Used 12 V 5 A power supplies

Continuing my power supply reviews, today I’m looking at some boards that aren’t new, but have relatively convenient output specs—which is why I ordered them for review in the first place.
As always, this review includes a detailed inspection, schematic, tests, analysis, and conclusions.

This time I didn’t hold back and ordered five of them at once. Partly so I could compare them, and partly because their specs really are convenient: they can power various LED strips, a mini PC, a monitor, and all sorts of other stuff.

The power supplies arrived in one huge bag, with each unit also packed in its own small bag. There was no additional protection against damage during shipping.


Specifications from the product page
1. Input voltage: AC 100 V–240 V, 50–60 Hz
2. Output voltage: DC 12 V
3. Output current: 5 A
4. Output power: 60 W
5. Operating temperature: -30 to +85 °C
6. Size: 102 x 45 x 26 mm

The boards look very similar, although there are some differences.


The wires have been cut short at both the input and output, which is typical of used power supplies. The components also vary from one unit to another—you can even tell just by their color.


The board is very compact, with tightly packed components. The power supply is noticeably smaller than the “people’s choice” model, even though its power rating is less than one and a half times lower.


The heatsinks are designed to run across the entire board, so the output diode heatsink extends onto the “hot” side, while the transistor heatsink extends onto the “cold” side. There’s protective insulation, using heat-shrink tubing in some places and special tape in others. It’s not a great solution, of course, but even well-known manufacturers use it.
On top of that, some parts of the heatsinks are a little crooked, and they’re also bent inward, so the power supply’s cross-section looks more like a trapezoid than a rectangle. At first I thought they’d been deformed during shipping, and some of them probably were, but a few heatsinks were secured in this position with hot glue, meaning they were already like this before shipping.


So, what do you get for the asking price?
1. A fuse and a decent input filter consisting of two inductors and an X capacitor. They somehow forgot the thermistor, though—there isn’t even a spot for one on the board.
2. They did remember the varistor: 14 mm in diameter, rated for 470 V.
3. The input filter capacitors are Jamicon, 82 µF, 400 V. One board had a different capacitor with lower capacitance, but a 450 V rating.
4. An FTA06N60D high-voltage transistor in an insulated package.


1. A transformer with a PQ26 core.
2. One transformer is wound so tightly that it’s hard to believe everything fit inside. You can see that some of the wires have double insulation, which is good.
3. There’s a Y capacitor, of course.
4. The output has two diode assemblies, each rated for up to 20 A.


There’s no consistency in the output filter capacitance either. Not only are the capacitors from different manufacturers, they’re also rated at 1000 and 1500 µF, in various combinations: 1000+1000, 1000+1500, or 1500+1500. There’s also a small 470 µF capacitor, but it doesn’t do much here because the schematic places it after the inductor, which has been replaced with a jumper. There is, however, a common-mode choke.


Since these power supplies aren’t new, I decided to check the components that are considered “consumables”—the capacitors. I didn’t check every single one, just a sample.
The high-voltage capacitors on three boards measure 81–83 µF, which is perfectly normal. One board’s capacitor is below spec at 74 µF, and the one with a different capacitor measures even lower, at 62 µF—but it was originally rated at 68 µF.
On the one hand, that’s enough capacitance. On the other, I’d replace the capacitors that are noticeably below their stated value. Their ESR is also higher.


Oddly enough, the output capacitors are in better shape: their capacitance is close to the stated value, and their resistance is within the acceptable range for capacitors of this type.
1, 2. 1000 µF
3. 1500 µF
4. 470 µF


The underside is simple and unremarkable: a green board that looks like phenolic laminate, with rubber spacers that were probably left over from the original enclosures.
One significant drawback: since the boards came in enclosures, there are no mounting holes, which is inconvenient.


Some of the boards look pretty good for used units. Still, I recommend checking and reflowing the heatsink mounting pads right away—there are two for each heatsink. On one board they were just loose, and on another the pads themselves had come off.


I noticed something odd: on one board, a 1 Ω resistor was soldered in parallel with the current-sense shunt. At first I thought the shunt itself was simply open, but no—the total resistance here was 0.4 Ω, compared with about 0.6 Ω on the others.

Also, at least one board was really dirty, with traces of water, especially around the PWM controller pins. Boards like this should be cleaned with alcohol first.


No, this isn't another demonstration of a dirty board. I was just turning the power supply over in my hands when I noticed that it actually has output overvoltage protection. It works by shunting the installed TL431 and is implemented with a zener diode and a regular transistor. A second optocoupler is often used for this kind of protection, but here they decided to use the same one as the main feedback loop.


Of course, I traced out the schematic, as always—partly out of curiosity and partly to help anyone whose power supply fails.

Among the odd and unusual features, I'll point out the two diodes and two capacitors in the PWM controller's power supply circuit. I saw this in another power supply quite recently, but I don't think I'd come across this approach before—or perhaps I just hadn't noticed it.


I strongly recommend making the first power-up through a regular 25–40 W incandescent bulb. In my case, the bulb was 15 W, so its filament glowed a little. That's normal.

With no load, the power supply draws about 0.3–0.4 W, which is a problem when using LED lighting where the power is switched on the mains side of the supply. The controller has a "green" mode, so with no load the power supply quietly "whines." The sound is quiet at load currents of 0–50 and 150 mA, louder at 100 mA, and silent at 200 mA.


Overload protection is set to 5.5–5.6 A. It shuts off immediately rather than gradually reducing the voltage. Voltage regulation is average: across the 0–5 A current range, the output drops linearly from 12.18 to 12.07 V.


Efficiency was 84–86% over the 0.5–5.5 A range, peaking at 2.5 A. The scale at the bottom is in 0.5 A increments.


The protection works very accurately: it tripped at 5.6 A, while the power supply ran steadily at 5.5 A.


The thermal test was run with the case open at three current levels—2.5, 5.0, and 5.3 A. Each stage lasted 20 minutes, for a total test time of 1 hour.

I originally planned just two stages, at 2.5 and 5 A, but the power supply behaved steadily and showed no obvious signs of overheating, so I raised the load current to 5.3 A and extended the test by another 20 minutes.


Thermal images at the end of each stage


And here are the temperatures measured with an infrared thermometer. The high-voltage transistor ran hottest, although the transformer reached about the same temperature.
Of course, temperatures would be higher inside the case. These supplies are designed to operate in a small enclosed box, but something tells me they were really intended to be rated around 12 V, 3–4 A.


As always, I checked how the output voltage changes with temperature. For this power supply, the change is relatively small and negative—that is, the voltage drops slightly as it warms up.
1. Cold power supply, no load
2. Start of the test at 5.3 A
3. End of the test at 5.3 A
4. Hot power supply, no load.

The difference between the output voltage of the cold and hot power supply was 30 mV, which is actually pretty good.


And of course, ripple—the bane of modern power supplies. Some people try to eliminate it, while others don't pay any attention to it at all.

1. Since the PWM controller has a "green" mode, with no load the output has short bursts at a frequency of 1.25 kHz.
2. At 2.5 A, ripple is about 100 mV, which is perfectly normal.
3. At 5 A it's naturally a bit higher, around 200 mV, but still acceptable.
4. At 5 A, ripple at 100 Hz is about 30 mV—nothing to complain about there either.


Near the beginning, I mentioned that the power supply has a spot for an inductor, but no inductor is fitted; the schematic shows it as a jumper. I was curious what the output would look like if the manufacturer had actually installed one.
I didn't have a compact inductor rated for 5 A handy, but there was an ATX power supply inductor lying on my desk. It's designed for high current, but has low inductance. Just what we need for the experiment, since compact inductors rated for 5 A will most likely have low inductance as well.
I desolder the jumper and solder the inductor to the back of the board.


Ripple at 5 A with different oscilloscope time bases. As they say, no comment needed—it’s all clear enough. You can compare it with the two lower oscilloscope traces shown above.
The capacitors haven’t been replaced with higher-quality ones, and no ceramic capacitors have been added in parallel, so the results could actually be improved further.


Later, I spotted a slightly more interesting board on Banggood with the same specs, but with an input thermistor, a higher-capacitance capacitor, and an output inductor. It costs $4.42 each — link
I kind of regret not noticing them sooner; maybe I’ll buy a couple while they’re still available.



Conclusions.
The overall quality is nothing special, of course. It’s clear these are used power supplies, and judging by their condition, they were obviously shipped in a sack. The capacitors vary in quality, capacitance, and manufacturer: some boards have a relatively decent Jamicon, while others have something less familiar. There are signs of water damage, so inspect the boards carefully before powering them up for the first time, and possibly wash them. It’s also a good idea to check the solder joints on the heatsink supports.
I’d also count the lack of mounting holes and an output inductor as drawbacks.

Still, all five boards worked, so there isn’t much to complain about. The input has an almost complete filter, including a varistor, but no thermistor, which I consider a drawback. It runs moderately warm for a 60 W unit; at least in that respect, I liked these better than the “popular” ones. Ripple isn’t too high even without an inductor, and overall, I was pleased with how stable they are.

In short, considering that these boards aren’t new, they’re pretty good. I liked them, but the component quality is inconsistent, so it’s better to buy several at once and keep one as a spare.

That’s all for today. I hope this was useful :)