Mean Well IRM-60-24ST power supply, sixth of six

And now it’s time for the last, sixth power supply from Mean Well. I decided to leave this one until last partly because its design stands out from the rest, and partly simply because I liked the way it looked. In any case, I found it interesting to take a closer look, and that’s what I’m going to tell you about today.
<cut />


From the manufacturer’s claims, I’d highlight its low no-load power consumption and the note that it is tested at 100% load. The latter at least greatly reduces the chance of getting a “surprise”, though it doesn’t eliminate it entirely.



The packaging is nothing new: just an ordinary cardboard box.


The power supply comes in two versions: one with screw terminals and one for soldering. In the first case, the model designation ends with the ST suffix.


As you’ve probably guessed, mine is the version with screw terminals; the “ears” on the sides of the case are for mounting it.
The case has all the necessary information printed on it: the full model name, input and output voltage, maximum continuous load current, and so on.


But, as before, I’ll include all the information as a page from the datasheet.
The power supply is available with five output voltages: 5, 12, 15, 24, and 48 V. The full list of options is actually much longer, since each model comes in two case styles and three power ratings: 30, 45, and 60 W.
In my case, it’s the 24 V, 60 W model.


The power supply is slightly larger than the EPS-65S-24 I reviewed a little while ago. That’s perfectly logical, though: inside, it’s a fairly similar PSU, but with an additional plastic case.


The manual gives dimensions for both versions. First, the one featured in this review.


And the version for soldering onto a circuit board.


As I mentioned, they differ only in the presence of the mounting “ears” and terminal blocks.


That makes the board-mount version a little more compact.


While looking for some photos for the review, I happened to come across this clever little board. The idea is to mount up to two of these power supplies on it and get a dual-rail output. From what I gathered from the description, this setup is used to power audio amplifiers. And why not? It’s actually quite convenient: take two power supplies and a board like this, and you’ve got a compact dual-rail source.


In this case, the manufacturer has used high-quality terminal blocks, so no complaints there. The case is marked to show what connects where.


Nothing on the bottom.


The bottom of the case can be removed, though you’ll have to wrestle a bit with the ten clips, which keep trying to snap back into place.


But there’s nothing particularly interesting inside; an additional board covers everything. The layout of its contact pads looks very much like the contact layout on the version designed for soldering onto a circuit board. Most likely, the other version simply doesn’t have this extra board. This is also supported by the fact that the power supplies have different heights: the one reviewed here is 4 mm taller.


The additional board is marked IRM-60ST-R3, while the power supply board is marked IRM-60ST-R2. It’s logical to assume there’s another board inside somewhere, marked IRM-60ST-R1.


There’s nothing else to inspect for now, so I’ll move on to testing.
Since the power supply has no adjustments—or any kind of indicator, for that matter—I’ll simply check the output voltage.
In this case, it’s 24.4 V.


The now-familiar test measuring the relationship between output voltage and load, as well as the power supply’s efficiency. The test is performed at two input voltage levels: normal, 220–230 V, and reduced, 105–110.


In terms of efficiency, the power supply performed slightly better than stated in the specifications. Its output voltage regulation was excellent. At a 24 V output, the maximum deviation was about 17 mV across a load range of 0 to 120%. 5 points.


Let's connect the oscilloscope and see what's coming out of the power supply. This test is also performed at two mains voltage levels: 220–230 V and 105–110 V. A filter recommended by the manufacturer, consisting of 100 nF and 47 µF capacitors, is connected ahead of the probe.


In no-load mode, short spikes with an amplitude of about 70 mV are visible. Changing the sweep rate reveals a slight sawtooth ripple.


Now for quite a few tedious oscilloscope traces. Since few people need them, I've tucked them under a spoiler.


For the final screenshot, I combined the waveforms of the high- and low-frequency ripple at input voltages of 225 and 105 V, measured at 100% load.
I have to say, I wasn't very impressed with the results. The manufacturer specifies a ripple amplitude of no more than 150 mV, but in reality I measured peaks with a total amplitude of up to 220230 mV. The peaks are very short, of course, but they're there, and that's a fact. The test was carried out correctly, as described, but the results were better with the previous power supplies.
This ripple is fairly easy to eliminate with an LC filter on the output, but apparently they decided that one should be added at the load input instead.


The power supply is rated for an operating temperature range from -30 (only with a 230 V input) to 70 degrees, but full power is available only at temperatures up to 50 degrees with a 230 V input, or up to 40 degrees for the 5 V output model with a 100 V input.
It is also specified that the unit can be loaded to 100% starting at an input voltage of 100 V.


Because of its sealed construction, I ran the test a little differently this time. I simply left the power supply on the table, with an ambient temperature of around 24-25 degrees, at 100% load and an input voltage of 230 V.


I measured the power supply's case temperature after 20 minutes of operation, then again 20 minutes later. After 40 minutes, the hottest spot was around 60 degrees, which means it was about 25 degrees above the room temperature.


The overload test showed that the power supply shuts down at a current of about 3.20-3.26 A, and this current is almost independent of the input voltage.
I'd call that an average result: the power supply goes into protection at about 130% load. I've seen power supplies with more headroom.


Shutdown process.


Response to an output short circuit.


As you can imagine, I got curious about what was inside after all.



But all I managed to dig out was a block completely covered in a rubber-like compound. It even smelled like rubber.
Around the heatsink, this compound came off fairly easily as a thin sheet.


After poking around a bit, I realized there wasn't much point in trying to get inside—the power supply is completely potted.


The second heatsink is visible on the other side, along with some insulating tape, which I slightly damaged while “digging out” the power supply.


If you look closely, you can see the potting compound on the diode bridge leads as well as around the high-voltage transistor.


After taking it apart like this, I decided to repeat the heating test out of curiosity. This time, though, I loaded the power supply to 110% and covered the potted assembly with a lid.


After half an hour, the hottest spot was at 70 degrees; 20 minutes later, it was at 75 degrees. All things considered, given that the load was 110% rather than 100%, I got roughly the same result as in the previous test—the difference isn't very large.


Since redrawing the schematic isn't very convenient, I'll include at least the block diagram from the manual.
The first unusual thing is the absence of a second optocoupler, which provides overvoltage protection. Of the seven Mean Well power supplies I've tested in total, this is the first one without it. Yet the specifications do claim output overvoltage protection.


Anyway, I decided not to stop there and kept going.
After putting in a bit more effort, I removed the additional board. That gave me access to the board for inspection and some measurements.
The power supply board itself measures 82x47x25 mm.


The power supply's circuit board. It turned out that there is actually one optocoupler, not two.


The teardown was made even more nerve-racking by the fact that, even several hours later, the input capacitor still had about 256 V across it. I'd already discharged it a little with the tester before taking the photo.
The photo didn't turn out very well; it was awkward trying to take pictures and measure the voltage at the same time.


The input capacitor is probably 120 µF (a standard value), while the output filter has a capacitance of 1080 µF. Presumably, it uses three 390 µF capacitors in this circuit.


The input section looks very similar to the EPS-65S-24: the input voltage monitoring circuit is connected in exactly the same way, and the input capacitor is just as reluctant to discharge. They did, however, add resistors to discharge the X-type input capacitor.
You can see from the board that there are two input inductors, essentially the same arrangement as in the EPS-65S-24.


The PWM controller is a GR8875 from grenergy, a company I hadn't come across before.
A small TVS diode connected in parallel with the output terminals protects against excessive output voltage.



Of course, I could have traced the circuit even in this condition, but in this case I figured that hardly anyone would repair a power supply that's been completely potted—unless they were doing it for the challenge. Besides, the circuit itself is nothing new, so I'll just show an example from the controller's datasheet.


Video version of the review


Based on the tests, my only complaint is the ripple level. Everything else is quite good. Output voltage regulation is excellent, temperatures are moderate, and the overload and short-circuit protection work perfectly.
The power supply has a very interesting design, and it's also available as a module for mounting on a circuit board. And, of course, there's a wide choice of output voltages and power ratings. The only thing missing, in my opinion, is a 3.3 V version—for complete happiness, so to speak :)

That's all. I hope you found the review useful, and as usual, I'll be glad to answer your questions in the comments.
I'd also like to thank Vladimir once again for providing the unit for testing.