What a Switching Power Supply Is Made Of, Part 2

I've already posted videos covering individual parts of a power supply, but after thinking it over, I decided I hadn't done it the right way—or, more precisely, not in a very logical order—so I decided to put that right.
This article kicks off a short series called “How It Works,” where I'll go through all the sections of a typical switching power supply one by one, explain what they do, and point out where components can fail.

As I've mentioned before, a typical power supply consists of the following sections:
1. Input filter and rectifier with filter capacitors.
2. PWM controller and inverter transistors.
3. Power transformer and snubber circuits.
4. Output rectifier, output filter capacitors, and feedback circuit.

A simplified block diagram would look like this. There are, of course, some exceptions, but the overall picture is much the same.
One exception is power supplies with switched capacitors, but those are pretty unusual.


Most sections can themselves be broken down into smaller parts, so I may cover some of them separately. But today I'll explain where a switching power supply begins. For example, I plan to make a separate video about power factor correction circuits.
A power supply starts with the input EMI filter, rectifier, and filter capacitors.


First comes the protection circuitry, which includes a fuse, varistor, thermistor, and a resistor to discharge the input EMI filter capacitor.


Next comes the filter that keeps interference from the power supply out of the mains.
It includes Class X and Class Y capacitors, as well as a common-mode choke.


And finally, the rectifier and filter capacitors.


I've already talked about the input filter and protection components, but I'll digress and cover them briefly here too.
A decent input filter looks something like this.


Or like this. You can also see the choke, capacitors, fuse, and varistor here.


Or here's the filter from a Mean Well power supply.


As I said, the filter in a switching power supply isn't just important; it's often quite complex too. Sometimes the filter has so many components that it's almost as complex as a basic power supply. Here's an example of a more complex filter circuit.


By the way, filters like these are sold as standalone units, for example by Mean Well.


What's more, the manufacturer even printed the circuit on top, which is quite unusual.
I've seen filters like these in Soviet-era computer equipment; I still have one lying around at home.


In any case, the key component in the filter is the two-winding choke, and it's easy enough to spot.


But some manufacturers are extremely stingy (they could use a good slap on the wrist) and put jumpers in its place. Needless to say, they don't filter anything.
You usually find them in the cheapest power supplies. Though I've seen exceptions: a more expensive unit had jumpers, while a cheaper one had a choke.


The fuse is just as important. To start with, there are different kinds, and sometimes it's replaced with a resistor altogether.
Of course, there are special fusible resistors, but ordinary ones are often used instead.
Fuses come in different types and sizes. I wouldn't call one as small as the one in the photo a good choice.


Here's the right kind: it's not only larger, but also protected with heat-shrink tubing. It's not a case of bigger being better—we're not sizing fuses according to Freud. A larger fuse simply has more distance between its leads, making the circuit break more reliable.


It's generally assumed that a fuse protects the equipment. That's true, but only half the story. If the circuit has a varistor, it will start clamping the voltage when it gets too high, eventually blowing the fuse and protecting the equipment. I've come across branded equipment that had more than 300 V applied to it after the neutral wire burned out. After replacing the varistor and fuse, everything worked as before.
If there is no varistor, the fuse only protects your electrical wiring.


Varistor markings are very simple: three digits, the first two give the value and the third is the multiplier. For example, power supplies use 470 V varistors marked 471.


And, of course, capacitors. I talked about them in a separate video, so I'll keep this brief.
First, Y capacitors are easy to mistake for varistors because they have similar markings, colors, and dimensions. But varistors usually have simpler markings, while Y capacitors are thicker and smaller. So just read the markings carefully.
With X capacitors, the one on the right in the photo, it's much simpler: look for an X1 or X2 marking and the rated voltage.


Safety capacitors usually have more markings, so you can even identify them by their appearance.


And here are X-type capacitors. They also come in blue cases; you can see them at the beginning of the video.


After all the filters comes the rectifier. Its job is simple: convert AC to DC. But even here, there can be surprises.
A diode bridge is usually used for rectification in switching power supplies, which is pretty obvious and straightforward.


But some manufacturers manage to cut costs even here. I have some dirt-cheap power supplies lying around that use a half-wave rectifier—in essence, just one diode.
With this design, ripple at the rectifier output will be significantly higher for the same capacitance. You could, of course, modify it by adding the three missing diodes, but if they cut costs that much, it's cheaper to throw it away.


A diode bridge can be made from individual diodes or built around a bridge rectifier, which is, of course, much more convenient.


By the way, someone once asked me whether a bridge rectifier should be mounted on a heatsink. It depends on a lot of factors, but if the power supply has passive cooling, it's better to screw a small heatsink onto it—for example, as in MEAN WELL power supplies.
The power supply in the photo is rated at just 150 W.


Low-power supplies usually have just one capacitor, although I've come across exceptions.
Most often, these power supplies are designed for a wide input voltage range.



In higher-power supplies, you'll most likely see a switch like this. It lets you select an input voltage range of 110 or 220 V.


There will be two capacitors next to it. This isn't a hard-and-fast rule, of course: some high-power supplies have one capacitor, and I'll definitely talk about that. You can also find low-power supplies with two capacitors; it's just that the arrangement shown in the photo is the most common.


I've come across some misconceptions and confusion online about what happens when you switch the voltage range, so I'll try to explain.
In the usual mode, the switch is open, and two capacitors connected in series are connected to the output of the diode bridge.
The resistors discharge the capacitors and help balance the voltage across them.


Since not everyone has 220 V at the outlet—and sometimes it's half that—they came up with a simple way to switch between the ranges.
When the switch is closed, the midpoint between the capacitors is connected to one of the input terminals, and the diode bridge starts working as two diodes.


If you rearrange the diodes a little, the circuit becomes easier to understand.


It becomes two half-wave rectifiers, connected so that one charges the first capacitor from the positive half-cycle, while the other does the same with its capacitor from the negative half-cycle. The two lower voltages add up to a full 300–310 V. This circuit is called a voltage doubler rectifier. This trick is only possible with AC, which, fortunately, won out over DC many years ago.


But this solution has one small drawback. Since the circuit operates in voltage-doubler mode, closing the switch with our 220 V mains voltage can end badly. The rectifier will try to charge the capacitors to 310 V each, while they are usually rated for only 200 V.
At best, their tops will bulge, and the whole room will smell like a can of milk.


I've had worse cases, too, when a capacitor simply burst and all that was left on the board was its rubber base.


The main thing in this situation is to make sure the flying top doesn't hit anything important, like an eye.


Another important question I get asked very often is how to determine the required capacitance of the input capacitor.
The usual recommendation is a capacitance in microfarads equal to the power supply's output power in watts, but there are some nuances here too. I'll try to explain and show them on the graphs.
In the first example, the mains voltage is normal and the capacitance has some headroom, so you can see small ripples.


Now the input voltage has sagged a little. Everything is fine, except that the ripple is approaching the yellow zone, though it's not critical yet.


Let's bring the voltage back to normal, but increase the load. You can immediately see the ripple amplitude increasing. This operating mode can already be harmful to the input capacitor, shortening its service life.


Keep the same power, but lower the input voltage. The ripple amplitude increases slightly because the inverter tries to compensate for the lower voltage by drawing energy from the capacitor for longer. It's harmful, but everything still works.


Let's lower the voltage even further. Situations like this do happen, and not necessarily for long. For example, starting a refrigerator or air-conditioner compressor on a weak mains supply can cause a noticeable voltage sag.
The voltage on the capacitor drops below the red zone, meaning we'll see 100 Hz ripple at the power supply output. That's bad.

One more experiment: raise the voltage a little, but reduce the capacitor's capacitance. The result is the same as before, except that the ripple amplitude is greater, which is now even more harmful to the capacitor.


Good-quality power supplies usually have capacitors with plenty of headroom. This is necessary for stable operation across a wide input-voltage range and to extend the capacitors' service life.
For example, a monitor power supply rated at around 40–50 W has a 120 µF capacitor, even though 47–56 µF would be enough if you calculated for 220 V alone. We don't think the manufacturer did this out of the goodness of their heart.


You can improve a power supply's performance by increasing the capacitance—for example, by adding a couple more capacitors in parallel. But keep in mind that the capacitors must have the same capacitance, and ideally should be identical in every other respect too.


Some power supply manufacturers do this: here, four capacitors are connected in two parallel-series pairs.


You can also use a single capacitor across the 310 V bus, but in that case it must be rated for at least 400 V.


I've modified power supplies this way for a high-power adjustable supply. Below you can see the desoldered input-voltage switch. I recommend doing this to other supplies too, because even a switched-off power supply can burn out once a year. :)


Another popular question: which capacitors are best to use—the brand, model, and so on.
Chinese power supplies often contain either counterfeit branded capacitors or simply cheap, unbranded ones. They are of course worse than branded ones, but experience shows that this is not critical in this circuit.


The main thing is to make sure the capacitor isn't a "matryoshka"—so it's best to measure its capacitance and then decide whether to replace it or add other capacitors in parallel.



Branded power supplies, of course, use decent capacitors; I haven't come across any counterfeits or no-name ones.


Now, about manufacturers. In fact, the requirements for the input capacitor aren't as strict as those for the output capacitors. But if you want to do things right, I'd recommend avoiding no-name parts and looking at branded capacitors.
There's a pretty wide selection, for example, Nippon.


Or Samwha, formerly Samsung—reasonably priced and good quality.


Nichicon, though they're more expensive and harder to find.


Rubycon capacitors are good too, but unfortunately they're often counterfeited. For example, the ones shown above are labeled RubyconG; as you can imagine, that's something entirely different.


I also recommend the very good CapXon KF series capacitors.


Or Jamicon.



I've left until the end a question I'm asked a little less often, but which is still important when choosing filter capacitors.
People have asked what maximum operating temperature rating to choose when buying capacitors to replace the originals in a power supply.
Generally speaking, both 85 °C and 105 °C capacitors will work fine. But if your power supply has passive cooling, I'd recommend using 105 °C capacitors; they'll last longer in that kind of power supply. If it has active cooling, I don't think you'll notice a significant difference.


To finish, here are a few photos of an almost vintage power supply. It was installed in some old computer, if I remember correctly, made in Bulgaria. It also had a keyboard with Hall-effect sensors, housed in a metal case—practically indestructible. But all I have left of it are the keys and sensors; I regret taking it apart now.
Anyway, this is a passively cooled power supply with active power factor correction—in other words, essentially the same feature that's being promoted as important today. And it was already around 30 years ago, and used quite widely.
The unit is rated at 270 W, although the board itself says 260 W. The only output voltages are 12 and 5 V.
It was made by Boschert. I was surprised to find that they still sell them, though refurbished.


And here's what my power supply looks like. I might give it a photo shoot of its own; I think it's earned one :)
Sorry about the dust—it's been on the balcony for many years, plus we moved and renovated the apartment.


That's all for today. As always, I'm looking forward to your questions and suggestions for topics for new videos and reviews.

I almost forgot again—the video itself. This was my first time filming in this format. If you think it's worth continuing this way, let me know.