36 Volts, 10 Amps, 360 Watts—or continuing to explore how power supplies work, plus a small bonus

In my previous review of a power supply, I touched on how to choose the right one. To be honest, I didn’t expect this topic to be so useful. In the comments, and even more so in private messages, people asked me about other aspects of choosing a power supply, how they work, and how to troubleshoot them.
In this review, I’ll try to answer most of those questions, and perhaps touch on some new ones too :)

I’ll start by saying that one of my upcoming projects required a 36 V, 10 A power supply. Actually, I needed two and ordered two, but since they’re identical, I’ll review just one.
I won’t say what it’s for just yet, sorry, but we’re going to take this power supply apart down to the last screw.

As always, let’s start with the packaging.
The power supplies arrived (in addition to the outer packaging) in plain white cardboard boxes. There were no identifying marks on the packaging, just two large boxes.
They look exactly alike, though I’d have been more surprised if they were different :)


The main difference between switching power supplies and those that use 50 Hz transformers is size. The second difference is price.
A 50 Hz transformer rated for this much power would be much larger. Although its design is far simpler, it would cost more because it contains more copper and iron.
Switching power supplies are also more efficient, which is why they’ve become so widespread in recent years, although “iron-core” transformers are more reliable.
Keep in mind, though, that branded power supplies are usually even more complex and expensive, since they use quality components, input filters, power factor correction, and so on. That’s why people more often go for simpler options from small Chinese manufacturers.
We’ll take a look at one such power supply in this review.
So far, we’ve looked at low-power supplies. This time I’ll cover a fairly powerful 360 W model, although next to 800–2000 W power supplies, even this one seems like a “little guy.”

As I mentioned above, switching power supplies are usually compact.
This power supply is about as tall as a matchbox—49 mm. It’s 215 mm long and 114 mm wide.


One side of the case is marked:
S-360-36
Power supply output: 360 W
Output voltage: 36 V
Maximum output current: 10 A
Input voltage: 110/220 V ±15%

The other side has an input voltage range selector. It’s unnecessary—and even dangerous—in our countries: if you switch it to 110 V and plug it into a standard 220–230 V outlet, you’ll most likely get a loud bang.
When repairing these power supplies, I usually snip this switch off right away, just to be safe.


A small fan is mounted on top of the case. At this power level, power supplies with passive cooling are extremely rare. I’ve only come across a few, and their complex design makes them very expensive, so they’re not widely available.
There’s a label nearby saying that the fan is automatically switched on and off depending on the temperature.
To get ahead of myself a little, there’s no automatic control at all. With no load, the fan spins slowly, but as soon as you put even a small load on the power supply, it immediately ramps up to full speed, regardless of the temperature.


In my previous review, I wrote that power supplies designed for high output current usually have separate terminals for connecting the load. This one is built that way: there are three terminals each for the positive and negative outputs.
The input terminals are standard: live, neutral, and ground.
There is also a power indicator LED on the left, along with a trimmer for adjusting the output voltage.


The terminal block has a protective cover that opens to 90 degrees and snaps shut when closed.


I have a habit of opening up a power supply before switching it on for the first time. I do this for safety, since all sorts of things can happen.
Everything inside this power supply looks fine, except for one small detail I noticed right away. The output inductor is quite large and almost touches the top cover, which isn't very safe. It won't give you a fatal shock, of course, but the power supply could be damaged. I'd recommend adding some insulation between the inductor and the cover. This is a common issue with many inexpensive power supplies, so it's not a flaw unique to this one.


As I mentioned above, the power supply is cooled by a small fan.
According to the markings, the fan measures 60 × 15 mm: 60 mm is its length and width, and 15 mm is its thickness.
The fan is rated for 12 V. Unfortunately, this is a cheap fan with sleeve bearings. If you plan to use a power supply like this somewhere, I'd replace it with something more suitable for reliable, trouble-free operation over the long term.
As I've mentioned in some of my other reviews, I usually use Sunon fans. In my opinion, they're quite well made and reliable.
One good thing is that the fan in this power supply is fairly quiet, which is great.


The power semiconductors are screwed to the aluminum power-supply case through small heat-spreading pads.
I'm not a big fan of this way of mounting semiconductors, but almost everyone does it. For example, Mean Well power supplies mount the transistor the same way, though they fit a rubber cap over it for safety.
Since this is a push-pull power supply, it has two high-voltage transistors rather than one.
There is one output diode, although the board has a footprint for a second one connected in parallel with the first. The second diode is fitted in models designed for lower voltage and higher current, but there's nothing stopping us from adding one here too. That would be more of a modification, though, and the measurements will show whether it's worthwhile.


Inspection complete. Let's power it up and do a quick check.
The goal of this check is to find the output voltage adjustment range and set the power supply to its rated output voltage, or whatever voltage is needed.
1. When switched on, the power supply showed 36.8 V at the output.
2. The lowest voltage you can set is 34.53 V. I expected the lower limit to be lower; I'll have to modify it for my application.
3. The upper limit was a real surprise. As I turned the adjustment, I started to feel a little uneasy. It reached 52.3 V, compared with the rated 36 V. I expected the power supply to give up while I was taking photos, but everything went fine. Still, I don't recommend setting the output voltage this high; ±10% of the rated voltage is generally considered normal.
4. Set the output to 36 V. Judging by the adjustment range, you can already tell that the control is very coarse, so I had to fiddle with it a bit to get exactly 36 V. In real life, though, there's no point in doing that; I only did it for the review :)


Let's take the power supply apart further.
The transistors were stuck pretty firmly to their plate, and I didn't want to pry them off, so I unscrewed the heat spreader plate too :)


I have no particular complaints about the circuit board. It's an ordinary, inexpensive build—not the worst I've seen, but not the best either. I'd give it 3 out of 5.
I did find one defect, though: one of the transformer pins wasn't soldered very well. The poor solder joint wouldn't have caused anything catastrophic, but it was disappointing.
The traces carrying significant current have been reinforced with an extra layer of solder.


Naturally, I drew up a schematic for this power supply. I did it just for the review, since I know this type of power supply circuitry well and usually don't need a schematic. But it might be useful to someone, as this design—with a few minor changes—is used in most power supplies of this power rating.
Even though I'm familiar with this circuitry, tracing the schematic from the circuit board wasn't very convenient and took longer than I expected.
The schematic is very similar to that of a classic computer power supply and, as experience has shown, is very repairable.
The schematic includes a shunt for current measurement. Its resistance is shown as 0.1 Ω, but when I checked it with a continuity tester, it seemed closer to a jumper.


Next, I decided to explain a little about how these power supplies work in general, especially since many of their sections are typical of almost all switching power supplies.
This block diagram shows the main sections of a switching power supply. These days, though, the oscillator and control circuitry are usually integrated into a single IC, and sometimes the IC also contains a high-voltage transistor.
Sometimes a Power Factor Correction circuit is installed at the input of a switching power supply. It's mandatory in high-power supplies that comply with European standards. I'll cover that another time, since it's rarely found in inexpensive power supplies.


I’ll use this block diagram as the basis for explaining this power supply, but first, a little theory about what happens inside a switching power supply.
The key to how a switching power supply works is PWM regulation. It’s worth noting, though, that switching power supplies without it do exist, but they aren’t regulated, meaning the output voltage depends on the load power and input voltage.
PWM regulation means changing the ratio of the switching element’s on-time to its off-time.
On a graph, it looks like this:


To put it in simple terms, I recently explained this regulation principle in a private message, so I’ll try to explain it again here.
Many of you probably remember problems like this: water flows into a pool through one pipe at a rate of X liters per minute, and drains through another at a rate of Y liters per minute.
I’ll use that principle to explain how this works.

First, imagine a very large reservoir (the electrical grid), a small reservoir (the power supply’s output filter capacitor), and various odds and ends for transferring water from one place to another.
There’s a tap on the barrel, and water flows through it to the consumer—or, in our case, energy flows to the load.
We can refill the barrel only a certain number of times per minute (there are other options, but we won’t get into those for now)—say, 100 times.
Our task is to keep the water level in the barrel constant at all times.
Since we can refill it only a certain number of times per minute, we’ll have to refill it with different amounts of water each time.
For example, if consumption is low, ordinary cups will do. But if the tap is wide open, we’ll have to use buckets.
In PWM regulation, this means a narrower or wider on-time for the power switching element.
If the tap is closed, we refill the barrel with thimbles—there’s still evaporation (leakage, current drawn by the feedback circuit, etc.) to compensate for :)

Using the feedback circuit, the controller monitors the power supply’s output voltage and adjusts the power delivered to the load so that the output voltage remains constant.
By the way, you can use this method to create feedback based on just about anything.
For example, in LED drivers, the controller monitors the current.
You can monitor temperature and adjust fan speed, or monitor light levels and adjust a lamp’s brightness, and so on.

This diagram shows:
1. Current in the transformer circuit (approximate)
2. The control signal for the switching transistor
3. Voltage across the output capacitor.


There are quite a few topologies used in switching power supplies. I’ve drawn some of the most common ones.
Here’s a brief overview.
1. Flyback converter. Used where a wide input voltage range is desirable and power requirements are modest (up to 100–150 W). Your tablet or monitor power supply most likely uses this topology.
2. Half-bridge converter. Another very common topology. I don’t think I’d be far off the mark if I said that 95% of computer power supplies use this topology. Its advantages are high power with a relatively simple circuit, and a smaller transformer, since the transformer has no air gap, unlike in the first topology.
3. Push-pull converter. This topology is rarely used in mains-powered supplies, but it’s widely used in the inverters of inexpensive uninterruptible power supplies.
4. Full-bridge converter. You could call it an “expanded” version of the half-bridge. Its advantages are higher power and half the current through the power switches compared with a half-bridge.
This topology is also used in more sophisticated uninterruptible power supplies.

There are several other topologies, but they’re variations on the ones above and are less common, so I haven’t included them in this article.


This time I also drew a color version of the schematic for the power supply under review, using colors to mark the main sections I mentioned above.
As I wrote, it's hard for me to name some of the colors accurately, so I'll clarify.
Red — Input power filter, diode bridge, power stage.
Red-violet (bottom left) — Control circuit for the inverter's power transistors.
Green — PWM controller IC and its supporting components.
Blue — Output rectifier, inductor, and filter capacitor
Light blue — Output current monitoring circuit
Violet — Output voltage monitoring circuit
Yellow-orange — Converter shutdown circuit that activates when the output voltage drops.

This schematic doesn't have the familiar component found in all the previous schematics: an optocoupler. That's because the PWM controller here is powered from the output voltage. The power supply starts up initially thanks to resistors R8 and R14. This approach was used in AT-standard computer power supplies. With the arrival of the ATX standard, the controller began to be powered from the standby supply, and these resistors were removed from the circuit.


Next, I'll show most of the sections and components using a specific power supply as an example.
Let's start with the mains filter.
This power supply has one, which is already good, since cheap computer power supplies use simple jumpers instead, while more expensive ones may have a multi-stage filter. This one is somewhere in between.


The power supply input has a fuse and an inrush current limiter—a NTC thermistor.
There's also an X2 capacitor to reduce interference emitted by the power supply back into the mains.


The two-winding common-mode choke is wound with fairly thick wire, although it could have been larger for this power rating.
The input diode bridge KBU808 is rated for 8 A and 800 V.


The input filter includes both Y capacitors and one regular high-voltage capacitor.
But in this case, using a regular high-voltage capacitor instead of a Y capacitor is safe: if the power supply isn't grounded, then even if the capacitor fails, the output will still be connected through the Y capacitor; and if the power supply is grounded, then there's nothing to worry about anyway :)


The input filter capacitors are marked 680 µF × 250 V.
If the markings are accurate, their capacitance is adequate, and the voltage rating even has some headroom.


But reality turned out to be a little different: the capacitors have a total capacitance of just 437 µF, which gives only about 220 µF when connected in series. Not much, though tolerable in principle.
Higher capacitance means longer capacitor life, less ripple, and more headroom for lower input voltage.
I think I'll replace them with something better later, but I haven't found suitable ones yet. These capacitors are 35 mm tall, and 40 mm is probably the maximum that could fit, while most of the capacitors I've found are 45 mm tall.
There's room on the board for a capacitor with a larger diameter, so I'll keep looking :)


The PWM controller and inverter section.
The PWM controller is a classic: the KA7500, which is almost identical to the TL494, probably the most widely used PWM controller. The only real rival in popularity is the UC384x.
The inverter's power switches are MJE13009


Unfortunately, the heat spreader is pressed against the case without any thermal paste. Testing showed that this doesn't cause any problems, but I'd still apply some for peace of mind.


The output transformer, rectifier, and filter capacitors.


The output diode is an SF3006PT, rated for 30 A and 400 V, which is more than enough for a 10 A power supply.
As I mentioned above, there's room for a second diode next to it, so you could improve the specs a little in principle, but in practice the efficiency gain would be tiny.


The output inductor.
Here it serves a somewhat different purpose than in flyback power supplies, which is why it's so large. I'll just say that its size is appropriate for the power supply's rated output. No complaints apart from its height.
The output filter capacitors.
The manufacturer installed three 1000 µF, 63 V capacitors.
I usually say that output capacitance should be 1000 µF per amp of output current. The requirements are less stringent in push-pull power supplies, and even name-brand units use the same (and sometimes lower) capacitance at this current. To be fair, though, branded power supplies use higher-quality capacitors.
The current shunt is also visible in the photo, along with space for additional shunts in higher-current versions.


The capacitance is fine here—almost exactly as specified.


After inspecting it, I put everything back together, leaving the top cover off, and moved on to load testing.
I used the same test setup as before, consisting of:
An electronic load
A multimeter
A non-contact thermometer
An oscilloscope
A pen and some paper :)

This time, though, I had to remove the electronic load's top cover because I was worried it would overheat at this power level.
The testing was much the same as before, except that I had to remove the top cover on the fly to measure the temperature. As a result, some of the temperature readings will be slightly high, since the power supply had time to warm up a little without forced cooling.


1. No-load mode, voltage set to 36.03 V; ripple is practically nonexistent.
2. Load current: 2 A. The voltage rose slightly to 36.06 V; ripple is within spec.


1. Load current: 4 A. The output voltage rose a little more; ripple is within spec.
2. Load current: 6 A. Output voltage is 36.09 V, which is an excellent result. Ripple is just 50 mV.


1. Load current: 8 A. The output voltage is almost unchanged. Ripple rose to 75 mV, but is still low for this current.
2. Load current: 10 A. The output voltage rose to 36.12 V—an excellent result, just 0.3% above the initial value. Ripple rose to 100 mV, which doesn't seem like a problem to me, especially considering that the power supply delivers 360 W and 100 mV is only 0.25–0.3%.
For comparison, on a 12 V power supply, that would be equivalent to 30 mV of ripple.
Unfortunately, the last test lasted only 15–16 minutes instead of my usual 20. The electronic load's overtemperature protection kicked in and shut off the load :(


After letting the load cool down a little, I decided to continue the test as an experiment, this time at 12 A. If we're testing, let's really test it :)
I decided to try this because the power supply's components were nowhere near their maximum temperature.
Unfortunately, the power supply lasted no more than a minute. I took a photo and captured the waveform, then there was a very quiet click (though with the load's fans roaring, maybe it wasn't so quiet), a tiny flash near the power switches, and the power supply went silent :(
I did have a suspicion that the electronic load was to blame. In a certain situation, when it overheated, it could have shorted the power supply's output (if the protection on the heatsink with the current sensor tripped first). It couldn't have heated up to that temperature in a minute, though. In any case, the power supply didn't hold up :(


The waveform just before it failed.
The ripple voltage is clearly within spec. What bothers me is the higher-frequency ripple, probably caused by ringing in the power circuits. I think this could be one possible cause of the failure, but I can't say for sure.


The thermal test went as usual: 20 minutes to warm up, measure the temperatures, increase the current by one step, and so on.
The results are easy to read from the table. The top row shows the temperatures measured with no load; I also checked that the thermometer gave the same readings on different components.


As a small bonus, I’ll briefly describe how I tracked down and repaired the fault in this particular power supply, along with the troubleshooting principles that apply to most faults in other units.

Summary.
Pros
The power supply delivered its rated power
Operating temperatures are within the normal range
Low ripple
A proper 220 V input filter is present
Excellent output voltage stability
Good repairability

Cons
Reliability issues under overload or short-circuit conditions
The input filter capacitors have lower-than-specified capacitance
The fan does not have the advertised automatic control.
Poor-quality output capacitors

My take. I was very disappointed that the power supply failed. It happened while it was delivering more than its rated power, but that points to a missing or improperly functioning overload protection circuit. At the same time, I was pleased with its thermal performance: even at maximum power, none of the components overheated. The exhaust air did have a slight smell of warm components, but that's common with new power supplies.
Even though I burned this power supply out, I can say it has decent potential, and as long as you don't overload it, it should work. This is mainly thanks to the maturity of this circuit design; it's hard to mess things up here, although reliability issues did crop up :(
I think I'll make a few modifications to it in the future, and I hope you'll see it (or, more likely, them) in one of my upcoming devices. I'm slowly preparing a review of it, which will also include a description of the modifications.

There may be a few errors in this review. If you spot any, let me know and I'll correct or add to it as needed.
All the repair information is based on my personal experience. There are many more possible causes of failure and ways to diagnose faults than I've covered, but describing them all would be very difficult and would make this article far too long.
I hope at least some readers will find answers to the questions they've asked me.

The store gave me a discount on the power supply. The original price was $30.2, and the discounted price of $26.99 will be available for one week.

The product was provided by the store for review.