Five boards from sealed 60, 100, 150, 200 and 300 W power supplies
To keep up with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about all new posts there, and might also write about something interesting now and then - link to my account
You can also subscribe to updates and new articles on the Telegram channel - https://t.me/KirichBlog
These power supplies were also sent as part of a comprehensive testing and selection program, but unlike the previous ones, they weren’t sent in the form they’re sold in.
The thing is, sealed power supplies are usually potted with compound, so it’s very difficult to see what’s inside, although I’ve had to “dig into” these supplies more than once. Anyway, the person who sent them made arrangements with the factory, and they sent the boards separately.
Of course, testing a bare board isn’t entirely fair, since potting compound transfers heat better than air, and the enclosure itself makes a decent heatsink. Still, we’ll try to assess the components, circuit design and output power, as far as possible, of course.
As I said, there are five power supplies here. I could have done five separate reviews, but first, that would be more work, and second, at least four of them have a lot in common. So I think it makes more sense to cover them all in one review.

Interestingly, the manufacturer did solder the cables in place, even though I didn’t really need most of them for testing. What’s even more puzzling is that none of the power supplies had the ground wire soldered on, despite having a pad for it, and the wire itself is tinned.

It’s worth highlighting three models separately, and I’ll start with those. They look the most alike to me: the 60, 100 and 150 W versions.

I’ll start in order of increasing power, of course, although there are a few nuances here. They’ll come up toward the end of the review, so let’s not get ahead of ourselves.
So, the 60W-12V, IP67 model - link

There’s no input filter, and the board doesn’t even have a footprint for one. There’s no output inductor either, but overall, while the design is simple, it’s perfectly adequate for this power level.

No complaints about the creepage and clearance distances for electrical safety. And since this power supply is potted, they’re excellent.

As I mentioned above, the cables were soldered in place, but the ground wire wasn’t. There’s a pad for it, though. Why? No idea. Maybe it gets connected to the enclosure somehow later? I’ll have to check continuity on the assembled power supplies.

1. The board is marked 75 W, even though the power supply is rated at 60 W.
2. The 47 µF input capacitor is quite adequate for 230 V, but a bit small for 115 V.
3. The 10N65 transistor is rated for 10 A and 650 V, in theory—more than enough headroom for this power level.
4. The interwinding capacitor is the right type; no complaints.
5. The output uses a 20 A, 100 V diode assembly, also with plenty of headroom. As a reminder, for switching power supplies with this topology, the output diode current is often taken as 3× the load current. So a 15 A diode assembly would be enough for this power supply; this one uses a 20 A part.
6. The output capacitors are 2 × 1000 µF, but rated for only 16 V, which is too low for flyback power supplies; 25 V is recommended.

The PWM controller is an LD7575; the current-sense resistors are 2 + 2 + 1.2 + 1.2 Ω.

The input capacitor measures 41 µF, versus the stated 47 µF. In this case, although it's right at the lower limit, that's still an acceptable tolerance.
The output capacitance matches the values printed on the capacitors.

Most of the tests were carried out with a ZKE EBC-A10H electronic load. I fitted small heatsinks to the transistor and diode, using just thermal paste to attach them.

Checking the protection circuit.
First, at 230 V, the power supply went into protection at a current of about 7.1–7.2 A, corresponding to 85 W.
At 115 V, protection kicked in sooner, at 5.9 A or 68 W. But toward the end, the voltage started to sag, so the actual maximum current was 5.6–5.7 A.

I could only estimate the efficiency very roughly. I still haven't found a suitable accurate power meter to replace the one I lost :(

Ripple was measured using a 0.1 + 1 µF filter. In previous reviews of high-power supplies, I connected the probe directly, because I wanted to see more. But for accurate ripple measurements, you should use a filter.
In both cases, the oscilloscope's 20 MHz bandwidth limit was enabled.

I've already written about measuring with a filter correctly. This wasn't my idea: the recommendation is to place a pair of capacitors right next to the probe tip. This removes the “spikes” from the oscilloscope trace—spikes that wouldn't reach the load anyway, especially if it has even a small capacitor at its input.
By the way, has anyone come across adapters or connectors like these for a standard probe?

Unfortunately, even with the filter, the ripple is quite high, and short spikes still get through. I tested with no load and at currents of 2, 4, and 6 A.
At 6 A (120%), the main ripple was about 400–500 mV peak-to-peak, and even higher if you include the spikes. You need to think carefully about what you're powering: powering an LED strip or a motor is one thing; sensitive equipment is quite another.
But in my opinion, these short spikes are more critical for the capacitors, and are partly determined by them. In other words, if we use high-quality capacitors with low ESR, they'll suppress most of these spikes. With cheap capacitors, the spikes aren't suppressed, so the capacitors heat up more and age faster. That's exactly why I pointed out that the capacitors here are rated for 16 V. With more headroom, the capacitor is physically larger and cools better.
That said, in this particular case the power supply will be potted, which helps keep component temperatures down because it effectively evens them out.

No complaints about the low-frequency ripple. Yes, some got through at maximum load, but it's not that critical. I think that when powered from 115 V, the 100 Hz ripple will become noticeably stronger, though again, that's not so critical.

As for temperatures, I tested several operating conditions, gradually increasing the load. The maximum current was 5 A; you can estimate the test duration from the graph.
I should also note the very good voltage regulation, both as the load changed and as the temperature changed.

Results at 4 and 5 A. I monitored the temperature of the transformer, the high-voltage transistor, and the output diode assembly.
The results were pleasantly surprising. Even with such small heatsinks, the hottest component was the diode assembly. The transformer reached about 92 °C even at maximum load, while the transistor reached only about 71–72 °C.
Overall, I think it can deliver 5 A continuously without any problems.
I would have liked to see the internals before the potting, because I would have stuck a small aluminum plate onto the transistor and diode assembly. It would have cost next to nothing, but helped dissipate heat more effectively.

Later, I ran a similar test, but this time with an input voltage of 115 V. It was a little simpler, though I don't think that made much difference here.

As expected, the transformer and diode assembly temperatures remained practically unchanged, but the transistor got hotter. This is typical when running from a lower input voltage.
Results at 4 and 5 A.
At 5 A, the transformer reached 90 degrees, the transistor was also around 90, and the diode assembly was nearly 130.

The second power supply, 12 V, 100 W - link

The layout is somewhat similar to the previous one, except that the transformer is quite interesting and fairly compact. It is also fitted into a cutout in the circuit board, which improves its thermal performance when potted.
There is no input filter or output inductor. By the way, this may be where a metal plate is attached to the transistor and diodes; its approximate dimensions are marked on the circuit board.

No complaints about safety: the clearances are adequate, and it is potted as well.

1. The marking on the board also indicates a power rating of 100 W.
2. There are now two 47 µF input capacitors, which is good to see.
3. The transistor is the same as last time: a 10N65.
4. The interwinding capacitor is the right type, no issues there. You can also see the transformer construction better here.
5. There are two 20 A output diode assemblies, which is more than enough for 8.3 A.
6. There are three capacitors instead of two, but they are still rated for 16 V

Even the PWM controller is the same, but here the current-sense shunt consists of five 1 Ω resistors, giving a total resistance of 0.2 Ω.

In reality, we have 80 µF on the input and 3100 µF on the output, which is quite enough for operation at 230 V.

As before, most of the tests were run with the EBC A10H load, but I had to use a more powerful one for the overload test because the current went over 10 A.
I also used the same cooling setup, with heatsinks simply resting on the components and held in place only by thermal paste.

When I connected it, I spotted an error in the board's silkscreen: the polarity is marked incorrectly :)

The protection tripped at around 13–13.3 A with 230 V input, dropping to 11–11.2 A at 115 V. In my opinion, it would be worth lowering it a little more, but that's just my preference.

I measured efficiency over the 1–13 A range, just to be safe, although that's already more than a 1.5× overload.

Ripple with no load and at currents of 3, 6, and 9 A.
The results are practically identical to those of the previous power supply: the main ripple component is around 400–450 mV at maximum current, rising to 1.2 V including the spikes.

The low-frequency ripple looks a little better, but that's because the capacitance per watt of output power is higher here. This should also help when running from a lower input voltage.

As before, the test was run at 230 and 115 V, but I could only get up to 7 A.

The thing is, the board had already heated up quite a bit at 7 A, even though the stated maximum current is 8.3 A.
The transformer got fairly hot, reaching around 96 degrees, and one of the output diodes exceeded 135 (the thermal camera goes into overload at that temperature). The other diode, however, only reached 109 degrees. I don't know why; perhaps the heatsink wasn't making as good contact, or maybe the circuit board layout or the characteristics of the diode assemblies had something to do with it.
. The high-voltage transistor reached about 75°C, which is perfectly normal.
It’s worth remembering that I tested it “in the open air.” I think that in a metal enclosure filled with potting compound, temperatures would probably be lower and the power supply would reach its rated maximum of 8.3 A.

The test at 115 V was shorter, but the maximum was still 7 A.

The transformer temperature was slightly lower, probably because the test was shorter. The transistor, however, got hotter, reaching 98°C. That was to be expected: the lower the voltage, the harder it has to work.
The output diode assemblies showed the same results as before.

The third power supply, 12 V, 150 W, 12.5 A - link

This board is noticeably different, mainly in its layout. The components are mounted on both sides, and the capacitors lie flat on the board rather than sitting in a cutout. The same goes for the transistor and diode assemblies.
The board still has no input filter, but it now has a thermistor, which was missing from the previous boards.

The transformer is installed like the one in the previous model, in a cutout. As I mentioned, this arrangement should help it cool better by transferring some of its heat to the enclosure. That’s probably why it overheated during the previous tests.
The assembly is neat, with adequate clearances.

1. The PCB marking matches the stated power rating of 150 W.
2. On the input side, there’s an X capacitor and a pair of Y capacitors in addition to the thermistor, but the ground wire is still not connected to anything.
3. The input capacitors are 47 µF each, but they’re slim, and the board itself looks very flat. Perhaps there’s a “slim” version of this power supply.
4. The high-voltage transistor is now rated for 15 A and 650 V; the previous ones were rated for 10 A.
5. The interwinding capacitor is the proper Y type, no complaints there.
6. The diode assemblies are still the same 20 A parts. That’s fine, though, since their combined 40 A rating still exceeds the recommended 12.5 × 3 = 37.5 A.

The PWM controller is still the same LD7575. The current-sense shunts are interesting, though: the 100 W power supply had a 0.2 Ω shunt, while this one has four 0.82 Ω resistors, giving us 0.205 Ω. So the protection current should be slightly lower than, or equal to, the previous model’s, even though this power supply is 1.5 times more powerful.

Once again, I’ll point out the very sensible design and transformer placement. Since the transistor and diode assemblies are mounted slightly above the component leads, heat is probably transferred to the enclosure using either thermally conductive rubber or a metal heat spreader. Something tells me the first option is more likely.
At any rate, I think they put some kind of thermal interface between the transformer and the enclosure, though even the potting compound improves heat transfer.

The input capacitors have a combined capacitance of about 81 µF, which is rather low for 150 W, especially when powered from 115 V.
There are three 1000 µF capacitors on the output, for about 3000 µF total. Essentially, we have the same setup as in the previous power supply, and once again the capacitors are rated for just 16 V.
So the difference between this model and the 100 W one comes down to two key things: a more powerful transistor and a larger transformer, plus a different method of transferring heat away from the components. In my opinion, that’s not quite right. If you’re scaling up a power supply, scale everything, not just a few individual components.

Here I installed slightly larger heatsinks on the transistor and output diodes. More importantly, I screwed them onto the components instead of just laying them on top.

The protection trips at about 15.7 A at 230 V and 11.5 A at 115 V. The 100 W model tripped at about 13.3 A and 11.2 A, respectively. It’s a little strange that the threshold changed only with a 230 V supply.

I measured efficiency over the 1–15 A range. Nothing unusual overall; it could have been a bit higher.

Ripple with no load and at 4, 8, and 12 A.
The large ripple was to be expected: this power supply uses the same controller, a slightly different transformer, but the same pair of output diodes and, most importantly, the same three capacitors as the 100 W version.
So at 8 A, we get roughly the same result as the 100 W model did at 9 A, while at 12 A we get 600 mV and some very large “spikes”. I'm practically certain that 16 V capacitors won't last long under these conditions, unlike the 100 W version, where they're operating under less stress.

There are some small 100 Hz ripples at low frequencies, but that was also predictable: the input still has the same 80 µF, while the power is one and a half times higher.

Temperatures at 9 and 12 A.
Overall, no complaints. Yes, at maximum power the transformer heated up to 108 degrees, but remember, it's operating in open air here.
The high-voltage transistor is around 100 degrees, one output diode assembly is at 116, and the other is around 105. The difference is understandable: the components are at the edge of the heatsink, and one assembly simply gets less cooling. I think temperatures will be a little lower in a case filled with potting compound.

With an input voltage of 115 V, I didn't raise the current above 9 A. The input capacitance is a bit low for this voltage, and the transistor usually runs hotter in this mode.
The results: transformer, 90 degrees; transistor, 100; output diode assemblies, 97 and 91 degrees. I wouldn't recommend running this power supply above 10 A for extended periods at 115 V.

Two more power supplies. Actually, one of them would more properly belong with the previous three, but it just so happened that I had these units separately. Partly because they differ from the others in their dimensions as well as their power rating.

The fourth power supply, 12 V, 200 W, 16.7 A - link
Actually, this power supply was fifth in all my tests, and there's a reason for that: the board marking misled me, but more on that later.

This one has both an input filter and a thermistor, and the capacitors are clearly more substantial, which already hints at a higher power rating.

The board is a little dirty, but I have no safety concerns; the required clearances are there.

1. Input filter. Simple, but it's there. The board marking is also visible here: 400 W.
2. Interestingly, a Y capacitor was installed instead of an X capacitor. This may be partly related to the poor soldering; I'll explain why I think so later.
3. The transistor is the same as in the 150 W model: 15 A, 650 V.
4. Some unusual shunts: one through-hole, the other SMD.
5. The transformer is flat too, with a very tightly packed winding.
6. Not only are there three output diode assemblies, they're rated for 30 A each, for a total current rating of 90 A. That's pretty good for a stated current of 16.7 A—almost a twofold margin.

The PWM controller section; the interwinding capacitor is visible here too. Unlike the previous three power supplies, this one uses a different PWM controller, the OBGZ12CP. The component farther to the right appears to be an emitter follower for driving the high-voltage transistor.

While inspecting it, I noticed something odd: there's a Y capacitor and a pad connected to it, but it's unclear where the pad goes or what it's for.

The person who sent me these power supplies also sent a photo of a similar unit's board, marked as 250 W, though that's not really the point.
Looking at the photo, you can see that the units are so similar they're almost indistinguishable—in other words, it's clearly the same board, with only minor differences. For example, the input interference-suppression capacitor is a different type, and the lower version doesn't have the Y capacitor I mentioned above, although there's a place for one. Oh, and the lower version has jumpers in place of the input choke.

The input capacitors have a total capacitance of about 200 µF, which is quite reasonable for 200 W, at least on 230 V mains. For 115 V, I'd like to see more.
The output capacitance is even a little higher than specified: it uses four 2200 µF capacitors, for an actual total of about 10,000 µF. Unfortunately, the capacitors are 16 V again. Unlike the previous three power supplies, though, there’s plenty of room here—you could easily fit capacitors rated for 25 V or even 35 V, with the same capacitance.

For testing, I installed heatsinks on the transistor and output diodes. The heatsinks look a bit oversized, but that’s because I was misled by the board’s labeling.

Most of the tests were run with an Atorch CL24-based load, but I had to use my 40 A, 600 W load to measure efficiency, since the current exceeded the CL24’s 25 A maximum.

The efficiency curve looks like this. I tested it from 1 to 29 A. At 29 A, the power supply starts reducing its output voltage, so we can say the protection kicks in at around 28.8 A. When hot, it’ll probably be closer to 28–28.5 A.
Honestly, when I got close to 30 A, I felt a little uneasy, especially given the power supply’s topology. But it survived :)

Ripple: I wasn’t expecting anything good here, given that this is a flyback converter running under these conditions, with no output filter inductor...
I tested it with no load and at 8, 16, and 24 A. Why those settings instead of the more appropriate 6, 12, and 17 A? Once again, it was because I got the labeling mixed up. The photo above even shows testing at 22 A.
The most relevant ripple measurement here is at 16 A, since that’s closest to the rated maximum. It came to about 0.5 V excluding spikes, or around 1.5 V including them. The lack of an output filter inductor is to blame. I was expecting worse, though I wouldn’t call this good.
But adding an inductor presents a dilemma: it would significantly reduce output ripple, but sharply increase the pulsed current through the capacitors upstream of it. Given that they’re rated for 16 V, that certainly wouldn’t do them any favors.

At low frequencies, everything looks fine. The last screenshot shows 100 Hz, but that’s normal since the power supply was running at 150% overload :)

Because of my mistake, I tested it under more demanding conditions.
First, the results at 15 and 18 A. As a reminder, the rated maximum current is 16.7 A. I have to say, the power supply behaved remarkably well under these conditions. Even at 18 A, the transformer was only at 86 °C, with no potting compound, and it was essentially overloaded. The transistor was only at 80 °C, and the output diode assemblies were around 90–100 °C.

But since I thought this was a 300 W power supply at the time, I pushed the load to 21 A and then 22 A.
Temperatures rose, of course, reaching critical levels: 105 °C for the transformer, 104 °C for the transistor, and 100–120 °C for the output diode assemblies (the one closer to the transformer ran hotter because it was warmed by it).
I ended the test at 22 A, with the load drawing about 265 W—quite a bit more than the rated 200 W. Honestly, when I realized my mistake, I almost felt sorry for the poor little board. Despite all my efforts to overload it, it just kept working.
Of course, if I’d left it running like that, the temperature would have continued to creep up. But it kept working at 130% load, which is a good sign—or, more precisely, it suggests that it should run just fine at its rated 200 W. Reduce the load to at least 180 W, and it should be even better.

And the last power supply: 24 V, 300 W, 12.5 A. Link to the 12 V version.
The Chinese sellers accidentally sent a 24 V board for testing, even though the linked model is 12 V. There’s a slight difference in efficiency: all else being equal, the 12 V version will always run hotter because it has higher output current, greater losses in the output diodes, and so on. The difference isn’t huge, but it’s definitely there.

This power supply is quite different from the previous one, even though they look similar in many ways. For one thing, it uses a different topology: an asymmetrical half-bridge, which is more suitable for high power.
Perhaps the only thing this power supply does worse than the previous ones—and other flyback converters—is its narrow input voltage range.
The board has an input filter, and there are also footprints for fuses. I assume that by replacing the two 400 V capacitors with 200 V ones, reconfiguring the jumpers, and changing the fuse position, it could be adapted for a 115 V input.

I have no safety-related concerns about the PCB layout.

The transformers on the previous board and this one are the same size, but, as they say, “there’s one catch”: the transformer in the left-hand power supply is more powerful. This is because a flyback power supply has an air gap in its transformer, so at the same frequency and with the same dimensions, it will have a lower power rating than a transformer without a gap. But a flyback converter doesn’t need an output inductor, which simplifies the circuit design.
Every topology has its advantages and disadvantages; there’s no getting around that.

1. Here’s what confused me: the power supply labels. The person who sent me these units said they were rated at 60, 100, 150, 200, and 300 W. That seemed clear enough, except the labels on the boards read 75, 100, 150, 400, and 300. Yes, the 200 W power supply is labeled 400, so I assumed the one labeled 300 was 200 W, and the one labeled 400 was 300 W. Though something inside me protested: a 300 (400) W flyback is beyond the pale...
2. Surprisingly, the board even has a varistor—rated for 560 V, admittedly, but still.
3. As on the previous one, there are two 120 µF capacitors.
4. The transistors are 10N45; judging by the marking, they’re rated for 10 A and 450 V.

1. The interwinding capacitor is a standard high-voltage type, rated for just 2 kV.
2. As I mentioned above, there’s an output choke, wound on a sendust core with a two-wire winding.
3. The output has four 20 A, 200 V diode assemblies connected in pairs, since this topology has what are effectively upper and lower diodes.
4. There are three 1000 µF, 35 V capacitors. That’s perfectly normal for this topology, as are the selected diode assemblies.

Looking at this, the hair on my head stood on end, and my hand started feeling around in the desk drawer for something heavy, like a hammer...
But then the whole picture came together, explaining both the wrong interwinding capacitor and the poor transformer soldering on the previous board—which, of course, doesn’t make the soldering itself any less awful.
I think this is what happened: someone asked for five boards without enclosures. Three were finished, one was almost assembled and just needed the transformer soldered in, and another was only half assembled. The unfinished boards were handed to some student and he was told to solder them. He somehow managed to solder everything, mixing up the capacitor in the process. They checked that the boards worked and shipped them, since they weren’t really intended for sale or for use in a finished product.
I’m planning to ask the factory about the capacitor: was this a mistake, or are they all made this way? If it’s a mistake, no big deal. If they’re all made this way, that’s a serious problem—it shouldn’t be done like this. I’m leaning toward it being a mistake, since the soldering is done by hand and everything is correct on the other boards.

The same soldering across the whole board, and the output resistor is actually made up of two resistors.

I hadn’t come across this PWM controller before, but I like it. It uses a rare type of overload protection: latch-off.
There are two main types of overload protection:
1. The PWM controller enters a cyclic start/stop mode
2. The PWM controller reduces the output voltage, effectively limiting the output power.
But there’s a third type: latch-off. When an overload occurs, the controller stops switching and won’t restart until the high voltage has been removed for a while—about half a minute. In some ways, this is the better and safer approach, but it can be less convenient in certain situations.

The total input capacitance is 220 µF, which is just about acceptable for this kind of power supply. More would be better—even the stated 120+120 µF, or preferably more still.
The output is fine: 3000 µF.

For cooling, I used heatsinks from the Atorch DL24EW load. I realized later that they were a bit small for this power supply, but I thought it was rated for 200 W.
There was one little problem: how to clamp the heatsink against the components. Then, unexpectedly, the transistors came to the rescue—they were absolutely perfect for the job. I never thought they’d come in handy; I was already thinking of throwing them out. I recommend reading about them and my experience.

As before, I used the CL24 load. I actually liked it; the only thing that takes some getting used to is that after a reboot it resets the current setting to 1 A, unlike the DL24MP, which remembers the last value.

The maximum output current is 14.8 A. At 14.9 A, the voltage drops and the power supply shuts down. I happened to catch this moment by chance.
In my opinion, the protection threshold is set quite well: although the rated power is 300 W, the protection trips at 350–360 W.

Efficiency across the 1–14 A range is pretty good for a power supply of this rating, although I would have liked it to be a bit higher, since this is a 24 V unit, not a 12 V one.

Ripple was checked with no load and at currents of 4, 8, and 12 A, corresponding to roughly 100, 200, and 300 W of power.
As expected, the results here are excellent. A total ripple amplitude of 75 mV at maximum power is impressive for a 24 V power supply. Note that there is no output filter inductor here; adding one could reduce the ripple even further.
A reminder: ripple is also assessed relative to the voltage, because, for example, 200 mV on 5 V and 200 mV on 48 V are two very different things.

The low-frequency ripple is also fine, basically. There is a little something there, but nothing significant.

Thermal performance was measured in four stages: first at 4, 8, and 12 A for about 20 minutes each, then the current was reduced to 10 A and the unit was run for another 45 minutes or so. The final step was necessary because the power supply started overheating.
Several components overheated: the transistors, transformer, and inductor. The transistors most likely overheated as a result of the transformer overheating, which in turn was caused by the lack of potting compound. The unit simply pushed its components beyond their normal operating conditions. In other words, it is designed to transfer heat from these components to the case through the potting compound. Without it, they had a harder time, especially the output inductor.
The output inductor's temperature is not as critical, because, to quote -
In short, the picture looked something like this (components installed in pairs heated up differently; I’m giving the maximum reading):
8 A — transformer 72, transistors 71, diode assemblies 86/57, inductor 86. Everything is excellent here.
12 A — transformer 107, transistors over 135 (the clamp they were mounted under was at 115), diode assemblies 118/75, inductor 121. The transistors are clearly overheating, at the very least.
10 A — transformer 95, transistors 101, diode assemblies 101/65, inductor 103.
One mistake was that I initially thought I was testing a 200 W power supply, so I installed relatively small heatsinks, even though the aluminum enclosure has a clearly larger surface area. In other words, I essentially pushed the power supply into an operating mode it wasn’t designed for.
I don’t see any problem with running it at higher power when it’s installed in a metal enclosure filled with potting compound.

You may have noticed that I gave two temperatures for the diode assemblies, and that the readings differed noticeably. The thing is, these aren’t simply diodes connected in parallel. They’re located in different parts of the output circuit and operate under different conditions. What’s more, sometimes the diode assemblies themselves are different.
The thermal image clearly shows that the diode assemblies at the bottom get hotter than those at the top. The proximity of the inductor or anything else has no effect, since the board is almost symmetrical. I showed the same thing earlier in my review of flat power supplies, where the higher-power versions had the same topology and the same temperature difference between the diode assemblies.

So, here’s what I’m getting at. This particular power supply could be made to deliver more output power with a few simple changes—not by the user, of course, but by the factory. All they’d have to do is put the hotter diodes at the top (as shown in the thermal image above) and the cooler ones at the bottom.
In practice, they’d need to swap them around—or, more precisely, change the board layout. That would essentially cost nothing, and they probably wouldn’t even need to change the component placement program.
This would make the heat distribution more even: the hotter diodes would be on one side of the enclosure, while the cooler ones would be on the side with the transistors, heating them less.
And if they also made the output inductor a little more robust—using a slightly larger core and winding it with thicker wire—they could increase the power a bit more. Unfortunately, that would raise the cost somewhat.
So there’s a way to make a real improvement, but apparently the designers don’t bother with details like this. In my opinion, though, increasing the power “with no strings attached” would be quite easy and worthwhile.

I’m planning to do some more experimenting soon to reduce the ripple. I wanted to do that for this article, but it’s already gotten too long, so I’ll wrap up and move on to the conclusions.
Conclusions.
The 60, 100, 150 and 200 W models are practically identical, so the conclusions are the same for all four. I have two complaints about them: high output ripple and 16 V capacitors. The 200 W model’s capacitors can easily be replaced with 25 V ones, which should extend their service life.
I have no complaints about the power output of at least three of them: the 60, 150 and 200 W models. I accidentally ran the last one at 130% of its rated power :) I do have some concerns about the 100 W model, which overheated. But remember, these boards are designed to be installed in a metal enclosure and potted with compound, which significantly changes the components’ thermal balance. As I mentioned, potting also helps extend the capacitors’ service life, since heat is their enemy, and the compound helps reduce their temperature to some extent. But in this case, that’s more of an assumption.
As for the input capacitors, I have no complaints about the 60, 100 and 200 W models. I’d like to see a little more capacitance, but what they have is basically fine. The 150 W board, however, has only 80 µF. That’s probably tolerable on our mains, but it’s too little for 115 V.
The 300 W model stands out because it uses a different topology. I’ll say nothing about the board’s assembly quality—I’ve already explained everything above—but I liked the design. What I liked even more was its very low output ripple, its interesting overload protection, its full input filter, and, well, the model in general. Yes, you saw some overheating in the thermal image, but that’s probably because the heatsinks are smaller than they should be and the board isn’t potted. The manufacturer could improve this model, though; I’ve described how.
The uneven heating does leave me with some questions about how the 12 V version will behave at maximum power. Unfortunately, I don’t have a separate board, and even if I did, it would still be difficult to assess. One option would be to at least measure the enclosure temperature where the transistors and diode assemblies are mounted. I’ll have to think about that.
In a nutshell. The 60, 100, 150 and 200 W models have issues with their capacitors and ripple; the 300 W model has uneven heating. Of all five, these were my favorites:
The 60 W model is decent, though not very powerful.
The 200 W power supply turned out to be surprisingly durable. I’m pleasantly surprised.
The 300 W model has a sound design and decent components. I simply liked it, especially because its output ripple is very low.
The 100 W model raises some concerns about heating; it might do better in a potted enclosure. Otherwise, it’s a decent power supply. I’d especially note that it should work reliably across a wide input voltage range.
The 150 W model would be decent overall, but its input capacitance is only 80 µF: acceptable for 230 V, but too low for a lower input voltage.
That’s probably all for now. I hope this was useful. I may have forgotten something, but if you have any questions, I’ll try to answer them. Though, for the most part, I’ll be happy even just to hear your comments. And thanks again to Vitaly for providing the samples.
аплатов блок питания форум от 0 до 70в
Questions and answers
Which sealed power supplies did the author like best?
Of the five models, the author singled out the 60, 200 and 300 W units. The 200 W model proved robust, while the 300 W model stood out for its circuit design and very low output ripple.
What ripple did the 60W-12V power supply show?
At 6 A, the main ripple amplitude was about 400–500 mV, with short spikes adding to it. The author advises taking this into account when powering sensitive equipment.
What issues did the author find in the 60–200 W power supply boards?
The author noted concerns about the output capacitors and ripple. The 150 W model has only 80 µF of input capacitance: acceptable for 230 V, but too little for a lower input voltage.
This model is included in the summary table: Comparison of power supplies: ripple, efficiency, and temperature rise
The figures from this review are in the measurement database: Power supplies
Discounted products
You may be interested in
Discounted products
All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!
Only registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.
Log inRegister



