A power supply with two voltage outputs can sometimes come in handy, but alas. + Modification.

Recently, single-board mini computers have become fairly popular (in certain circles)—Cubie, Raspberry, and so on.
But sometimes enthusiasts use these devices to build NAS storage, torrent downloaders, or something similar. These applications usually call for high-capacity 3.5-inch hard drives, which in turn need two supply voltages: 12 and 5 V.
This power supply was intended for just such an application.

As I mentioned above, this power supply provides two voltages, 5 and 12 V, and is designed to be built in—in other words, it has no enclosure.

As always, I'll start with how it arrived. It came in an envelope, packed in an antistatic bag.


The power supply is relatively compact: 125 mm long, 54 mm wide, and 25 mm high.


The first thing that stood out when I unpacked this power supply was how clearly it differed from the store photo. To be honest, that was quite disappointing.
The 220 V connector is mounted directly on the board.



A cable for connecting the load is also included.
Judging by the cable, I can guess that this power supply is intended for a specific application, rather than as a general-purpose supply for DIY projects. That said, one doesn't rule out the other.


The board is single-sided, with components mounted on both sides.
There are quite a few components on the back.
The mounting holes are visible; the distance between their centers is 116 × 47 mm.
The ground contact on the 220 V power connector is not connected to anything.





The power transistor is a fairly popular 4N60.
The store photo clearly shows a PWM controller with an integrated power switch, something like a TOP244Y. It clearly has more than three pins, at least.


I couldn't make much sense of the transformer markings. The first number (80) doesn't look much like a power rating—it's too high—and the third (36) seems too low.


The output power diode assembly for the 12 V rail is marked B10100G, rated for 10 A and 100 V.
The photo isn't great, but you can try to make it out. I didn't see any point in desoldering it.
The output diode for the 5 V rail is a 5 A Schottky SR540.


The output capacitors are two 1000 µF, 25 V units, supposedly low-impedance types.
Considering they operate at 12 V, that's pretty good.
Though on the 12 V rail, there's a “specially trained” jumper in place of the output inductor.


The 5 V rail also has 25 V capacitors, but these are 1000 + 470 µF.
Here, though, the manufacturer did install a noise-suppression inductor.


Out of curiosity, I measured the ESR of the capacitor bank on the 12 V rail. It came out to 8 mΩ, or 16 mΩ per capacitor.
Average, but perfectly normal for this class of power supply, especially considering that cheap power supplies often have much worse capacitors.


The 5 V rail is a bit worse, but its total capacitance is lower too.


The schematic for this power supply. I don't think I got anything wrong.


That wraps up the visual inspection. Next, we can move on to measurements and testing.
First, of course, let's power it up with no load.
The 12 V output is high, while the 5 V output is low. That doesn't tell us much yet; we'll need to check it under different loads.
But at least it works.


Let's move on to further testing.

After measuring the voltages, I moved on to measuring the temperatures of various power supply components.
The measurements were taken with the same loads used at the end of the power test, i.e. at 47 W.

And finally, what power supply test would be complete without measuring output ripple?
All ripple measurements were taken under the same loads used at the end of the power test.
Something unbelievable is going on on the 12 V channel; the oscilloscope probe is set to 1:1.
Apparently, the specially trained jumper wasn't trained very well, and it isn't filtering anything :(




The 5 V channel looks great—I’d even say excellent.




Since I really wasn't happy with this result, I replaced the jumper with a simple, untuned inductor that I desoldered from some monitor.
The 12 V channel has clearly improved; now it looks great too.




After that, I tried one more experiment: I connected a hard drive to the power supply.
I couldn't be bothered to look for a cable with a connector, and they're not very easy to solder anyway, so I soldered some wires directly to the board and connected them to the PSU.
Well, what can I say? This is pushing it a bit: 14.6 V on the 12 V channel is, to put it mildly, a little too much…


This little inductor “ate” most of the ripple on the 12 V channel, and it's installed in the original spot.


But if you think I stopped there, you're very much mistaken.
Looking at the measurement results, I decided to try to improve things. The obvious solution was to add one turn to the transformer on the 5 V channel. Then the PWM controller would lower the voltage on the 12 V channel, after which I could change the resistor values in the feedback circuit and call it a day.
I added an extra turn (I won't show all the photos, just what it looks like) and got about 10.7 V on the 12 V channel and 4.67 V on the 5 V channel.
I thought that was good enough, but when I started changing the resistor values, I hit a snag: I can't raise the voltage because the PSU goes into protection. The best I managed was about 10.9 and 4.72 V.
I don't understand why yet.


What a shame. I could have used the case from an old TV tuner, put this PSU, the Cubie, and a 4 TB hard drive in it.


But I didn't stop there either, and I did finish modifying this power supply.

So, to sum up:
Pros
The power supply works.
With a fairly simple modification, it can have low ripple.
The temperature is perfectly reasonable; at least, I think the power supply is fine in that respect.
The output capacitors aren't exactly top-tier, but their capacitance, ESR, and voltage measurements are all perfectly acceptable.

Cons
The power supply shown in the store listing was clearly a different model—that's the one I fell for (and the photo clearly showed a branded unit, too).
The voltage fluctuations are completely unacceptable.
No inductor on the 12 V rail.

My take: this power supply left me with very mixed feelings. On the one hand, although it's not the same as the one in the store photo, it's fairly average. I've seen much worse power supplies (for example, the ones supplied with USB adapters for hard drives). But the voltage variation is really extreme. There was a modification that should have worked, but for some reason it didn't; I'll investigate why the chip goes into protection mode.
The second option is to switch the feedback loop to one rail and let the other one run free.
One possible use for this power supply would be to pair it with a 5 V controller and use it to control LEDs connected to 12 V.
The 5 V rail surprised me: a decent diode, 1500 µF total filter capacitance, a good inductor—and such a weak rail.
I have a feeling the designers simply made a calculation error somewhere.

The power supply for this review was provided by the tmart store.

I hope this review helped someone. I tried to provide as much information as possible.