Methods for protecting electronic devices and keeping them running when the mains voltage is too high

I don't think I'm far off if I say that quite a lot of people have had electronic devices fail because of overvoltage. It happens for various reasons, but the result is almost always the same: the power supply burns out. Just as there are different causes, there are also different ways to deal with the problem, and that's what we'll talk about today.

A reminder: to keep up with new topics on the blog, I recommend following me on Instagram, where I'll post notifications about all new topics and maybe write about something interesting too - link to my account

You can also subscribe to updates and new articles on the Telegram channel - https://t.me/KirichBlog

By overvoltage, in this case, I mean the device input being exposed to a voltage above 242-252 V for an extended period.

There are many causes of overvoltage in the electrical grid. The most common include wires crossing in the wind or snow, broken wires, a lost neutral, and, more recently, at least one more: the destruction of substations. There are also transient overvoltages, for example from lightning strikes, but that's another topic and calls for different protection methods.

What do people usually use for protection? At the very least, you can install an overvoltage protection relay. It's a simple, effective, relatively reliable solution, although it does have certain drawbacks. For example, if the relay fails, all the devices connected to it are left unprotected.
In household use, though, it's the simplest and most effective option, especially since homes have many different appliances that are difficult to protect individually.


Another option is to protect devices locally, and some appliance manufacturers offer this. I won't claim whether it works well or poorly; the point is that it exists.


And today I'd like to talk about this kind of protection and show you a real-world example.

There's nothing innovative about this kind of protection. So why isn't it built into everything? Because, in the short term, it would at least make things more expensive, and in the long term it wouldn't pay off. The first constraint comes from buyers themselves: these devices are simply more expensive. The second comes from manufacturers, since it's easier to deny a warranty claim than to make a reliable device.

What do manufacturers usually do for protection? At most, they install a varistor, which in some situations can indeed protect the device by breaking down and, as a result, blowing the fuse. Unfortunately, this kind of protection doesn't always work, it's not present in every device, and sometimes the varistor's rating is so high that it can only help with high-voltage transients.



Most often, the varistor is rated at 470 V, and in some situations it really does help. But this is one-shot protection; here's an example of it working.


Another option is the OVP function in power supplies. Again, it's far from always implemented; in fact, it's more common for it to be absent than present. As an example, here's a power supply circuit based on a chip from Power Integrations, with a UVP/OVP protection circuit. The manufacturer states that

In plain English: when the input voltage exceeds a certain level, we disable the high-voltage transistor. Since it is rated for 700 V and there are no voltage spikes caused by its operation, the power supply can withstand high voltage.



The third solution is to design the power supply from the outset to operate at this voltage. Essentially, it’s simple: the maximum voltage we can get from the mains is 380/400 V (except in unusual circumstances), so if we build a power supply for an input voltage of 400 V, it no longer matters what the input voltage is.
This approach is convenient in its own way, but it costs more. Also, at 220/230 V, and especially at 180–190 V, it operates less efficiently, reducing its efficiency; efficiency is also reduced by the use of higher-voltage power transistors.
Below are examples of power supplies rated for operating ranges of 100–520 and 80–580 VAC.

There is also a separate category with automatic 115/230 V switching (there were even dedicated switching control chips), but these power supplies never became widely used and were limited to models with a mechanical switch.


There are quite a few reviews of voltage relays, and I have about 5–6 reviews of them myself, with tests and sometimes even circuit diagrams. High-voltage power supplies are rarely used, though, so I’d like to focus on protection at the power supply level..

As an example of overvoltage protection at the load, we can look at the LRC-60 LED light drivers, especially since lighting is exactly the kind of application where a failed power supply is more noticeable. And since similar drivers are also used for street lighting, group protection may be inconvenient: if it fails, we lose the entire lighting branch, not just one light.


I ended up with two drivers, although essentially they’re the same driver: one is an IP20 version, the other IP66.


This is an updated version of the drivers. Three years ago, I reviewed both the drivers and the power supplies from this manufacturer.


One of those power supplies has been powering the kitchen lights ever since. The other had a more unusual fate: its transformer went into a soldering station, and its PWM controller went into another power supply for repair :)



The operating voltage range refers to the range in which the driver powers the load, but it also states that the driver can withstand an input voltage of up to 380 V for a relatively long time.


Since the protection ratings differ, one driver is potted, while the other is “bare.”


I’ll use the IP20 driver as an example when looking at the internals.


Well, I’m a simple guy, and I could have taken apart the potted driver too, as I’ve already done in one of my reviews, but I think that when you have an unpotted driver in your hands, that would be overkill.
By the way, the photo shows the previous model. It could also withstand 380 V at the input, but the key difference in the new one is the active power factor correction.


You can already see the differences between the drivers here: the new one is on the left, the old one on the right. The new one has no high-voltage capacitors.


I got sidetracked there. Let’s get back to the article.
Which components determine whether a power supply can withstand high input voltage? Let’s also look at this driver as an example.
1. A varistor on the mains input, after the fuse, of course (though I have come across one before the fuse...). It’s rated at 680 V, which corresponds to about 475 V AC, so here it’s there to protect against transient interference. By the way, a reminder: the voltage marked on this varistor is its DC rating. For AC, that corresponds to the peak voltage, whereas Epcos varistors, for example, are rated by RMS voltage.
2. An X capacitor. I had to desolder it to check the rating—or, more precisely, the voltage. This one is rated for 440 V, which is quite important, since power supplies usually use 250–300 V capacitors. There are two of these capacitors here, one before and one after the common-mode choke.
3. Another varistor, rated at 470 V. Judging by its unusual circuit connection, it’s there to suppress isolated pulses, but it has no effect on continuous voltage.
4. The interwinding Y capacitor, like X capacitors, comes in different varieties—or, more precisely, two types: Y1 and Y2. The former is more resistant to high-voltage surges, but in this case the capacitor is not only Y1-rated, it is also rated for up to 400 V.


1, 2. In fact, in some situations safety is further improved by connecting capacitors in series. This is especially true of Y capacitors, since a person's life may depend on them. The photo shows pairs of 250 V Y capacitors connected in series.
3. The same is done with X capacitors, although much less often.
4. In some situations, two input fuses are also fitted to improve the safety of the power supply—one on the neutral and one on the live wire. Sometimes two fuses are connected in series, for example, one before and one after the varistor, with different ratings, but that's more of an exception.

In any case, you can never have too much safety. People sometimes don't realize that the live and neutral input terminals on a power supply are marked for a reason: it's a safety measure, since the fuse must be installed in the live wire.


The control board is a submodule to save space, but the main point here is the high-voltage transistor, since it also determines how well the power supply withstands high input voltage. WML08N80M3, 800 V, 7 A—not bad at all. Someone might ask why all power supplies don't use transistors rated for such high voltages. The answer is simple: the higher the transistor's voltage rating, the higher its on-resistance usually is, which means a greater voltage drop, more heat, and lower efficiency. With this type of circuit, 500–650 V transistors are more commonly used.


And of course, a few tests and a practical demonstration. Besides the driver, I also needed a load, which in this case will be two LED panels.


Each panel has 54 LEDs arranged in a 6P9S configuration, giving a total voltage of about 26–27 V. The driver is rated for up to 60 V, so I'm using two panels. Great little panels—they're bright, and I even ran them without extra cooling. They do get hot, though :)


With both panels connected, the driver outputs 52.8 V. That's determined by the panels themselves. The current is about 900 mA, which is determined by the driver. The current and voltage are indicated in the model name: 60-900.
I chose these panels for a reason: I wanted a load close to the maximum 54 W, and got about 47 W. Of course, I could have used an electronic load in CV mode, but that would be much less illustrative.


Since I had the driver and load on the bench, I quickly estimated the efficiency and checked the claim that it has power factor correction.
It's hard to say much about the efficiency: I got around 90%, while 92% is claimed. As for the power factor, no question there—it was somewhere in the 0.97–1 range.


And of course, I tested how well it withstands high voltage. The manufacturer specifies 380–400 V, but I didn't want to stop there and raised the voltage by almost another 10% (though I raised it even higher while filming). I also tested its resistance to an output short circuit.
1. 00:00-00:45 - Gradual change in input voltage.
2. 00:45-1:00 - Sudden change in input voltage.
3. 1:00-1:35 - Output short-circuit test: first brief shorts, then a sustained one.

I also checked the output ripple amplitude. It's specified as up to 2%, and in reality it was a little lower. One division on the oscilloscope screen is about 5%; the main ripple (excluding high-frequency noise) is well within that figure.
During testing, I found that after a brief short circuit the driver restores its output immediately. After a sustained short, it takes about 10 seconds to recover once the short is removed.


The manufacturer also has a similar demonstration video, though it's a bit simplified. The waveform looks nicer, of course. I suspect the oscilloscope has some additional filters enabled to limit the bandwidth. I used the standard 20 MHz bandwidth limit, and the signal was taken from the shunt through a typical C+C network.


If you don't feel like watching the video, the main point can be summed up in one photo: the driver shuts down at around 303–306 V and turns back on at 290–295 V. Last time, I complained that there was no hysteresis after the protection kicked in. Here it's clearly present, which is a plus.


We've essentially looked at three main ways to deal with overvoltage on the mains:
1. Cutting power at the circuit level using a voltage monitoring relay
2. Protection built into the device: it doesn't fail, but shuts down while the voltage is too high.
3. The device tolerates voltages up to 380/400 V and continues to operate


Personally, I prefer the third option. I use the first one at home, but the second is just as viable; it all depends on the use case. When it comes specifically to protecting lighting drivers, I would also prefer the second option. That way, the lights will stay on in an emergency, which can be very important. I’d like to hear what you think about this.

That’s all for now. I hope you found it useful.