A brief review of a low-power compact power supply

This review is a sort of continuation of my look at compact modular power supplies, which began with a review of a similar model from Hi-Link. This time, though, I got my hands on a unit from a less well-known company, with a different output voltage but the same 3 W power rating.


As last time, I have a compact, practically sealed, low-power supply in my hands. These power supplies are used where high power simply isn't needed, and their design also lets you mount them directly on a circuit board.

I ordered a pair, as they say—just in case I lose one and break the other.


At first glance, it's a dead ringer for the Hi-Link HLK-PMxx series.


1. The manufacturer's logo is on top, along with the basic specs: input voltage range (90–264 V), output voltage (5 V), and output power (3 W).
The bottom is smoothly covered with potting compound, so we can consider the power supply sealed, at least in theory.
2. I immediately noticed that the leads were a bit crooked. It looks as if the internals were installed at an angle before potting—and both power supplies are the same.
3. Out of curiosity, I dug out the previous power supply, also rated for 3 W. It turned out to be a little longer (34 vs. 30 mm); the height and width are the same, 15 and 20 mm respectively.
4. The pin layout is different too, and in this case it's reversed: on the Hi-Link, the input pins are close together and the output pins are farther apart, while on the unit under review it's exactly the opposite.
I prefer the layout on this model; it offers slightly better electrical safety.


Now let's move on to the tests. I used:
1. ZKE EBC-A10 electronic load
2. Uni-T UT-181A multimeter
3. Thermal imager
4. Oscilloscope



An automatic test to check whether overload protection is present and determine its threshold.
The test showed that the power supply:
1. Has protection: when it trips, the output voltage drops to zero and recovers after 1–2 seconds.
2. Has a maximum output current of about 600 mA, which is right at the rated power limit (5 V × 3 W = 600 mA).


Further testing revealed a few more quirks.
1. With no load, the output voltage is 4.96 V, which is within spec.
2. If I apply a 100 mA load and then remove it, the output stays at around 4.78 V for a few seconds...
3. After a while, the voltage rises fairly quickly back to its original 4.96 V.
4–8. As the load current increases, the output voltage rises as well.
9. Attempts to load the power supply at 600 mA often tripped its protection. Even after it returned to normal operation, I had to wait about 20–25 seconds before it could handle a load properly; before then, even a 100 mA load would trip the protection again.


This is roughly what the previous test looked like as I increased the load current. The power supply doesn't actually deliver 0.6 A, so I ran the remaining tests at a maximum current of 580 mA.


For a more accurate ripple measurement, I powered the load from a regulated transformer power supply. The oscilloscope probe was connected through a filter made from two capacitors (1 µF + 100 nF). I also checked that there was no ground loop with the computer (I haven't fully galvanically isolated the load yet). Disconnecting the computer didn't change the waveform or its amplitude.


The ripple is quite high: from 300 mV at 200 mA to 400 mV at 580 mA. That's a lot for a power supply with a 5 V output.


Someone once asked me to compare two oscilloscopes: a newer one powered from the mains and an older one running on a battery. In practice, I got roughly the same results as above, just a little lower: 250 and 350 mV at 200 and 580 mA, respectively.
You can also see here that the intervals between the controller's switching cycles aren't uniform, as was also visible above. The switching frequency with no load is noticeably lower than under load.


The temperature performance, on the other hand, is excellent. The power supply first ran at 50% load for a while; then I loaded it at 580 mA and left it running for more than half an hour. The case temperature reached 50 degrees. The potted construction plays a big part here, as it also serves as a decent heat spreader.


Last time, I didn’t take the power supply apart, partly because photos of it are available online, partly because I’d bought it, and partly because it’s extremely difficult to open without damaging it.
This time, though, I had two power supplies, and one of them had been bought specifically to take apart.

So, I removed the plastic casing. It wasn’t easy, but after a while I managed to get it off.


Inside, I found a circuit board with flux residue that hadn’t been cleaned off properly. By the way, at first I thought the board was tilted because something was sticking out and pushing it out of alignment when it was fitted into the case before potting. But it turned out nothing was in the way, so how did the leads end up crooked...?


I don’t really have any solvents for removing potting compound, so I decided to try acetone. As it turned out, it softens the top layer of the compound a little, so you can scrape it off bit by bit.


After repeating the process—pour, scrape, pour—I finally removed some of the compound and found that:
1. The input capacitor is 4.7 µF
2. The output capacitor is 100 µF, 16 V
3. There’s an interwinding noise-suppression capacitor. What surprised me was that its potting compound softens relatively easily with acetone—more easily, in fact, than the compound used to pot the power supply.


At some point I got thoroughly fed up with it. At first, things were moving along fairly well and I could scrape off about 0.5 mm at a time, but then progress almost ground to a halt, so I decided that was enough.


All in all, it’s a very basic power supply. It’s controlled by a PWM controller, and, as you’d expect, there’s no input filter.


The voltage is regulated via the PWM controller’s supply winding, and there’s no output filter either. Hi-Link does have an output filter, which is why it’s a little longer.


Last time, people discussed in the comments what you could use a power supply like this for. Well, how about powering a small clock? :)



And, of course, the verdict.
Yes, friends, this is clearly no Hi-Link. Still, this particular power supply works, doesn’t get too hot, and keeps the voltage within acceptable limits. But it also has excessive ripple, an output current that barely reaches the rated value, and a strange short-circuit protection algorithm.
There are also some less obvious drawbacks: the contacts are arranged opposite to those on Hi-Link modules, so swapping one for the other will make things go bang. The contacts are also a bit crooked, which just looks sloppy.

I’ll put the second power supply to use, of course, but I wouldn’t recommend it. For about the same price, you can get a Hi-Link that delivered one and a half times its rated current in my tests without any problems.