Multifunctional EM-126T Wi-Fi relay from Novatek-Electro

It's been quite a while since I last tinkered with Novatek devices, and now I've got several candidates for review.
Today we're looking at a multifunction relay that combines several devices in one. But, as always, more on that in the review.


The link in the title goes to the Rozetka store. For more information, here's the link to the manufacturer's website. For readers in Russia, it will be a different link, or this one. The price is 4938 RUB.

I'll get a little ahead of myself and say that, for the most part, this device is very similar to the previously reviewed EM-129. It's noticeably cheaper and more convenient for non-technical users, but has lower power and a different design.
I'll also note that I've been using the EM-129 for over a year now, and it's been behaving itself. At my request, the software has also received a couple of changes, which I'll get to later.

A little background. We have an old fridge at home that's been acting up a bit after many years of use. Ideally, we'd replace it with a new one or get it repaired, but something always gets in the way. Anyway, it has this "problem": it can work normally for a while, then it "gets stuck" and stops switching off. But if I unplug it for a few minutes and then plug it back in, it starts working again, switches the compressor off normally after half a minute, and goes back to normal operation.
This hadn't caused any particular problems until my wife and I decided to go to the seaside for a week. I thought I really couldn't leave things as they were. I solved the problem at the time using another device, also from Novatek, though it only supports local control over the network. Like the EM-129, the one I'm reviewing can connect to the Overvis cloud service, and I thought it would be handy to have something at home that I could control remotely if needed, while also using it for monitoring—just "in case."

To make it clear what we're talking about, here's a list of what the device can do
1. WiFi-controlled outlet, with local control from the front panel also supported
2. Thermostat
3. Timer, which can be combined with the thermostat, so you can set different temperatures for different times.
4. Protection against overvoltage and undervoltage, with five trip thresholds and corresponding response times. Two thresholds are adjustable.
5. Measures voltage, current, power (active and apparent), and frequency. It has configurable protection against exceeding the power limit and frequency deviation, and can also record all measured values autonomously (for about a month).
6. Tracks energy consumption in kWh, with total, daily, weekly, and monthly counters linked to the cost of electricity.
7. Can work either through the Overvis "cloud" or via a direct IP connection. Developers can download the API.
8. Wide operating voltage range, approximately 50 to 400 V.
9. Built-in real-time clock with up to 5 days of backup operation, powered by a supercapacitor and synchronized over the internet using the NTP protocol.


So, the first thing we see is a fairly large box. Inside, the relay and its manual sit in a cardboard tray.
The packaging lists some basic specifications and other useful information.


There are two versions of this relay: the EM-125 and the EM-126T. The first is a couple of dollars cheaper, while the second has a temperature sensor and comes in two versions, 126T-1 and 126T-2, with 10 m and 1.8 m cables, respectively. Mine seems to be some third version, since the packaging lists a cable length of 1.6 m.


As always, the package includes a very detailed manual covering both the device and how to configure it.
The warranty is 3 years, which is rather unusual for Novatek, whose devices more often come with 5–10-year warranties.
Naturally, the manual is also available in electronic form, which is what I usually use.


Specifications. There are a few typos, but it's generally clear. It also says that the built-in clock has 5 days of backup operation, with a footnote specifying that this applies after the device has been powered for 30 minutes.


It looks very neat: light gray, with a well-molded case, good-quality plastic, and well-fitted parts. Honestly, it's nice to hold.


1. A backlit button on top, used for control, configuration, and indicating the operating mode. It lights up red or blue. Personally, I think green would be less annoying than blue, but you can adjust the brightness, which I really appreciate.
Below it is the outlet, rated for a maximum current of 16 A. It has protective shutters.
2. The cable, with a temperature sensor at the end. According to the description, it's a 10 kΩ thermistor.
3. A sticker with a QR code on the side, used to connect to the Overvis service. It's odd that they don't include any spare stickers—what if the one on the side wears off, tears, fades, and so on...
4. On the back, opposite the socket, there’s a plug for connecting the device to the mains. And once again, I have to praise the build quality and the plastic housing: everything is very solid, nothing wobbles. Five stars.


The layout of the controls and the device dimensions, as shown in the manual.


Naturally, I opened it up. To do that, you need to peel off the side label and remove the three self-tapping screws holding the two halves of the housing together.


Inside the device are two circuit boards. The larger one has an STM microcontroller, EEPROM, an op-amp, a 1 mΩ current-sense shunt, and an ESP module.
The assembly isn’t exactly perfect, but it’s reasonably good.


The power supply is on the second board. I didn’t remove it, but it’s already clear that it’s probably built around an LNK-series PWM controller from Power Integrations, with no galvanic isolation. On the left, there’s also what appears to be a pre-regulator for more reliable operation at elevated input voltages. The device gives the impression of being pre-production, since at least one resistor has clearly been installed by hand.


1. The wires are crimped and look neat, but as a perfectionist, I can’t help noticing the yellow-green wire. Has Odeskabel really stopped making blue and brown wire?
2. An Omron relay handles switching the power circuit. Its rated life is 100k to 1M cycles, depending on the load. There’s also a 420 V varistor nearby (that’s the RMS voltage rating in this case).
3. The power board is literally packed with capacitors. You can also see the converter’s inductor and a polymer capacitor.
4. A 0.33 F supercapacitor is used as the backup power source for the clock. There’s another polymer capacitor next to it, which is nice to see.


Let’s move on to how it works and its settings.
Connecting it is extremely simple. I did manage to mess up the connection a little, but never mind.

I scan the QR code and follow the link to overvis, where my EM-129 is already visible. You can also check the electricity meter reading accumulated so far.
The device was added to the list, but naturally it told me I was doing it wrong and needed to configure the connection first. It suggested reading the manual—but who reads that before getting started? :)


I open the manual (thankfully, it’s already available electronically) and refresh my memory about the process, which, once again, is essentially no different from the EM-129.
Hold the button down for seven seconds, then release it. The indicator flashes blue and red. Open the list of nearby networks on your smartphone, find the one named EM-126T, and connect to it.


Once connected, open a browser on your smartphone and go to em.com, where you’ll find the device settings menu.
1. The device then automatically searches for nearby networks, but for some reason it doesn’t find my new, good router.
2. No problem—I enter the network name and password manually.
3. I save the settings. The device reconnects and suggests following a link, although that’s no longer necessary since I’ve already added it to overvis.


The indicator has quite a few different states:
blue (flashing every 0.5 s) – connecting to a Wi-Fi access point;
blue (steady) – load is on;
blue (flashes every 5 seconds) – power is on, load is off;
red (flashing every 0.5 s) – the automatic reclosing timer is running;
red (steady) – a fault has occurred, or the device has been locked out due to a fault;
blue-red (flashing every 0.35 s) – Wi-Fi connection setup mode is active;
blue-red (flashing every 0.1 s) – factory reset is in progress;
purple (flashing every 0.1 s) – firmware update in progress.

I also managed to put the device into an undocumented mode with blue-red flashing, but more on that at the end of the review.

And I’ll mention again the handy indicator brightness adjustment, which ranges from barely visible to quite bright.


1, 2. When you open the device, you land straight on the “Status” screen, which shows both current readings and the values accumulated by the counters. By the way, at my request, the developers did reduce the size of the on/off button. The second screenshot shows what it looked like before. The point is that my entire apartment is powered through the EM-129, and I think it’s obvious that an accidental shutdown wouldn’t be good—since the router would go off too... I also asked the developers to consider adding a confirmation prompt before carrying out an action, to completely prevent accidental presses.
The display also shows measurements of active and apparent power, voltage, current, and mains frequency.
3. Click in the upper-right corner to open the device options menu.

Yes, if the energy counters aren't counting, you just need to reset them once. This can happen with a new device.


Since I now have two devices, I can quickly switch between them. And because this is just a browser window, there's nothing stopping you from opening two windows at once.
On the right are the results from more than a year of operation of the EM-129.


But let's move on to the settings menu. Almost everything here is identical to the EM-129.
1. Set the device name and its access password.
2, 3. Set the temperature correction, hysteresis, and operating mode—heater or cooler.


1. Network connection options. In the new firmware, the developers added automatic Wi-Fi reconnection. This caused problems connecting to my repeater, so an option to disable this feature was added in a later software version. By the way, the device downloads firmware automatically, then notifies you when an update is available and offers to install it, so the main thing is to have internet access.
2. Time synchronization settings. Nothing unusual here; you can also sync the time with a PC or smartphone.
3. Scheduled vacation. To be honest, I never quite figured out this feature—maybe because I've never been on vacation...


1. Enable remote access via the cloud. Everything looks fine, and the defaults are sensible.
2. Service settings. I left everything as is, except for the button backlight brightness.
3. Below are buttons for rebooting and resetting the counters, as well as a full reset procedure and firmware version information.


Like the EM-129, this device can also protect connected equipment. The available options are:
1, 2. Protection against high and low input voltage. You can set the threshold (240–290 and 160–230 V) and the delay. For the upper threshold, the delay is 0.2–2 s; for the lower one, 0.2–60 s.
3. Overcurrent protection. Here, too, you can set both the current limit (1–16 A in 0.1 A increments) and the delay (0.2–10 s).


1, 2. Overpower protection (100–3680 W in 10 W increments); you can choose whether it uses active or apparent power.
3. Mains frequency protection.


An observant reader will notice that the device has a temperature sensor, measures temperature, and even has settings for measurement accuracy and hysteresis. But the obvious question is: where are the options for responding to these parameters?
They're all in the “schedule” menu, which is similar to the timer menu on the EM-129 but also lets you trigger actions at temperature thresholds. In other words, it's a more advanced kind of timer.


During testing, I noticed a minor “quirk”: in portrait orientation, only the temperature is shown at the top of the screen, while in landscape orientation or on a PC, both power and temperature are displayed. It seems minor, but I don't see why all the information couldn't fit in portrait orientation too.


Let's move on to the tests. I'll start by measuring power consumption and checking the temperature. No issues here: power consumption matches the stated maximum of 2.5 W, and the case gets slightly warm. It will of course heat up a bit more under full load, but there's clearly some headroom.


Voltage measurement and device operation. Since the device has no screen of its own, I placed a smartphone next to it to display the measurements sent by the relay.

It starts up at around 40 V, connects to the network reliably, and sends data as expected, but displays the voltage as zero.
At 50 V, it starts displaying the input voltage, and the reading is accurate right away.
It continues to measure with good accuracy at higher voltages.
It switched off at 255–256 V with the threshold set to 255 V (the default), and worked without any issues.
It measures accurately up to 350 V, but above that it starts underreporting noticeably: at 360 V input, the reading is already too low.
Although the stated maximum is 400 V, it handled an increase to 450 V without any problems.


During the test, the device logged every activation. And while in normal operation a reading is recorded every five minutes, all fault events are logged.


After half an hour with an input voltage of 400 V, the case temperature was around 35–37 degrees.



I estimated the power measurement accuracy rather than actually measuring it. It looks fine to me, especially since I know my wattmeter reads about 1% high—I once compared it with a multimeter.
It’s hard to determine the minimum threshold. It displays power around 5–6 W reasonably accurately, since both the wattmeter itself and the fact that the EM-126 included the power consumption of both the wattmeter and the load introduce significant error.
At power levels around 130 and 2000 W, there are no issues either, especially given that the manufacturer specifies an accuracy of ±0.3 A.


Here are a few oscilloscope traces showing the voltage protection in action.

1, 2. Switching on after the lower threshold was triggered.
3, 4. Switching on after the upper threshold was triggered.

Here and in the other tests, switching on and off almost always happened on the same half-cycle of the sine wave.


1, 2. Voltage rising from normal to 260–270 V. The default delay is 0.5 s, and that’s about what I observed. The setting can be adjusted from 0.2 to 2 seconds.
3, 4. Voltage rising from 240–250 V to around 320 V or higher. The specified delay is 0.02 s; in practice, it seemed a little longer, though it depends on where in the sine wave the timing started. This delay cannot be adjusted.


I came across a promotional video on the Russian version of the website :)


At this point, I’d really like to praise this device, but after using it for about a week and keeping an eye on it, I kept finding myself wondering why...


Let me explain with a few examples.

First, the statistics—or, more precisely, the event log. According to the description, the device can store up to 10,000 events, after which it naturally overwrites the oldest ones in a loop. In normal operation, it logs an event every 5 minutes, which works out to 288 events a day, or a little over a month of records. I had the impression it was more, but I checked just now, and yes, the EM-129 showed 1 month and 6 days.

This is genuinely useful, but all that convenience is undermined by the fact that you can’t select a specific date. So, for example, if you want to see what happened a month ago, you have to download every record one by one! And you can’t download them all at once; you have to do it one day at a time. Given the download speed, that means clicking the button for about 5 minutes just to see what happened a month ago.
And why not reserve some memory for averaged weekly, monthly, and yearly graphs?

I’ve already asked the developers to add date selection, but so far nothing has come of it. The idea of a graph covering a longer period just occurred to me. I can’t imagine it would be difficult to implement. And of course, it would be nice to have the option to save the results.

Examples of daily graphs showing the refrigerator’s operation when connected through the EM-126


But the previous points were about improving the device; what follows are actual complaints.

As I mentioned above, I was having problems with my refrigerator, so I connected it through the EM-126 as an experiment—in other words, I was recreating exactly the scenario described in the introduction to this review.
So I’d check on my smartphone from time to time to see how it was doing, though naturally I wasn’t going anywhere in this case.

Then, one time, I tried to connect and got a message saying there was no connection. Well, that’s not entirely surprising: a refrigerator, especially an old one, is a pretty nasty load. I think anyone who has used a “Sinclair” will remember. It’s possible Wi-Fi auto-reconnect had kicked in—there’s an option for that now—but it seemed to happen rather often.


All right, I waited a bit, but it still wasn’t working. I switched the device off and on; it blinked and lit up blue, as if everything was fine. But nothing—still said there was no connection.
I checked the router and saw that the device was online. The button on the EM-126 was glowing blue, indicating it was connected, but there was still no connection... I power-cycled it, with the same result. I unplugged it, took it to another room (where the connection conditions were comparable), plugged it in, and it worked. I took it back, plugged it in, and it seemed to work again.

In the screenshot, the disconnections are where I disconnected the device from the network while trying to restart it. I don’t know what happened, but it was strange.


The previous glitch wasn’t all that serious, since the device kept working and would eventually reconnect if necessary. The next issue is much more serious.

First, just as an additional note: the logs don’t always record power consumption, even though the accumulated kWh readings are recorded correctly (highlighted in green). Sometimes, though, the device does manage to record short peaks, for example when the compressor starts. I don't know how the memory-write logic works, and I wouldn't call this a design flaw—it's just an observation.


But back to the problem. I'm working late at night, and sometimes I go to the kitchen for a smoke. That's when I notice that the refrigerator hasn't switched on for quite a while. I go over and see that it's completely off, while the EM-126 is blinking its LEDs strangely: blue, then a brief, faint red flash. It looked as if the device was trying to start up but didn't have enough power.
I unplug it and plug it back in. The relay clicks, and the refrigerator starts up.

So, apparently, during another compressor startup, the device either froze or rebooted in some strange way. Either way, something went wrong: it cut power to the output, and if I hadn't noticed, I would have found a puddle under the refrigerator in the morning.
What's more, if I'd been away from home, there would have been nothing I could do, since the device was, naturally, offline.

The screenshot below shows that the device saved its last data at 00:40, and the next set at 1:03, after I had rebooted it.


During the inspection, I also found a safety issue.
The manual states the electrical safety class and a rated insulation voltage of 450 V, and the device itself appears to meet those specifications.


But the remote sensor doesn't: its cable is clearly marked for a maximum of 300 V. I even opened the device again to make sure there really was no galvanic isolation. Surely Novatek's engineers couldn't have messed up that badly—but apparently they could.

Though I don't have to look far for an example. I still remember how their ATS burned out my computer because of a badly routed circuit board.


Of course, you'll say, come on, surely this cable can handle mains voltage? Well, probably it can, but either way, the device no longer meets its stated electrical safety class. Besides, the remote sensor uses a cable intended for low-voltage applications.

Let me explain one more complication. Besides conductor cross-section and insulation strength, there's another factor: the cable's mechanical strength. Say you have a 230 V supply and a light fixture with a 10 W bulb, and you need to figure out what cable to use to connect it to an outlet.
Your first thought might be: the current is low, so a 0.1 mm² conductor will do. No, it won't, because a cable that thin could easily be accidentally broken, so you choose one with a larger cross-section. I think you get the idea.

There's another issue with cables like these. Below is a line from the manual stating that the minimum monitored temperature is -10 degrees. The cable used is actually quite good and flexible, but only while it's warm. Try bending it in freezing weather. I tried bending an "alarm" cable at -20, and it just snapped on the first bend. Here the limit is -10 and the cable is a little different, but I doubt it'll fare much better—and it may be carrying the live wire.
There's also an issue with the upper limit: 90 degrees is specified, but the cable in the photo above is marked for a maximum of 80...



Conclusions.
As often happens, I'm of two minds about this device. On the one hand, it's put together quite well, has a neat enclosure and a decent outlet assembly that at least looks reliable.
Another advantage is that it's not just a "WiFi button": it's an overvoltage protection relay, and it can measure current, voltage, power (active and apparent), and temperature. It also has a timer and keeps logs that you can access either through the Overvis cloud or directly on your local network. By the way, the relay remembers its state before power is cut and restores it when power returns, unless an event occurs that should change that state.
It's also convenient that the device integrates with the Overvis cloud service, although in reality it just provides access to the device; there's no way for devices to interact with each other, which is a shame. I really don't understand why they couldn't add integration—for example, with an EM-129 installed at the mains input. Set a power limit, let it detect EM-126-series devices, and have it switch off the least important loads when there's an overload. Damn, I should charge for that idea—it's basically just a software feature.

Update: it turns out the integration feature is available, but only in the EPS 1+1 power-flow management system.


Now for the not-so-good parts. Some readers will of course ask: why is it so expensive when there are WiFi outlets from Chinese manufacturers that cost several times less? Yes, you're right, but only partly, since the EM-126 is a considerably more capable device.
Essentially, it's similar to the EM-129, but in a different enclosure, with lower power, a temperature sensor, and a lower price.

And here I want to criticize the developers. I would consider the price reasonable if the device were better polished—in other words, if it were in the "expensive but high-quality" category. There's definitely demand for devices like this. But software shortcomings, limited functionality, and failure to meet the stated electrical safety specifications mean I can't do that.
So what's stopping them from adding:
1. Communication between devices—in other words, creating an ecosystem
2. The ability to select a specific date without having to load all the dates in between
3. Viewing average results for a week/month/year
4. Saving logs to a PC for analysis

And, of course, there are the hardware issues:
1. The device operates in harsh conditions, so adding filters would be a good idea. A spike on the power lines sometimes knocks out the WiFi connection on both the EM-126 and EM-129, although it happens an order of magnitude less often with the EM-129.
2. Improve safety: either add galvanic isolation and use a thin cable, or skip the isolation but use a safer cable.
3. I'd also consider adding an internal WiFi antenna for a better connection, though it works fine as is.
4. And the wires inside definitely shouldn't be yellow and green.

For now, I see two key problems: the temperature sensor's safety, and the fact that the device managed to freeze and cut power to the load even though the mains were within normal parameters.

Out of curiosity, I started testing the second device. I'll report back later, but I can say that it's been running for several days now and is much more stable. The connection is more reliable, and there haven't been any freezes or dropouts.


That's all for now. As usual, I'd be very glad to hear your questions or just your comments.