Five More Boards for Sealed 36, 60, 100, 150 and 200 W Power Supplies

Once again, power supplies from a Chinese factory have arrived for testing. As you may remember, these tests are aimed either at finding models that offer the best balance of quality and price, or at testing products and getting factory-level issues fixed. I want something simple: power supplies with a decent level of quality that are still affordable.

To keep up with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about every new article there, and might also write about something interesting from time to time: link to my account

You can also subscribe to updates and new articles on the Telegram channel - https://t.me/KirichBlog


Sealed power supplies are considerably harder to test. The reason is that they’re not really finished products, but rather semi-finished assemblies whose thermal behavior once potted can differ significantly from that of the bare board. Still, I tried to make the test conditions as close to real-world use as possible.
The manufacturer was kind enough to accommodate our requests and sent boards for five models in the state they’re in before potting. I’ll explain later why that matters.
These power supplies will eventually be sold in enclosures, and I’ll be able to provide a link to them then.

These models caught my eye when the factory first sent photos. Even then, I had a good feeling that they might deliver positive results. Now they’ve arrived, I’ve tested them, and I’d like to share the results.

The manufacturer thoughtfully packed each power supply in a small box. I assume they’re sold in the same boxes, so you can get an idea of their approximate size.


Just five boards, “as is.” The first things that caught my eye were the color of the capacitors and the fact that they’re all marked as coming from the same manufacturer. To be honest, I don’t know what’s under the heat-shrink tubing, but it’s quite possible they really are all from the same company. That could be either a plus or a minus, depending on the measurement results. Put simply, they’ll probably all be equally good or equally bad. I think you remember those capacitors whose actual capacitance was half the stated value.



Yes, all the power supplies are 12 V and differ essentially only in power.

I’ll start with the smallest one, rated at 36 W.


It looks neat. The components don’t have any additional cooling—or rather, the transistor transfers heat to the board, while the output diode is left “in the open air.” Though once the board is potted, the operating conditions will clearly be less demanding.


1. There’s no input filter or thermistor, but there is a varistor—though it comes with a “catch.”
2. The input capacitor is 33 µF, which is quite enough for our mains voltage and 36 W of power, but a bit low for 115 V.
3. The high-voltage transistor, PWM controller and associated components are all implemented in the conventional way.
4. The transformer and Y capacitor next to it look fine too.
5. The output diode is rated for 3 A, which seems a bit low for a 3 A power supply. It would be better to use a pair of diodes in parallel, but let’s see how it handles the temperature and remember that the board will be potted.
6. There’s no output filter either, and there’s just one output capacitor: 1000 µF, 16 V. For reliability, I’d prefer a 25 V capacitor.


I have no complaints about the PCB itself; everything looks great. But remember the “catch” with the varistor? It’s connected across the mains before the fuse, which is a violation of the rules. A varistor should always be placed after the fuse.



The capacitors’ actual capacitance turned out to be quite close to the stated value, which is already a nice surprise. I don’t think anyone likes it when a manufacturer tries to pull a fast one, even on something like this.


I tested the low-power supplies with a ZLEtech EBC-A10H battery tester, which let me produce neat graphs. The connection was, of course, four-wire.
So, at a mains voltage of 230 V, the cold unit delivered a maximum of 2.9–3 A, and the warmed-up unit did about the same.
At 115 V mains, the board managed 3.1–3.2 A when cold and almost 3.2 A when hot. That’s actually strange; usually, at a lower input voltage, the protection kicks in sooner.

In any case, you can see that the protection threshold is set right up against the rated power. Beyond that, the controller starts reducing the voltage instead of shutting down completely into protection. In any case, it would be nice to raise the protection threshold a little—just 5–10%—and it would be absolutely perfect.



First, the high-frequency ripple. As expected, it is quite large because this model has no output filter and uses a flyback topology. It also has just one output capacitor.
The main component is around 500 mV, with a peak-to-peak value of up to 2 V, including the spikes.


The low-frequency ripple is generally fine: there’s no 100 Hz component. The large peak-to-peak value is due to the high-frequency component.


During the tests—or, more precisely, after they were over—I started wondering: what if...
1. Check the output capacitor’s parameters
2. Try replacing it with a genuine Panasonic capacitor
3. Try eliminating the effect of lead length, since the capacitor in this model is mounted on its side.


So I took a new genuine Panasonic capacitor and the original one, compared their parameters, and found that they were practically identical—which is certainly good news.


Next, I made three comparisons—or rather, six. As a reminder, I usually measure ripple with a C+C network in front of the probe. This makes the oscilloscope trace look a little cleaner and more accurate.
But sometimes you need to see the whole picture, so the measurements on the left are taken with just the probe, and those on the right with 1 µF and 0.1 µF capacitors connected in parallel with the probe.
First pair: original capacitor
Second pair: Panasonic.
Third pair: the same Panasonic, but soldered flush to the board.

In all cases, the load current was 2.5 A, and I used the same load.
You can see that the Panasonic barely reduced the spike amplitude, but it did reduce the main component. This is because the capacitors have almost the same ESR, but the Panasonic has about 20% more capacitance.
But when I eliminated the effect of lead length by soldering the capacitor vertically, the spikes also got smaller.


The spikes got smaller because capacitor leads are often made of steel and have noticeable resistance, which degrades the filter’s performance.
That’s why this kind of circuit is usually laid out using the connection shown on the right. With the layout on the left, the trace resistance adds to the capacitor’s ESR, making it less effective at suppressing high-frequency ripple and interference. Keep this in mind when routing your boards.

Put simply, the capacitance affects the main part of the ripple, while the capacitor’s ESR affects the spike amplitude. In other words, you can solder the original capacitor vertically and get spikes like those in the last pair of screenshots, but the main component will remain the same as in the first one.
In practice, these spikes are easily suppressed by the wires, especially if there’s a capacitor after them. They have a greater effect on the capacitor’s service life and on radio interference.



I tested the temperature rise in four modes: 1, 2, 2.5, and 2.8 A. I didn’t go any higher because the protection might kick in.
I should note the excellent stability of the output voltage: it didn’t fluctuate with either the load or the temperature rise. Well done.




Results at 2.5 and 2.8 A. The only thing I can really fault is the output diode’s temperature. Once it’s potted in compound, it will most likely run cooler than it did here, exposed to the air.


The high-voltage transistor was barely warm, which bodes well for operation at 115 V. A lower supply voltage usually affects its temperature in particular, so I ran a relatively short test at 115 V; there wasn’t much point in going longer.


And yes, the transistor got warmer, but it was still nowhere near a critical temperature. The temperatures of the other components barely changed. So this power supply should work just fine at 115 V, too.



The second power supply, rated at 60 W.

This one has heatsinks on the transistor and diode assembly. That’s why I want to thank the manufacturer for sending the power supplies exactly as they’ll be potted in compound. Last time, it wasn’t entirely clear whether the components would have heatsinks or not, and that matters when potting the board, too.


The overall build is just as neat, and I have no complaints so far.


1. There’s still no input filter, but a thermistor and an X capacitor have been added, which is certainly useful.
2. The input capacitor is 68 µF, which is plenty for 230 V, but still a little low for 115 V.
3. The transistor is rated for 10 A and 650 V, more than adequate.
4. The PWM controller marking is slightly different: the previous one was 281Q12, while this one is 281P08, but something tells me they’re one and the same.


1. Transformer, Y capacitor, optocoupler—no issues.
2. The output rectifier is rated for 30 A and 100 V, with plenty of headroom.
3. Another photo of the optocoupler and feedback capacitor.
4. There are two output capacitors, 2200 µF and 680 µF. Unfortunately, both are rated for 16 V, but at least this one has an output filter, which is great to see.


No issues with the circuit board: the clearances are adequate, and there’s a gap under the Y capacitor. And yes, the varistor is installed correctly here.


The input capacitor is slightly undersized at 60 µF instead of the specified 68, but it’s within tolerance in principle. No issues with the output capacitors.


The protection is set rather crudely and trips somewhere around 8–8.1 A. The power supply handles the load extremely well, but nearly 100 W is a bit much for a 60 W power supply.
A reminder: ideal protection should work like this—if the overload is brief, allow a high current; if it’s sustained, allow 10–20% above the rated value. This isn’t always possible to implement, but I’d really like to see the protection current not set so high. Say, 30%, but not 60–70%...
Or add thermal protection.



I have no complaints about the ripple. As usual, I try to check it in four conditions: no load, and at 33, 66, and 100% load. Here, I tested at 2, 4, and 6 A.


At low frequencies, 100 Hz starts to show up at 6 A. It’s not exactly critical, but it will increase when powered from 115 V.


The temperature test was run at 2, 4, and 6 A, or 24, 48, and 72 W—above the rated 60 W.
At 4 A, everything looks good: the transformer is at 63°C, the transistor at 55°C, and the rectifier at 78°C.
I had no complaints at 6 A either, but apparently the transistor photo didn’t save (that happens sometimes: you press the button, but the photo isn’t saved), so I’ve included the output capacitor temperature instead. In any case, the transformer was at 82°C, which is excellent. The transistor temperature was also fine: it was only 55°C at nearly 50 W. The only concern is that the rectifier heated up to 107°C, which is a bit high, though still within spec, especially considering the 20% overload.




The third power supply, rated at 100 W.



Here, the transistor and rectifier are intended to be mounted to the aluminum case, which acts as a heatsink. Otherwise, this is essentially a larger version of the previous two models.


1. The input now has a full filter, plus a thermistor and varistor. The specified capacitor value is 100 µF, which, as before, is fine for 230 V mains but a little low for 115 V. There’s a decent 0.2 Ω, 2 W shunt near the transistor.
2. As in the 60 W model, the transistor is rated for 10 A and 650 V.
3. The PWM controller photo didn’t come out well, but it looks similar to the ones in the previous power supplies.
4. A proper Y capacitor is used between the windings.
5. The output rectifier is the same as in the 60 W model, rated for 30 A and 100 V, and it has plenty of headroom here too. As a reminder, for a flyback converter, it’s normal to choose an output diode rated for three times the load current.
6. The output has three 1000 µF, 16 V capacitors and an inductor to reduce ripple.




No issues with the circuit board: the clearances are more than adequate, there’s a protective slot at the narrow point, and the varistor is downstream of the fuse.
By the way, the varistors are rated for 560 V peak. They’re only there to suppress transient spikes, since the design doesn’t allow for repairs or fuse replacement.



And once again, I have no complaints about the capacitor values; they’re all within measurement error.


Since the circuit board is designed to transfer heat to the case, I had to improvise again. I used a couple of heatsinks from electronic loads. By the way, I’m not a big fan of devices that dissipate heat through just one side of the case. It would make more sense to mount the transistor and diode assembly on opposite sides of the case.


I ran the first load-capacity test with the same load, but 10 A wasn’t enough to test either the cold or the hot power supply. I should note, though, that the output voltage regulation is excellent, as it was on the previous models.


I had to continue the tests with a different load. It turned out that the protection starts reducing the voltage at somewhere between 10 and 11 A, which is perfectly normal for a power supply like this.


I made a rough estimate of the efficiency—rough because the power is quite low, so the wattmeter’s measurement error becomes significant.


Ripple was measured with no load and at currents of 3, 6, and 9 A, which is close to 33%, 66%, and 100% of the rated load.

The high-frequency ripple is excellent: even at maximum load it was around 25 mV.


The low-frequency ripple is also pretty good; at least, there’s no 100 Hz component.


The long-duration test was run at 3, 6, 9, and 10 A, for a total of more than 2.5 hours.


Thermal images after testing at 6, 9, and 10 A.
I have to say the results are quite reasonable, especially considering that the power supply was tested outside its potted case.
In fact, even 9 A is already above this power supply’s rated output: it’s rated for 100 W, but was delivering 108 W. I ran the 10 A test as an extra, especially since that was already 20% above the maximum rated power.

The results:
At 6 A: transformer 71°C, transistor 43°C, diode assembly 70°C.
At 9 A: transformer 90°C, transistor 55°C, diode assembly 97°C.
At 10 A: transformer 101°C, transistor 63°C, diode assembly 106°C.

As you can see, there’s some headroom for higher ambient temperatures, and the input transistor also has some margin for operation at reduced voltage. Overall, I have no complaints here. The results will of course be different inside the case, but I don’t think the difference will be that great.



While I was at it, I wanted to show how the temperatures of different capacitors compare. They’ll of course even out inside the case, but I just wanted to show how operating conditions affect them.
The first reaches 88°C, heated by the high ripple level and nearby components.
The second reaches 77°C, mostly heated by ripple.
The third, however, is only at 55°C. There’s almost no ripple here, so it’s mainly heated by other components.

Here’s the point: capacitors before the inductor always operate under more demanding conditions because of the higher ripple level, so they fail noticeably sooner. The capacitor after the inductor has an easier time because there’s almost no ripple.
That’s why the first capacitors need to be selected more carefully. There are usually more of them, and they have a shorter service life. There’s one more thing to consider, though: the output capacitor can also suffer if the load itself has a strongly pulsed current draw.




Fourth power supply, 150 W.



The design is almost identical to the 100 W model. In fact, all four models use the same circuit topology; they just add a few components (filter components and more capacitors) and increase the transformer and capacitor capacity. That’s basically it.


1. The input filter is almost complete; no complaints. The capacitor is 120 µF, acceptable for 230 V but too small for 115 V.
2. The transistor is now rated for 16 A and 650 V, compared with 10 A on the previous models, which makes sense.
3. The PWM controller is the same as on the previous models, but it looks like an emitter follower has been added to increase the PWM controller’s drive capability and improve the operation of the high-voltage transistor.
4. Y-type interwinding capacitor.
5. There are now four output capacitors, compared with three on the 100 W model.
6. Diode assemblies rated for the same 30 A and 100 V, but here there are two in parallel, no questions there.



The circuit board is almost fine; I have a minor concern about one of the traces, which is relatively close to the output circuits, but there should be about 4 mm of clearance there. Plus, the board will be potted, so I wouldn't call it a critical issue, though I would have liked a little more clearance, or a slot.
I do have a concern about the small solder droplets on the transformer. I would have rejected this if it weren't for the potting compound, which will keep them from getting where they shouldn't.


The input capacitor's capacitance is within tolerance, but I'd still like to see more, say 140–150 µF.
The output capacitor is fine.


I've already installed three heatsinks on the board here: one on the transistor and a pair on the output diode assemblies. The transistor has the same heatsink as the one used during the 100 W model tests.
But I should point out that the power components are laid out correctly here: heat from the transistor is carried away to one side panel, and heat from the diodes to the other.



All the tests here were done with a load based on the CL24. By the way, this load is for sale. One has already gone, and I'm still using the second one myself; at least it's been tested and works great.

To make testing easier, I decided to use the load's own voltage measurement with a four-wire connection.


The protection is set to about 14 A, meaning the voltage is still normal at 13.8 A, but starts dropping at 14 A. Well, I think that's fine; at least it'll operate more reliably.


Efficiency is about 86.5%—neither good nor bad, just an ordinary result.


High-frequency ripple is also very low. These are the results with no load and at currents of 4, 8, and 13 A, so no complaints here.


But low-frequency ripple at 100 Hz shows up, especially at maximum current. This is because the input capacitor's capacitance is still a bit too low for this power level. Accordingly, I wouldn't recommend using this power supply with a 115 V input, unless you significantly limit the power—I think to no more than 100–120 W.


The thermal test followed the usual routine, at currents of 4, 8, 13, and... 12.6 A, for about 20–25 minutes per stage.

1. No-load voltage.
2. 4 A, cold power supply
3. 8 A, warm power supply
4. 13 A, board already noticeably warmed up



Everything was going well, but as the unit warmed up at 13 A, the output voltage started to drop slightly, meaning the power limit was kicking in, apparently because the current-sense shunt's resistance was drifting.
I first reduced the load current to 12.7 A. The voltage returned to normal, but then started dropping slightly again.
I lowered the current to 12.6 A and finished the test at that level.

Basically, the protection is set precisely to 150 W, which I think is actually a good thing: it's much harder to burn out this power supply than models with the protection current set too high.
The power supply held its output voltage very well, regardless of load power and temperature. Kudos.




Results after testing at 8 and 12.6 A.
At 8 A, I have no complaints at all—well, almost none. The temperatures of the output diode assemblies seemed a little odd: the left one (closer to the transformer) is noticeably hotter than the right one. Sure, the left one gets some heat from the transformer, but the difference seems rather large to me.
At 12.6 A, the temperatures were as follows: transformer, 105 °C—rather high, though it may be a little lower once potted; transistor, 70 °C, leaving some headroom for operation at 115 V, but the low input capacitance is a problem; output diode assemblies, 107 and 79 °C. The left one is right at the limit, while the right one has plenty of headroom.




And now the fifth power supply, 200 W.




This model uses a different topology, which is the right and logical choice: using a flyback at power levels above 150 W no longer makes sense or offers any practical benefit. You can tell just by looking that this uses a half-bridge converter, which is excellent for this power level. The alternative is a single-switch forward converter, but those are much less common.

I should note that, as with the 150 W model, the power components are laid out properly: the transistors are along one side and the diodes along the other.



1. The input has a proper filter with a thermistor and varistor. The input capacitor is 150 µF; it could have been larger, but given that this power supply is designed for a narrow input-voltage range, it’s fine.
2. The transistors are rated for 16 A and 500 V.
3. The PWM controller is different, of course: NCP1252.
4. Two 30 A, 100 V diode assemblies. They aren’t connected in parallel here; instead, they operate in what are effectively the upper and lower legs.
5. The output has a pair of 2200 µF, 16 V capacitors. I have no complaints about them.
6. As expected, there’s also an output storage inductor, which is a feature of this topology. A litz-wire winding would have been a better choice for the transformer, but that’s not essential.



No complaints about the circuit board: it has the required clearances and slots.


The input capacitor actually measures about 133 µF, which is within the margin of error, so that’s fine. The output is excellent.


Here I had to enlarge the heatsinks and also position the transistors and diodes at odd angles. Otherwise, I’d have had to drill new holes in the heatsinks, and frankly, I couldn’t be bothered.


Let’s connect the load using the same four-wire setup, since I really can’t be bothered to poke around with a multimeter too :)


Efficiency was measured over a range of 1–21 A and came to about 87%, which is fine. At 21.5 A, the overload protection started to trip. It’s a latching type here, rather than one that reduces the output voltage. Voltage regulation is excellent: across the full power range, from 12 to 250 W, the voltage dropped by about 0.02 V.

In general, I think the protection current is set too high; I’d recommend lowering it to about 18 A.


Ripple with no load and at currents of 6, 12, and 18 A.
No complaints, though that’s largely thanks to the topology used in this model. In any case, around 100 mV at over 200 W is good.



I also had no particular concerns about the low-frequency ripple, although some small 100 Hz ripple started to appear at 18 A. It’s not critical.


The thermal test was run at 6, 12, and 17 A, covering the full power range and even going a couple of percent beyond it.

1. No load
2. 6 A, power supply cold
3. 12 A, warm
4. 17 A, warmed up
5. 17 A, hot
6. No load, hot.

As you can see, the maximum difference was just 20 mV—an excellent result.


The temperatures are really impressive; I couldn’t help admiring them :)
At 8 A, the transformer is at 59 °C, the hotter of the two transistors at 51 °C, the diode assemblies at 68 and 85 °C (the uneven temperatures are a feature of this topology), and the inductor at 62 °C.
At 17 A, the temperatures went up, but are still good: transformer 70 °C, excellent; transistor 70 °C, excellent; diode assemblies 94 and 120 °C, with the latter clearly too hot; inductor 83 °C, excellent.

My only real concern is one of the diode assemblies. Potting it with compound might help, or it might not. But the temperatures of all the other components—and the fact that it runs passively at 200 W—are really impressive.


Impressed by these results, I decided to check the capacitor temperatures too, and they were even better: 50 and 48 °C at 17 A, after the unit had warmed up. You have to admit, that’s a very good result. I think this power supply should last quite a long time.



So, let’s try to sum up everything discussed above in the form of a few conclusions.
1. The 36 W model: I liked almost everything. I have two concerns: the varistor is before the fuse, and the ripple is a bit too high, though it’s perfectly adequate for undemanding loads.
2. The 60 W model: my main concern is the protection current being set too high, although the power supply works fine even at 120% load. Ripple, temperature, and safety are all fine.
3. The 100 W model: I have no real complaints. It’s a very good example of a power supply with low ripple, properly tuned protection, and acceptable operating temperatures. It would be even better if the transistor and diode assembly were on opposite sides.
4. 150 W model — the transformer temperature is a little on the high side, but it should be fine in a potted enclosure. The protection is also calibrated very precisely, so it should be harder to burn out. Otherwise, it’s pretty good. Though it could really use a larger input capacitor, around 150 µF; that would be great.
5. 200 W model — I genuinely liked this one. It’s well designed, with low ripple and little heat. If I had to nitpick, it would be one of the output diode assemblies—or, more precisely, how hot it gets.

A few observations about all the models:
1. The circuit boards are designed quite well, with proper clearances and slots—in other words, good safety.
2. The capacitor capacitance is within the stated tolerance. Testing the 1000 µF 16 V capacitor also showed that they have low ESR, which is certainly good for their service life.
3. Of course, you can find minor flaws in all the models, but to be fair, they don’t spoil the generally good impression. Maybe if they were just smoothed out a little with sandpaper (rather than having to file them down), they’d be excellent.


That’s probably all for today. It was a big review with lots of tests, and I’m a bit worn out, but I really hope the information was useful. And thanks to Vitaly for providing the boards for testing.