Another Low-Profile Power Supply, 12 V 20 A

This is my third power supply review, and this time it’s a 240 W, 12 V model. It’s very similar to the previous one: low-profile, budget-friendly, and another invitation to play roulette. Still, it’s best to read the review to find out all about it, roulette included.


The power supply in this review and the previous one do have a lot in common, but if you take a closer look, they’re actually not all that similar.

The packaging is identical, well, except for the size.


The previous power supply looked thin and very long, but next to this one even it seems tiny. The case is 308 mm long and the same 53 mm wide.

This time it really is a 12 V power supply, as I ordered. The only odd thing is that it’s rated at 250 W, but also listed as 12 V and 20 A. Maybe the rules of math have changed, but I’ve always thought 12 × 20 = 240. Let’s chalk it up to a calculation error :)


Last time I complained that the product page said 12 V, but they sent me a 24 V model. This time, you’re invited to play a guessing game from the start: the title lists both options, 12 × 20 and 24 × 10, with no way to choose.


This time the terminal blocks are covered, and they’re noticeably larger and sit straighter, too. Overall, everything looks fine.
Cooling is active, which is normal for power supplies this compact and powerful. There are also a couple of stamped recesses on the bottom for the component mounting screws.


It’s also a bit taller: 22 mm versus 19 mm for the previous model.


The cover fits a little more snugly, but it’s still held in place by clips along its entire length.


The fan speed isn’t adjustable. I expected it to be noisy, but was pleasantly surprised: it runs very quietly. The fan is installed so that it blows cool air into the case.


I’ve taken apart all kinds of power supplies, but this is the first time I’ve held one this long. 31 cm is really a lot. When I opened the case and saw the circuit board, I couldn’t figure out what all that empty space was for. A little later, I realized it was room for the fan :)


When I started inspecting the insides, I immediately noticed that the transformers in both power supplies looked very similar, even though their power ratings differ by a factor of two. The one in the 240 W version is actually a little thicker, by about 3 mm. Since it has active cooling, the manufacturer decided to draw more power from it.


There isn’t even an X capacitor on the input side, let alone a common-mode choke or other small components. The input capacitor is recessed into a cutout in the circuit board, and there’s no heatsink on the bridge rectifier.


The output has a storage choke (not to be confused with a filter choke) and a pair of 3300 µF 16 V capacitors. I would have used four of them, preferably rated for 25 V.


1. The input capacitor is rated at 150 µF and 450 V. It looks like a name-brand part, but its capacitance is low for the rated power, even taking into account the claimed “narrow” input voltage range.
2. The inverter uses two 13N50 transistors, rated at 12.5 A and 500 V.
3. The transformer is also wound with a bundle of thin wires.
4. The output has two MBR30100 diode assemblies, rated at 30 A and 100 V.


Surprisingly, the actual capacitance was close to the stated value. I’m used to capacitors that look like name-brand parts having much lower capacitance, because they’re usually not genuine.


And there was a mishap, too. When connecting the alligator clips to the capacitor, I accidentally pressed on it a little harder than I meant to. There was a small click, and the solder joint on one of its leads broke...
It probably couldn't have done that on its own, but either way, it happened—and, among other things, it immediately showed that the components inside aren't secured.


Unlike the previous unit, this board is also screwed to a standoff attached to the case. This connection is used to ground the case, but the power supply output is not connected to this ground in any way.




The board has narrow sections around the transformer, which also serves as a structural brace.


The PWM controller is a "timeless classic": the KA3845, formerly known as the UC3845. Chinese engineers most often use one of two controllers in power supplies like this: the UC3845 or the TL494, depending on the topology.


Here, the manufacturer has added RC snubbers across the output diode assemblies. The feedback is taken correctly, almost right at the output terminals, then routed along long traces to the voltage regulation circuit, located near the PWM controller.


I traced the schematic from the board and tried to make sure everything matched, but I can't guarantee it's 100% error-free.



The inverter topology is very unusual for a power supply like this: it uses what's known as an asymmetric half-bridge, also called a "single-ended half-bridge". This topology is most often used in welding inverters. In power supplies, you usually see:
1. Low power: single-ended flyback
2. Medium and high power: single-ended forward converter.
3. Medium and high power: half-bridge.
4. High power: full-bridge.


Essentially, it's a hybrid of a conventional full-bridge inverter, but with two switches replaced by diodes, and the inverter itself operates in single-ended mode—that is, both transistors turn on at the same time.


Since my computer-connected electronic load has a maximum power of 150 W, I'll have to use a less sophisticated but more powerful one for the tests.

In this case, the power supply's no-load output voltage is about 12.1 V.


Next came the efficiency test. The key test points were spaced at 2 A intervals of load current, but I'll show the photos in order: 0, 4, 8, 12, 16, and 20 A.
Unfortunately, the power supply couldn't deliver the rated 20 A, and its output voltage dropped to 8.9 V at that current.


I had to go back to the last step where the voltage was still acceptable and increase the current gradually. It turned out the maximum was only 18.1 A; beyond that, the voltage drops linearly as the load current increases. In other words, the power supply can actually deliver only about 215–220 W at most.


Efficiency wasn't too bad, staying at 82–84% across almost the entire load range. The horizontal scale is in increments of 24 W of output power.


I checked the thermal performance the usual way, increasing the load current in steps at 20-minute intervals. There were three stages: 6, 12, and 18 A. The last value was chosen to correspond to the maximum output power mode.
The power supply was operated with the cover on, since it has active cooling. Then the cover was removed, and thermal images and individual measurements with a non-contact thermometer were taken as quickly as possible.

After 20 minutes under a 6 A load (72 W)


Another 20 minutes later, at 12 A (144 W).


And the final measurement, after 20 minutes at 18 A (215 W)


The overall photos show where the hot air exits. The case also gets noticeably warm, of course, but due to the way the measurement works, it appears cool.
The overall photo of the disassembled unit shows that the hottest component is the input bridge rectifier. This is partly because it has no heatsink, and partly because it is exposed to hot air warmed by the transformer and inverter transistors.
Unfortunately, the input capacitor is also exposed to hot air, so its thermal conditions (and therefore its service life) are far from ideal.


After an hour of warm-up, the output voltage under load had dropped to 11.9 V. This was most likely because the current was right at the overcurrent protection threshold.
After removing the load, the voltage was 12.122 V, compared with 12.114 V when the power supply was cold—an excellent stability result.


Throughout the test, the temperature stayed within reasonable limits for every component, although it was noticeably hot inside.


And finally, let's see what is happening at the output. The load also has a switching power supply, but I hope it won't have much effect on the measurement results.


Power supply operating modes: no load, 4, 8, 12, 16, and 20 A.
Two types of ripple stand out here: the main ripple, up to 200 mV peak-to-peak, and the “spiky” ripple, reaching 1.6 V. The difference is that the latter is much easier to suppress by adding an output inductor. It is most likely caused by the inverter diodes switching. Snubber circuits or faster diodes on the “hot” side might help.


Ripple at 100 Hz was measured in four modes: no load, and at 6, 12, and 18 A. At maximum power, the “humps” caused by insufficient input capacitor capacity are clearly visible.
The vertical scale here is 500 mV/div, not 200 as above.


My final conclusions will be very similar to what I wrote at the end of the previous review, though there are a few differences.
The power supplies are built to roughly the same standard—or, more precisely, they are equally poor. There are no filters at all; the only component that reduces interference is an interwinding Y capacitor.
On top of that, this power supply could not deliver the power claimed by the manufacturer. It managed only about 90% of the rated output.

That said, there are some positive differences. For example, the case is actually connected to ground here, so there is a chance of less interference. I also prefer the circuit design, though it is partly let down by cost-cutting on components. Output ripple is lower—or, more precisely, ripple like this is much easier to deal with.

A major drawback is the combination of active cooling and no overtemperature protection, so this power supply should not be left running unattended.

That's all from me. I hope you found the review useful.