UM-U120S power supply, 12 V 10 A

It's been quite a while since I last reviewed a power supply—one of those "feeders" that power all sorts of devices that need less than mains voltage but more than a battery can provide.
So, at the suggestion of one of my regular readers, I ordered several models to test. Today I'll be looking at the first one.


To get ahead of myself, this power supply managed to surprise me, but not with how it performs—with some of its circuit design "quirks."
But first, as always, let's take a look at the packaging.

The power supply came in a plain white box with nothing on it but the part number.


There are three versions of this power supply:
5 V, 15 A
12 V, 10 A
24 V, 5 A.

I have the 12 V version on my bench. You may have noticed that the 12 V and 24 V versions are rated at 120 W, while the 5 V version is considerably less powerful, at just 75 W. That's because low-voltage power supplies usually have lower efficiency.

It's built around a U-shaped aluminum chassis that also serves as a heatsink. There's no protective cover.


I won't repeat the dimensions; they're shown in the drawing.


There is one three-position terminal block at the input and one at the output: live/neutral/earth and 12 V/common/12 V, respectively.
The terminal blocks differ slightly in design. The input one has a protective ridge to prevent the wire from being pushed in too far.
The output terminal block may also be a different type, with four positions.


Terminal assignments shown in the drawing


To access the circuit board, remove the four screws holding it in place and the two screws clamping the power components. There's a protective pad underneath the board.


The board is generously coated with sealant, but even so, you can see that it has a full mains input filter and an output inductor to reduce ripple. That's partly why I chose this power supply for the review: at least from the outside, it looks like a properly designed unit.

The high-voltage transistor and output diodes have extra insulation, and there's even thermal paste in places.


1. In addition to the fuse and input EMI filter, there's a 470 V varistor.
2. There are two 82 µF input capacitors, plus an additional inductor that forms a CLC filter with them. You can also see the thermistor here.
3. The PWM controller is a fairly well-known OB2269; I've seen it in other power supplies. The interwinding capacitor is the correct type: Y-class.
4. The high-voltage transistor is a 12N65 in a fully insulated package.
5. The output has a pair of MBR20U100 diode assemblies. Interestingly, the transistor and diodes even have insulation fitted over their leads.
6. The output also has three 3300 µF, 16 V capacitors and an inductor. It's unusual to see such a large output capacitance in a relatively low-power supply, but the fact that the capacitors are rated for just 16 V is a problem.
I suspect the manufacturer tried to make up for the capacitors' mediocre quality by increasing their capacitance.


The assembly is fairly neat. It's no Mean Well, of course, but the board is relatively clean and the soldering is tidy.


Rather than go through the rest of the power supply's components, I decided to sketch out its circuit.


But while doing that, the unit managed to genuinely surprise me with some very original design choices. I've highlighted them in color for clarity:
1. The thermistor is placed after the capacitors, not before them.
2. The filter inductor is on the negative rail, not the power rail.
3. An unexplained Zener diode and resistor are connected to the PWM controller's supply circuit.
4. An equally unexplained Zener diode is connected in parallel with the optocoupler's LED.

Honestly, I can't think of a single explanation for this circuit design other than that the designer was in a hurry and made mistakes. Or, alternatively, the power supply may have been modified and some of the “extra” components were left in, with the thermistor simply installed in the wrong place.


Let's put it back together and connect it.
The output is about 12 V, and no-load power consumption is around 1.1–1.2 W.
The adjustment range is 11–14.3 V, although you can actually turn it down further, to about 9 V. But below 11 V, with no load, the power supply becomes unstable and starts cycling on and off.


Voltage regulation under varying loads is average, although since I connected it through a terminal block, there may have been some voltage drop across it.
Below are the no-load voltage readings and readings at currents of 3.5, 7, and 10.5 A.


It has overload and short-circuit protection, which trips at around 12.3–12.5 A. Once the overload is removed, the output voltage recovers automatically.


In my opinion, the efficiency could be a bit higher. It’s around 84–85% across the main operating range.
The graph below shows load current along the horizontal axis, from 1 to 12 A in 1 A increments.


I checked ripple with no load and at currents of 3.3, 6.6, and 10 A. The probe was connected using a filter made from 1 + 0.1 µF capacitors.
The main ripple component is fairly small, but there are some unpleasant spikes with a peak-to-peak amplitude of up to 200–220 mV at maximum current. I wouldn’t call that excessive, but given the output inductor, I expected less.


The low-frequency ripple is fine, thanks to the large input and output capacitance. The test conditions were the same as above.


As always, I checked the power supply’s thermal performance in 20-minute stages, gradually increasing the load.
Tests were run at 2.5, 5.0, 7.5, and 10 A.

Unfortunately, I couldn’t run a long test at 10 A: after about 5 minutes, the power supply shut down due to overheating. To be clear, that was after running at 7.5 A long enough to warm up, followed by another 5 minutes at 10 A.


Thermal images after the 5 A and 7.5 A stages, and after switching it back on at 10 A. Since the shutdown was unexpected, I switched the power supply on again for the last photo, waited for it to shut down, and took the thermal image.
The thermistor gets hotter than any other component, even though it serves no useful purpose in this power supply.
That makes the transformer the component most critical to overheating. It reached about 100°C here, which is high but still tolerable. If it gets any hotter, the protection will simply shut the power supply down, so that part works as it should too.


While it was warming up, I checked voltage regulation, and have no complaints here. The reading on the left is with the power supply cold; the one on the right is after it shut down due to overheating.


The overtemperature protection doesn’t latch, and the output voltage comes back. But unlike after an overload, when it recovers in 1–2 seconds, this takes about 30 seconds.


Conclusions.
When I first ordered it, what caught my eye was that, at least on the outside, the power supply looked properly designed. Yes, there are some cost-cutting measures: unknown components are used instead of name-brand ones. But overall, it looked pretty good.
The tests also produced fairly good results, in principle. But to get the full rated 120 W continuously, you’ll need to provide at least a gentle airflow for cooling. And 7.5 A is already close to the limit, so extra cooling wouldn’t hurt there either.
The ripple amplitude is relatively low. I’ve seen power supplies with both lower and higher ripple, including both name-brand and no-name models.
It’s good to see the manufacturer included overtemperature protection. That makes this power supply much harder to damage.

That said, the power supply has a few oddities, especially the thermistor placement. It should be before the capacitors, not after them; as wired here, it just heats the air.

In short, it seems pretty good overall, but I expected a little better. Still, at the current price, it’s a decent competitor to the popular budget power supplies.
I’m thinking of converting this power supply into a UPS for LiFePO4 batteries, along the lines of what I did here. It’s well suited to the job, and the thermal protection will be useful for improving safety.

The store provided coupon BG0784b9 that should bring the price down to $11.99. It’s valid until August 15.

That’s all from me. I hope you found this useful.