RPS-120-27 power supply made by Mean Well

I often mention Mean Well power supplies in my reviews, and even more often in the comments. This brand’s power supplies fall into a price range that isn’t exorbitant, while still offering very high quality. One of my regular readers sent me this power supply specifically for testing, and I’d like to thank him for that. He also asked me to put it through its paces, which is exactly what I’ll be doing today.
If you’re interested in branded, but not exactly cheap, power supplies, I think you won’t be disappointed.



First, a little about Mean Well. It’s a Chinese company that specializes in various power supplies. The company is relatively young, at only about 35 years old. I often use its power supplies and recommend them to others. Usually, I’m talking about the RS and LRS series. These power supplies are designed for long-term operation under relatively demanding conditions and undergo a preliminary burn-in test. The specifications state that they can withstand an input voltage of 300 V for up to 5 seconds.
Unfortunately, things aren’t always so rosy with Mean Well. They had a series of plug-in power supplies that had a fairly high failure rate. I have one of those somewhere at home, though it’s already taken apart.

But I digress.
So, the power supply was bought from the US store mouser.com. I didn’t ask much about the intermediary; I think the person who sent it to me will answer that question if he wants to.

The power supply arrived in its “original” packaging: a resealable bag, with the unit inside a bubble-wrap envelope and wrapped in soft material. The contact side of the power supply had extra protection.


First impression: very neat. No complaints there.


The power supply has a 4 × 2-inch form factor—or, to put it in familiar terms, the size of two matchboxes side by side, since a standard matchbox is 50 mm long, almost exactly two inches.
The power supply is 29 mm tall.
There are two versions of this power supply: a bare board and one in a perforated enclosure.


For comparison, here’s the power supply from my previous review, the one under review here, and the “people’s choice” model, which I also reviewed several years ago.


The input capacitor has a label listing all the main specifications:
Input voltage: 100–240 V
Input current: 2.3–1.1 A.
Mains frequency: 50/60 Hz
Output voltage: 27 V
Output current: up to 3.15 A without forced cooling, or 4.5 A with forced cooling.
Power: 85 W without forced cooling, or 121.5 W with forced cooling.



The manufacturer was right to list only the basic specifications, because if you read the full specs, they might leave an inexperienced user a little... puzzled.


On top of everything else, this power supply has a few special features (something has to justify the price).
For example, it has various protection features and a safety rating that allows it to be used in medical equipment.


It’s rated as compliant with the MOPP and MOOP requirements of IEC 60601-1 3rd—more specifically, 2 × MOPP.
The table is shown below; the power supply under review meets the requirements in the bottom row.
Power supplies for medical equipment must first and foremost be safe for people in terms of electrical safety, so they have improved insulation. They must also be safe for nearby equipment in terms of electromagnetic interference, but that’s covered separately.
In any case, this power supply was intended for a bench power supply, not medical equipment. Still, extra safety can’t hurt a bench power supply either—as a bonus, albeit not a free one.
By the way, I’ll come back to this topic in one of my planned reviews.


The power supply is packed very tightly—I’d even say VERY tightly. I’ll come back to this later, both during the review and in the conclusions.


The first thing that catches your eye is the input EMI filter, which takes up a fairly large area of the circuit board. It consists of two common-mode chokes, as well as X- and Y-class capacitors. Power is connected via a standard connector with a 3.75 mm pitch, but since only two of the three contacts are fitted, the distance between them is 7.5 mm.


Here are the main components of this power supply.
1. There are two fuses at the input, one on the live wire and one on the neutral. This solution is necessary to improve safety, and ideally this is exactly how it should be done because of the way switching power supplies work. Interestingly, the fuses are not only different colors, but are also soldered in differently: the distance between the holes in the board is greater for the brown one than for the black one, so it sits a little crooked.
2. There are also two thermistors at the input, installed on the live and neutral lines.
3. Although the input connector has no ground pin, the power supply can still be grounded: the ground connection is brought out to the pads around two mounting holes, one on the input side and one on the output side.
4. The bridge rectifier is mounted on a heatsink. The layout is so cramped that one component presses against another. The bridge rectifier is secured with a countersunk screw. At one point I was concerned that the inductor was pressed against it, but I relaxed when I saw that there was a rigid spacer between the screw and the inductor.


1. The input capacitor has a capacitance of 180 µF, which is quite good for an input voltage of 220–240 V, but may not be enough for 100–120 V. Like some of the other components, the capacitor is held in place with silicone sealant.
2. The heatsink is U-shaped, with a high-voltage transistor 22NM60N mounted inside. It is rated for 16 A, 600 V, and an on-resistance of 0.16 Ω. An excellent transistor.
And this brings up a perfectly reasonable point about repairability. As you can see in the photo, for example, the bridge rectifier cannot be replaced without removing the heatsink. At least they left a “gap” for the transistor, although its surrounding components are right next to it, so the heatsink would still have to come off for a repair.
3. The input and output capacitors are made by Rubycon. Mean Well often uses capacitors from this manufacturer in its power supplies; they are known for their fairly high quality.
4. As you can see in the fourth photo, the input capacitor is essentially “hanging in midair”—or, more precisely, supported by silicone. Underneath it are:
A thermistor that monitors the power supply’s temperature to protect it from overheating. It is about 10 mm from the transistor heatsink.
The PWM controller’s supply capacitor. Another point against the designers: it is very awkward to replace.
Several SMD components
One of the two Y capacitors connecting the primary and secondary sides of the power supply.
The ground wire.


1. The output diode assembly is also mounted on a heatsink.
2. The diode assembly is marked STPS20170CT and rated for up to 20 A at up to 170 V. Since the maximum specified output current is 4.5 A, it has almost a 1.5× safety margin.
Unfortunately, repairability here is just as poor as with the components on the first heatsink: the diode assembly can only be removed together with the heatsink.
3. The transformer is wrapped in tape so thoroughly that it is practically impossible to get to. Nearby are the snubber components (on the right), the fan connector (in the center), as well as a trimmer and the power indicator LED.
4. The output filter consists of two 680 µF, 35 V capacitors, one 100 µF, 35 V capacitor, and an inductor.
In my opinion, the output capacitance is low, but I see this quite often in power supplies from this manufacturer. My biggest concern, though, is where the capacitors are installed: they are squeezed between two heat-generating components—the output diode heatsink and the transformer. I’ll skip the fact that one of the capacitors cannot be replaced without removing the heatsink, but I wouldn’t put capacitors in such a hot spot. Still, the manufacturer offers a three-year warranty.


The ground wire connects the primary and secondary sides of the power supply.
The photo may make it look like you can pull it out easily, but trust me, that’s an optical illusion—the wire is almost taut.


It also runs between the heatsink and the transformer, creating a gap. The wire and heatsink are both held in place with the same silicone sealant.


I have no complaints about the soldering quality at all: everything is very neat, the board is clean, all components are additionally secured with adhesive, and there are protective slots where needed. The board is double-sided.


The high-voltage section of the power supply.


The low-voltage section has unpopulated component positions. I assume there is an “extended” version of this power supply. Interestingly, there are no SMD jumpers on the board, and only one regular jumper (not counting the ground wire).


1. The power supply is controlled by an NCP1239 PWM controller made by ON Semiconductor.
2. To the right are resistors that discharge the capacitor installed before the bridge rectifier. There is a footprint for a dedicated IC on the board, but it is unpopulated.
3. Unlike cheap power supplies, this one has two optocouplers: one handles feedback, while the other protects against output overvoltage if the feedback circuit fails.
4. The optocouplers are also secured to the board with glue, and there is a protective slot beneath them.
5. Two Zener diodes are installed on the low-voltage side near the optocouplers. The lower one provides overvoltage protection.
6. The fan is powered through a 7812 regulator, but there is no speed control—or rather, there is in theory, but at a load current of 500 mA, the voltage at the regulator's input rises to 14 V and the fan runs at full speed. As the load on the power supply increases further, the voltage rises to 15–16 V.
On the plus side, this output can be used to power additional low-power devices, for example in the same bench power supply, but the capacitor before the 7812 needs to be increased.
Unfortunately, the transformer windings are connected, so there is no galvanic isolation between these outputs. But as far as I can tell, they are wound separately after all, so there is a chance they can be split into two (if needed).


Of course, I redrew the schematic. I apologize in advance for any errors; it wasn't very easy to do.
The values of the small capacitors are approximate, since they would have to be desoldered from the board for an accurate measurement.


I highlighted several sections of the schematic in different colors.
1. Orange — output overvoltage protection. If the output voltage rises above 32–33 V, the optocoupler disables the PWM controller.
2. Green — overtemperature protection, using the same PWM controller input. The overtemperature and overvoltage protection circuits have a latch that resets when the power supply is disconnected from the mains and the supply capacitor discharges.
3. Pink — voltage regulation circuit.
4. Blue — additional 12 V output for powering the fan
5. Red — an unidentified component connected in parallel with one of the input choke windings. The board labels it as a diode; it has a glass body with color bands and looks like a high-power Zener diode, a spark gap, or even a capacitor (some capacitors came in this type of package).
The diode at the bottom: I'm not entirely sure its anode is connected as shown in the schematic. There weren't many other options, since the cathode is connected as shown.



The circuit is quite complex. For comparison, below is the schematic of a “people's” power supply, which is larger, has almost half the power, and has neither overtemperature protection nor a fan output.
I think even someone with little knowledge of electronics will notice the difference.



The PWM controller is quite feature-rich, but one item looks suspicious:
Frequency Foldback Down to 26 kHz and Skip Mode to Maximize Performance in Light Load Conditions.

And to be honest, this particular feature really worried me. Basically, there's nothing wrong with it except for one thing: at minimum load, the controller drops the frequency to its lowest setting, which often leads to problems like “buzzing” and output ripple.
This mode is considered “energy-saving,” but in practice it only benefits everyone except the consumer. If it weren't there, the difference in power consumption wouldn't be that significant, but marketing is winning for now: everyone has a “green mode,” so we need one too.


PWM controller block diagram. Lots of timing circuits and comparators. The controller has two main operating frequencies—65 and 100 kHz—plus 26 kHz in energy-saving mode.


Simplified application circuit. It shows the minimum required connections. Interestingly, it also includes an overvoltage protection circuit, but it's connected somewhat differently. In this circuit, an increase in voltage sends a high level to the Fault input (the temperature sensor input). In the actual power supply, the optocoupler pulls this pin to ground, so the logic works in reverse.
Looking at the block diagram, you can see that there are actually two comparators: one responds to an increase in voltage on this pin, and the other to a decrease. The Mean Well designers didn't use the first mode; they simply combined the overtemperature and overvoltage protection circuits.


All right, time to move on to the tests.

As I mentioned above, there's a trimmer on the board that lets you adjust the output voltage.
The power supply was initially set to about 27 V; the minimum and maximum voltages are approximately 24 and 30 V, respectively.
The adjustment is quite smooth, and I had no trouble setting it to exactly 27 V.


Next came tests of the power supply's load capacity and ripple level.
1. No load; the oscilloscope trace shows that the power supply is operating at a very low frequency. I'll show it more clearly later.
2. Load current: 1.1 A.
3. 4.5 A, the maximum declared current
4. Then I got a little carried away and raised the current to 5.5 A. I didn't notice any particular problems, although it was clear the power supply was running at its limit.
The power supply "rings" in every mode, especially noticeably when running with no load and under overload.


The test was performed at currents of 1.1, 2.3, 3.4, 4.5, and 5.5 A, corresponding approximately to load currents of 25, 50, 75, 100, and 125% of the rated power.
In no-load mode, the frequency is very low—if I understand everything correctly, about 1.2 kHz. The sweep is set to 500 µs per division.
As the load increases, the frequency immediately rises to 26–30 kHz. If you keep increasing the power, it then goes to 50 and 70–75 kHz.
This seemed a little strange to me, since the specified operating frequencies are 26, 65, and 100 kHz.

In any case, ripple amplitude at the maximum settings is about 75 mV, which seems high to me. But according to the power supply datasheet, that's still half the specified value (150 mV).


I also measured ripple at 100 Hz. The amplitude is slightly higher here: up to 100 mV at currents up to 4.5 A, and up to 150 mV in overload mode at 5.5 A.


For the thermal test, I used the same load currents—1.1, 2.3, 3.4, and 4.5 A—but also ran a test at 5 A.
During the test, the power supply's output voltage rose slightly. When I removed the load at the end, I expected to see a voltage well above the initial setting of 27 V, but the result was better than I thought: it was only 28 mV higher.
It's worth noting that the manufacturer specifies only 85 W without forced cooling. I loaded it to 135 W, and in a confined space at that. In other words, the power supply was overloaded both in terms of power and cooling.
By the end of the test, I could clearly smell overheated varnish, but everything was working normally. The whole test took 1 hour and 40 minutes.


Thermal image after the test.


After that, I let the power supply cool down and decided to make the test more challenging. I took a TN 61 transformer and connected its windings to get 100 V in autotransformer mode.


I confirmed that the voltage at the power supply input barely changes with the load (1.1 A on the left, 5.5 A on the right), then moved on to repeat the tests, this time with a lower input voltage.


In this mode, the high-frequency ripple increased slightly, but I wouldn't say by much: it went from about 75 mV to about 85 mV.


Ripple at 100 Hz stayed at about the same level, but when I raised the load current to 5.5 A, it increased to 1.2 V. The low input voltage and insufficient input capacitor capacity are to blame. Ripple amplitude starts rising sharply at around 5.1 A, so the power supply effectively delivers its rated specifications even at the minimum input voltage of 100 V.
To be honest, when I tested it at a load current of 5.5 A, I expected the power supply to fail, but everything went fine. I also checked the short-circuit protection; there were no problems here either. The power supply shut down and automatically turned back on when the overload was removed.


I also tested its immunity to input transients. The same TN61 transformer served as the "noise generator." The power supply did great in this respect: it was quite difficult to provoke any response, but I eventually managed it :)
Note that the ground wasn't connected, so the power supply's immunity to interference was reduced.


The test at 50% load current is on the left, and at 100% on the right.


Mostly out of curiosity, I also decided to check the power supply's efficiency, especially since the manufacturer claims it's 90%, though without specifying at what voltage.
But I ran into a small problem here. At normal mains voltage, everything was fine, but at a 100 V supply voltage, my wattmeter started showing something strange.
As an experiment, I connected the wattmeter before the transformer and got more realistic readings.
The wattmeter is connected after the transformer on the left, and before it on the right. Load currents: 1.1 A and 4.5 A.


A quick explanation: the photos are arranged in pairs. The test at 220 V is on the left, and at 100 V on the right. The first photo on the right shows the power consumed by the transformer at no load. Since the power measurement error is fairly high with this method, I’ll give an approximate efficiency in the results.


At 220 V, temperatures stayed at a perfectly acceptable level, especially considering that the power supply was running without active cooling and in a confined space.
At 100 V, however, the temperature rose noticeably, and efficiency dropped accordingly. The final test ended with an emergency shutdown triggered by the overtemperature protection.
The temperatures shown for the transistor and output diode are those of their heatsinks at the component locations, since it wasn’t possible to reach the components themselves.


So, what can I say in conclusion? In my view, “opinions are divided,” and there are reasons for that.
I know that people usually skip through all the tests shown here and read the conclusions. That’s fair enough: I usually include the testing process “for reference,” while the results and summary table give a better picture of the conclusions.
I’ve shown the table; now for the conclusions.

I don’t know—maybe I expected more from this power supply, but the test results were somewhat disappointing.
First, the design. Overall, the build quality is high, but I have to point out a few shortcomings: the components are packed too tightly, repairs would be awkward, and the output capacitors are poorly positioned. I understand that power supplies like this are essentially disposable, and in more developed countries people simply throw them away and buy a new one. But where we live, power supplies are often repaired, and I wouldn’t want to desolder a heatsink just to replace a capacitor.

Its performance is also a mixed bag. The power supply performed very well in the load tests, and the output voltage stability is good too. It has both short-circuit and overtemperature protection, as well as overvoltage protection circuits. But I’d rate the ripple as high, even though it remains comfortably within the specified limits. Efficiency is roughly as claimed: around 89–90%, against a stated 90%.

The biggest disappointment was the “green mode,” which makes the power supply noticeably “buzz,” and that’s quite unpleasant. I was glad to see that a fan can be connected, but disappointed that its speed isn’t adjusted according to temperature. Adding speed control wouldn’t be difficult or expensive, but they didn’t do it.

Overall, the build quality is good, repairability is poor, and performance is somewhere between average and good, with a few caveats. I’d expect reliability to be good, since temperatures stay low under normal operating conditions and the components used are high quality.

That’s all for now. As usual, I’ll be glad to answer questions and read your comments. I hope you found the review useful.