged that trace into the PCB calculator – I guessed 2in long and 1/4 wide with 1oz copper. Calculator says it will be 60C over ambient. This is an enclosed box so it does not radiate heat very well. Since it can’t radiate the internal temp will just keep rising. Temp inside the box has exceed 125C which is melting point of the plastic. Very close to the ignition temperature of wood/fabric (160-180C). Given that the plastic is charred it may have exceed 250C. Need to measure it to see exactly how hot it got. This is definitely something that could start a fire.
That trace probably needed to be 3oz copper which would heat up only 5C. Or you can use 2oz copper and make it wider.
You can make an AC switch like this much more cheaply by using something like a Monolithic MP157 and a hot ground power supply. You can eliminate the transformer. But… when programming the flash you will have to power the unit using an isolated power supply
@Jon Smirl, your MP157 is a huge unsafe energy waster at 100mW no load (advertised). Itead uses Dialog Semiconductor’s iW1700 in their power supplies with 5mW no-load (advertised): https://www.itead.cc/power/ac-dc-power-module-5v-500ma-v3.html . Which shows that they care about the stuff they make and don’t go for cheapery-cheap. The recall discussed is also rather positive than negative event, showing again that they care, and that’s the first such case for a Chinese company I remember at all.
I’m on for more of their stuff, but of course good luck to folks who’s going to run 16A thru *any* PCB traces, Itead’s or not.
https://ac-dc.power.com/products/linkswitch-family/linkswitch-tn2/
That is 30mW no transformer but it is a dime more.
Not obvious which is best for the environment – it takes energy to make the transformer. We’re only talking pennies of electricity here.
@Paul M
@The Cageybee
They have CE certification -> https://www.itead.cc/wiki/images/f/f7/CE_Certificate_for_Sonoff_Series.pdf
However this only has to do with interferences, nothing to do with safety.
You need TUV for safety certification in Europe, ETL or UL certification for the US as mentioned by other commenters.
Hey Samsung started a trend up and behold Sonoff POW and Sonoff BOOM 😉 Atleast ITEAD is honest about this, recalls their devices and unlike Samsung actually knows what the problem is and does not repackage reship the same faulty devices as replacements…
does sonoff pow has a temperature sensor ?
@cnxsoft
I’m afraid CE covers a lot more than interference.
From Wikipedia:
“When a manufacturer puts the CE marking on a product it implies that it complies with all the Essential Health and safety requirements from all the directives that applies to its product”
Also, from Jerry Shi ” You know we even do not have CE before, but we do the testing for new products.”
So, straight from the horses mouth, he admits to not having actually complied/fulfilled the certification process. Kinda makes that CE document not worth the paper it’s written on.
@pandada8
No.
TBH I had reservations right from the start. Any consumer electronic device that encourages you to cut the cords on another device and wire them in by the consumer is probably skirting with the law. I’m not sure about the rest of Europe, but in the UK, cutting cords and wiring in oneself would be illegal, unless singed of by a government approved and qualified electrician.
If you install this kind of device without the necessary sign off, a persons home insurance would be invalid and could open you up to criminal prosecution if anything went wrong.
From the pictures (and looking at my devices from 12/2016) I would say that the tracks are not the problem. It seems that the fuse holder and/or the fuse itself generated extreme heat. And this is understandable – this type of sockets and fuses are not designed for 20A rating (or even 16A). I could imagine how this fuse had turned into light bulb for some time to melt the case.
For me this is the key – I will remove (jump) the fuse with 1.5mm2 wire on my devices (3 pcs, 12/12/2016) and will add external circuit breaker. The relay is also questionable, but maybe the easiest way is to short the contacts too. The shunt also must be checked if it can handle the current and the heat – I see 0.001Ohm on the schematics, smd 2512 – at 16A this is 0.256W so it is safe – at least if we don’t look at overload condition. Type C circuit breaker will tolerate some 200% overload for several minutes, so if one uses 25A breaker we might get some 50Amps for 2 minutes – “will it melt?” (sure, 25A over regular EU socket is not correct, but I often see it).
I sent a ticket to ITEAD 20 days ago about getting replacement units for my th16 ‘s and pow ‘s that I had recently received and still no reply 🙁