Another power supply, this time 24 V, 20 A, and 480 W

By and large, power supplies aren’t all that different from one another, but this time things went differently. A lot of things were different, and I’ll go over them, highlighting the key points. I think it’ll be useful.
I’ll try to keep this review short—or at least not too long :)

The differences started with the packaging. First, there was a special “window” in the box, so you could see the sticker with the power supply’s model name. Handy.
Second, the power supply was sealed in plastic film, which I hadn’t seen before either.


On the outside, this power supply is practically identical to the previous 360 W model: same dimensions, same fan grille.


In my reviews, I almost always show a photo of the terminal block. I started doing that after someone commented that some power supplies have covers that don’t open all the way. And now I’ve come across one like that myself. It turned out later that this can be fixed, but straight out of the box the cover wouldn’t open fully. Not very convenient.


The terminal labels aren’t on a sticker; they’re stamped into the cover. There’s also a warning label next to the fan.
The cover is quite thin, and it’s even dented in one spot.


As usual, there’s a trimmer for adjusting the output voltage, as well as an indicator LED.
The power supply is marked S-480-24. The output current is 20 A. I’ll probably never understand why power supplies are labeled “LED Power supply.” What do LEDs have to do with it if the power supply is universal? I guess they sell better that way.
There’s a warning sticker, as well as a 110/220 V switch.
The power supply was made in late 2016, so you could say it’s fairly recent.


When I took the cover off, I actually froze for a moment :) Finally, something different from the classic TL494-based power supplies I’ve seen so many times. There’s practically nothing inside. As they say, that’s no coincidence.
The case is also a little different. Usually the cover is held on by six screws; here there are two screws and a pair of tabs at the top.


To make the difference between a “classic” power supply and this one clearer, I took a couple of comparison photos with the previous 12 V, 360 W power supply.



First, I checked how the power components are mounted. Even though there’s a 99% chance of no problems when transistors or diodes are installed in pairs, I still inspect the mounting.
The transistors and diodes are clamped to the aluminum case with bars. There are no heat-spreading plates, so the power components are pressed directly against the case.
No complaints: everything is straight and neat. They even applied thermal paste. At first, it might look like there’s far too much, but in fact only a very thin layer remains under the components.

If you look closely at the second photo, you can see the markings on the circuit board, which suggest it was designed for a 360 W power supply.


A 60 mm fan cools the internals. It seems fairly powerful, which is also suggested by its size relative to the power supply’s rating. It’s not very loud, but you can hear it.


The first test is measuring the output voltage adjustment range.
1. The power supply was initially set to a little over 24 V.
2. The lowest setting is about 14 V, but the power supply is unstable at this level, so I had to switch the tester to display minimum and maximum values. Apparently, the power supply is underloaded in this mode, and the PWM controller isn’t getting enough power, so it keeps restarting.
3. The power supply starts working stably at around 20 V.
4. The highest setting I could get was about 27 V.
5. I set it to the standard 24 V and noticed two things. The adjustment is quite coarse. I don’t see why they made the range start as low as 14 V; they could easily have limited it to 20–27 V, which would have made the adjustment smoother.
6. But there’s another problem: as it warms up, the output voltage drifts upward a little. You can see this from the MAX reading and the time next to it.


Since I was measuring the voltage, I also measured the capacitance of the input and output capacitors.
The input capacitors have a total capacitance of 313 µF, which is a bit low for 480 W. The output capacitors aren’t much better, at around 7000 µF; I’d like to see more there too. But as I’ve pointed out many times, branded power supplies with similar specifications have about the same output capacitance.


Now we can take it apart properly and see what differences the Chinese engineers have in store for us.


The first “surprise” was waiting for me almost immediately. While taking it apart, I also noticed that there are five mounting holes for the board, but only four screws. It wasn't the middle one that was missing, as is usually the case, but a corner screw.
To get ahead of myself a little, I found the screw when I accidentally knocked the board near the end of the inspection. It was presumably under the transformer. Not good.


The power supply input has a filter to suppress interference from the power supply getting back into the mains. The filter uses a typical configuration for power supplies like this.


1. Before the filter, there is a fuse and a pair of thermistors to limit inrush current. I'm sometimes asked why the live and neutral wires are marked on power supplies like this. The reason is that there's only one fuse, and it's usually on the live wire. So if the power supply fails, the electronics are not only disconnected from the mains, but also won't remain at live potential.
2. Next comes an interference-suppression capacitor and a two-winding choke wound with fairly thick wire.
3. All the interference-suppression capacitors that affect safety are the correct Y2 type. The filter uses just one ordinary high-voltage capacitor, but it doesn't compromise safety.
4. The bridge rectifier is made up of four 1N5408 diodes, which I don't think is a great choice at this power level. Only the active cooling saves the day. On the plus side, you can see a spot for an additional capacitor nearby. You could fit a 400–450 V capacitor there to help the ones already installed.


The four filter capacitors look unusual; two is what you'd normally expect. They have the logo of a well-known company on them, but don't be fooled: these aren't genuine capacitors. You can tell by the uneven heat-shrink at the top of the case.
The stated filter capacitance is 470 µF, with a 2S2P configuration. The actual capacitance is 313 µF. I doubt genuine capacitors would have that much variation, and their size says it all.


Interestingly, the transformer is about the same size as the one in the previous 360 W power supply. But the power supply under review operates at twice the frequency of the previous one.


1. This time, MOSFETs are used instead of the bipolar transistors seen in previous reviews. The transistors are IRFP460, but they look different from each other, which may mean they're used parts. In a proper production run, you'd normally expect the transistors to come from the same batch, not to mention look alike.
2. The output diode assemblies are much the same story. Both are marked 43CTQ100, but they look different.
3. The output inductor is wound with four strands of wire and is relatively small, especially compared with the previous power supply models I've reviewed.
4. The output capacitors are from an unknown manufacturer, rated at 35 V with a capacitance of 2200 µF.

As usual, there's no output interference-suppression choke. In fact, you rarely see one in high-power power supplies, at least Chinese ones.
There's a high-power resistor next to the capacitors, which is why the output voltage "drifts" as the unit warms up.


In my reviews, I usually inspect the circuit board and most often write something like, "The board is clean and the soldering is neat." Not this time: here, it's the opposite.


But what surprised me most was the circuit board layout. It's usually recommended to place power-stage components as close together as possible. More precisely, the components in each power stage should be grouped together.
Here we see a bunch of long traces running from the power transistors to the transformer, with a power-supply trace and ground running alongside them. In my opinion, this isn't a great design and is likely to cause a lot of radio-frequency interference. The only thing that helps is the power supply's all-metal case, which should be grounded.


Most of the output section consists of fully tinned copper pours, which is the right approach at these currents.
But look a little lower and you'll see a handful of components: these are part of the feedback circuit. On the other side of the board, directly above them, is a load resistor (I've marked it in the photo) that gets noticeably hot. The heat affects the components and the voltage "drifts"; even precision resistors don't help. It's not a big deal here, since the drift is small, but it's there. Perfectionists can simply lift the resistor off the board, which will also reduce the heating of the electrolytic capacitor next to it.


Credit where it's due for the resistors under the mains filter. Not only are the resistors installed in pairs at a minimum, but there are four of them in the PWM controller's supply circuit. There are also resistors both before and after the bridge rectifier. The first ones discharge the input EMI suppression capacitor, while the second ones discharge the power supply filter capacitors.


The power supply is built around the popular PWM controller UC2845, which makes it a single-ended power supply. Another important difference, since the previous ones were based on the TL494. Both PWM controllers were developed at roughly the same time, so they are now among the most classic controllers used in power supplies. This is an advantage, since power supplies like these are easier to repair.


There were a few blunders, too. Chinese power supplies and blunders are brothers for life; only the severity varies.
In this case, I immediately spotted an unsoldered lead on the snubber for one of the output diodes—not good.
There are also tiny solder balls all over the board, along with traces left by wave soldering. These may have no effect at all, or simply burn off the first time the unit is powered up without causing any problems—or they may take out the power supply. The defects are very easy to fix, but the production engineer is clearly getting paid for nothing, if there even is one.


I haven't reviewed a power supply with this kind of topology before, so I was doubly interested in drawing out its schematic. The board may look like it has very few components, but that impression disappears when you look at the schematic.


Next, I divided the schematic into functional blocks. The colors may not have enough contrast—sorry, there wasn't much choice.
1. Red — high-voltage power (hot) section
2. Blue — low-voltage output (cold) section, feedback circuit, and fan power circuit.
3. Green — PWM controller and its standard supporting components.
4. Orange — presumably the soft-start circuit and output short-circuit protection.
5. A color I can't identify — the diode near the transformer, part of the transformer saturation protection circuit.

Most component values and reference designators match the actual board, but the values of some SMD capacitors are approximate, since I didn't desolder them.


This power supply uses a single-ended forward topology, whereas the more common low-power single-ended supplies use a flyback topology.
In the block diagram, I've highlighted the forward converter blocks (on the right) that aren't present in the flyback circuit (on the left). The forward converter adds a diode and an inductor, and one of the transformer windings is connected with reversed polarity (this is important).
There's another difference, too: in a forward converter, the transformer core has no air gap, which is essential in a flyback converter.


Forward converter topology (especially single-ended) is very similar to a classic step-down converter.
In both circuits, the input switch "pumps" the output inductor, which delivers energy to the load through a diode during the off period. The difference is that in a forward power supply, both the transistor itself and the transformer and one of the output diodes act as the switch.
Here are the corresponding blocks, marked with the same colors for clarity. I think it's now clear why I wrote above that this power supply has no output filter inductor: the one installed is an energy-storage inductor. You absolutely must not short it out.


Forward converters are usually used at higher power levels, while flyback converters are used at lower ones. This is because the flyback transformer has an air gap, and its size starts to become significant. Also, spikes are harder to control, and the circuit may operate less stably.

But high-power forward converters have their share of complications, too. In this case, an extra diode and transformer winding have been added to the circuit. This circuit protects the transformer from saturation under abnormal conditions (for example, a short circuit at the load). In the color-coded schematic, this block is marked with the "color I can't identify."
The quote describing this circuit is taken from here (warning: intrusive ads may appear).

This circuit has several significant drawbacks. First, unidirectional currents in the transformer windings require measures to reduce the core's one-sided magnetization. Second, when the switch opens, the energy stored in the transformer's magnetizing inductance cannot discharge on its own, since all the transformer terminals are left "floating." This causes an inductive spike—a rise in voltage across the switching transistor's power terminals—which can lead to its breakdown. Third, shorting the converter's output terminals will inevitably destroy the power stage, so thorough short-circuit protection is required.

The drawback associated with core magnetization by unidirectional currents is common to all single-ended topologies, and is successfully addressed by introducing a non-magnetic gap. An additional winding is used to suppress overvoltage, “discharging” the inductive component with current г3 during the idle phase, as shown in the figure




To avoid burdening readers with unnecessary theory, I’ll move on to the practical part—more specifically, the tests.

The test setup is the same as in my other reviews and consists of:
1. Electronic load
2. Multimeter
3. Oscilloscope
4. Thermal imager
5. Thermometer
6. Wattmeter, not reviewed.
7. Pen pencil and paper.


There is already some ripple at no load, though in this case it’s not a concern.


For the test, I used a combination of resistors and an electronic load.
1. First, I connected two resistors, which drew about 4.8 A. The electronic load brought the total load current up to 5 A.
In my opinion, the ripple is rather high for a 25% load.
2. The same pair of resistors drawing 4.8 A, plus 5.2 A from the electronic load, for a total of 10 A.
Ripple is over 100 mV, and the output voltage has risen slightly. Although that’s a side effect, it’s a useful one in this case.


1. Two resistors drawing 4.8 A, plus 10.2 A from the electronic load, for a total of 15 A.
Ripple has increased quite substantially. The oscilloscope is set to 50 mV per division, with the probe at 1:1; you can work out the rest for yourself.
The output voltage has risen a little more.
2. I added a third load resistor to the two already connected, bringing the total to 7.2 A, plus 12.8 A from the electronic load, for a total load current of 20 A.
Ripple has increased further and is now very noticeable. At the selected measurement range, the oscilloscope screen is barely enough to display it.
The output voltage has also risen slightly, but I should point out one thing. I wrote above that the voltage rises as the unit warms up, but during the test it remained rock steady. Any fluctuations were within the last digit. In other words, increasing the load current raised the voltage, and it stayed there until the next test step—so that’s a plus.


Efficiency measurements have become an integral part of my power supply tests, and I tested this unit too—especially since it uses a different circuit design.
Here are the results:
Input — Output — Efficiency.
7.1 — 0 — 0
144 — 120 — 83.3%
277 — 240 — 86.6%
414 — 360 — 86.9%
556 — 480 — 86.3%

In my opinion, the efficiency is quite respectable—better than that of the previous power supplies I’ve reviewed.


Now, about the temperature and how it’s distributed among the components.
The input bridge rectifier and transformer get the hottest, but in both cases the temperature stays well below critical, so I can confidently say that the power supply could deliver 550–600 W. I’d especially like to point out how cool the power transistors stay: even at maximum power, they never got above 52 degrees.
The test followed my usual routine: 20 minutes of warm-up at 25% power, then 20 minutes at 50%, and so on. The test took about an hour and a half in total, since I decided to extend the final test a little.
For the most part, it didn’t matter how long I tested this power supply, because devices with active cooling reach thermal equilibrium very quickly. The temperature is almost the same after 20 minutes as it is after an hour. With passively cooled power supplies, it takes much longer, so I try to test them for longer.


There was, however, one minor but unpleasant issue typical of power supplies with active cooling. The components stay at a normal temperature mainly thanks to the constant airflow inside the case. When I removed the cover for testing, I noticed the temperature rising quickly. Unfortunately, this is a characteristic of all actively cooled power supplies, and at loads above 50%, running with the fan stopped usually ends badly.

This most often happens because the power transformer overheats. I often stress how important it is to monitor the transformer’s temperature: if it gets too hot, the ferrite loses its properties.
Put simply, here’s what happens:
Imagine a pump (the inverter transistors), a control circuit (the PWM controller), a tank (the transformer), and a valve (the output diodes).
The pump pushes water (let’s say) into the tank, then pauses while the outlet valve drains the water, and the cycle repeats.
The more power we need, the more water we pump into the tank. But then it overheats, and the tank’s capacity shrinks by a factor of five. The control circuit doesn’t know this, though, and keeps pumping as before. Since the cylinder has become smaller, the pump starts working under a heavy overload. There are then two possibilities: either the cylinder bursts or the pump burns out. Since the cylinder is very sturdy, the pump burns out, most often taking the control circuit and fuse with it.
That’s why it’s important to keep an eye not on the transistors, which can reach temperatures of up to 150 degrees, but on the transformer, whose limit is 110–120 degrees.

The power supply has no fan monitoring or thermal protection, so if the fan stops (because of dust or a jam), the power supply will most likely burn out. This can happen with many power supplies, so it’s important to keep the cooling system in good condition.

The photo shows the transformer temperature rising from 92 degrees to 100 in just 20 seconds with the cover removed. In fact, the temperature was lower to begin with; it had simply risen by the time I removed the cover and took the first photo.


On the other hand, the load resistors got really hot during the test, reaching around 250 degrees even with airflow. The electronic load stayed considerably cooler, even though it was dissipating almost twice as much power. However, after the last tests, one of the thermal switches in my load ended up burning out, and it kept trying to shut down well before it overheated. I still haven’t gotten around to making a new version.


Conclusions.
I won’t go through all the pros and cons; instead, I’ll sum up what I found.
The power supply passed the full-load test, and its temperature stayed within the normal range—in fact, below it. This suggests it should also work normally at considerably higher power levels.
But the build quality leaves a lot to be desired. I’m also disappointed by the undersized input capacitors and the poor quality of the output capacitors. This device looks more like a kit for assembling a decent power supply, except it’s been equipped with whatever parts were at hand.

So on the one hand, I don’t want to criticize it too much: the power supply works, and works well. On the other hand, little things like solder splashes, a loose screw, and so on call for some “file work.”

The store provided a coupon for this review: S480power. With it, the price comes to $22.99. In my opinion, even with these shortcomings, the price is quite reasonable. If you don’t mind checking it over and making a few fixes, it’s a perfectly good option. If you want something you can just buy and use, it’s better to get a Mean Well, but it will cost more. The coupon is valid for two weeks.

That’s all for now. As usual, I’m looking forward to your questions and comments. And I’ve got a 600 W power supply waiting for me.