Two More Classic 250 W and 360 W, 12 V Power Supplies

Another installment in the power supply review series, and once again it’s a classic: switching power supplies built around the seemingly eternal TL494 and its equivalents.

As a reminder, 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


This review will be a little shorter than the previous one, first because there are two power supplies here, not three, and second because these units aren’t all that different from the lower-power models. Still, I’ll run the same tests as before—what would a review be without them?

So, here are two power supplies, also packed in cardboard boxes, though you can already see the difference: one has passive cooling, the other active cooling.


As you’ve probably guessed by now, the passively cooled one is rated for 250 W, while the actively cooled one is rated for 360 W.
The 250 W model measures 20 × 110 × 49 mm, slightly larger than the previous models.
The 360 W model is even larger, at 215 × 114 × 49 mm.
By the way, I’ve never really thought about this before: are there any standards for the dimensions of enclosures for power supplies like these? Something like a power-to-size ratio, or similar?

The mounting hole positions are different too, of course, which makes sense since the enclosures are clearly different.
The labels on both units list the basic specifications. For some reason, though, the 250 W model is rated at 20 A, which works out to 240 W; 250 W would be closer to 21 A.


The terminals, adjustment, and indicators are all more or less similar. The outputs are brought out to two groups of three terminals, and the terminal blocks have covers with latches.


A few observations before the review. The photo shows that while the layout of the power supplies is similar, there are some noticeable differences; at the very least, one is a mirror image of the other. The fanless model has the same layout as the previous units, while the actively cooled version has its components arranged in mirror image.

To be fair, I’m not a big fan of active cooling, but unfortunately, once you get above a certain power level, a fan is usually fitted. Of course, you could make a much higher-power unit with passive cooling, but that would drive up the price significantly—either through a pile of aluminum or through more expensive components and circuit design.
I’ll go further: I’ve more often seen power supplies based on the same TL494 with passive cooling rated at up to 200 W, but this one is 250 W. Let’s see how it handles that.


This also got me thinking: I seem to remember seeing passively cooled power supplies with the same “mirror-image” layout. I’m talking about the “classic” designs, of course, because I remember that the layouts in the Mean Well LRS 200 and 350 are the same, even though the first is passively cooled and the second has a fan.



Let’s start, of course, with the lower-power model, PTA-250W-12V - link.
By the way, the product page gives the correct current rating: 20.8 A, which corresponds to a power rating of 250 W.

This model has passive cooling, and personally, I think that for this circuit design, construction, and component selection, this is about as far as you can go without active cooling or making the unit more expensive.


1. There’s an input filter, but this unit has two thermistors, unlike the previous models.
2. Capacitors from an unknown manufacturer with the rather wild name Borunda, rated at 470 µF
3. The layout and circuit design are typical of power supplies built around the TL494, KA7500, and similar controllers.
4. The transistors are the familiar 13009s as well.


The manufacturer made more changes in the output section. First, a heatsink has been added for the output diode assembly—a fairly standard solution that can genuinely reduce the thermal load on the output rectifier. Second, unlike the 120 W model, the output inductor here is wound on two cores instead of one.


1. The output diode assembly is rated for 30 A and 100 V, which seems marginal to me; I’d have used two in parallel.
2. There are now two current-sense shunts, which makes sense given the current is over 20 A. But the inductor is still wound with just three wires, which seems inadequate to me, especially since there’s clearly room for more wire.
3. The output capacitors are three 3300 µF, 16 V units. No particular issues here.
4. Y capacitors are used in critical locations, which is fine too.



The actual capacitance of the input capacitors is 320–330 µF, compared with the stated 470 µF, i.e. In fact, they’ve cheated us by one step if you go by the E6 series (100, 150, 220, 330, 470 and 680).
The output capacitors are fine; I have no complaints about their capacitance.




The board is extremely easy to remove: release the transistor and diode assembly clips, undo one screw, and that’s it.
This photo was taken after the tests so as not to affect the test conditions.



Overall, I have no complaints about the assembly or PCB layout.


Slots have been cut into the board at the narrowest points, so everything looks fine here.


The full adjustment range is 10.66–12.54 V. Usually, the upward adjustment range is wider; here it barely reaches 12 V, which feels unusual. Still, it should be fine for minor compensation when powering a load through relatively long wires.


The efficiency of this power supply is also pretty ordinary, at around 82%. You get what you pay for: if you want higher efficiency, be prepared to pay more :(
I tested it from 1 to 23 A; at 24 A, the power supply went into overload protection.


No complaints at all about the high-frequency ripple. These are the results with no load and at currents of 7, 14 and 21 A. Even at maximum load, the peak-to-peak ripple is about 50 mV, which is an excellent result.


The 100 Hz low-frequency ripple is, of course, worse, but that was predictable: the filter capacitance is 160 µF, and we’re drawing 250 W.
As they say, it’s not exactly terrible, but it could be better if they’d installed capacitors with a real capacitance of 470 µF, rather than just claiming that value.


And of course, here’s where quite a lot of Chinese power supplies fall short: load testing.
As usual, we start by recording the voltage with no load. I set it to 12.028 V here.



1. We load it to about 50% of its rated power; the voltage drops to 11.974 V.
2. Let it run like this for about 20–30 minutes, sometimes 40. I usually just wait until the power supply reaches a point where its temperature is barely rising. Then I increase the load current to 100%. The photo shows 20 A; a little later I raised it to 21 A. The voltage dropped a little further—that’s how voltage depends on load current.
3. Wait another 30–40 minutes and check again. The voltage has dropped slightly; that’s how voltage depends on the feedback divider heating up.
4. Remove the load. The voltage rises to 11.991 V. This reflects both the voltage’s dependence on load current and its dependence on temperature, depending on which one you’re measuring against.

So, the voltage drops by about 50–70 mV for every 10 A of load current, and by about 40 mV between the hot and cold states due to temperature. In my opinion, that’s an excellent result, especially considering that there’s a trimmer resistor in the feedback circuit, and its resistance usually varies more with temperature than that of fixed resistors.



Now, what about component temperatures?
First, the results at 50% load.
1. Transformer winding: 57.7 °C, excellent
2. Transistors: 48.6 °C, excellent
3. Output diode assembly: 73.9 °C. That seems rather high for a 50% load. Assuming the room was at 25 °C, that’s a temperature rise of about 50 °C. At full power, we’d expect another 50 °C or so, bringing it to at least 125 °C.
4. Output inductor: 71 °C. I’d have liked that to be lower, too.



Full load: 250 W.
The overall picture is, of course, worse here.
1. Transformer winding: 107 °C. Basically fine; I’ll explain why later.
2. Transistors: 76.2 °C, excellent, with plenty of headroom.
3. Output diode assembly: the thermal camera is over its range, showing more than 136 °C. I later found out it was around 140 °C—actually higher than the expected 125 °C, so it’s clearly overheating.
4. Output inductor: 123 °C, too hot.

Based on the test results, I can say that without forced airflow, this power supply won’t be able to deliver 250 W continuously. If the ambient temperature rises, things will get even worse. In my opinion, this is a very good power supply rated for 200 W continuous output, but it can deliver 250 W only for relatively short periods.




The second power supply, PTA-360W-12V, link.
This is a 12 V, 360 W, 30 A model.

This one uses active cooling. To be fair, the fan is fairly quiet, but it has no temperature control and runs constantly, which is disappointing.



The layout is typical; there isn't much new you could come up with here.


It has an input filter and Y capacitors. There are two thermistors, connected in parallel.
The input capacitors are still Borunda, but this time they're rated at 680 µF.


The PWM controller section is the same classic design I mentioned earlier: KA7500, also known as TL494, DBL494, and so on.


1. The transistors here are D209L, essentially the same as 13009 but made by a different manufacturer. The only thing that really sets this model apart from the previous ones is the use of a heat spreader.
2. Unlike the previous power supply, whose output inductor is wound on a powdered-iron core, this one uses sendust, as do the 150 W and 200 W models, so this solution should be more efficient. But once again, it's wound with just three strands, which is a bit light for 30 A.
3. The diode assemblies are rated for 30 A and 100 V, just like the previous model, but here there are two in parallel.
4. The output capacitors are 3300 µF, 16 V, just like the previous model.



There are three current-sense shunts, compared with two in the previous model, which had one and a half times less current. Of course, you could use two or four shunts for the same current, but all else being equal, you can think of this as one shunt per 10 A.
The fan connector is next to them, with a spot for soldering a temperature sensor nearby. But the circuit is primitive: a thermistor in series with the fan. The thermistor was probably meant to sit against the output inductor.


The transformers in both power supplies are very similar, but the 360 W model has a slightly larger one. The core dimensions are 44 × 40 × 13 mm for the 250 W model (left) and 46 × 40 × 13 mm for the 360 W model. In practice, they're almost the same power, but active cooling saves the more powerful version from overheating.


1. The input capacitors have a total capacitance of 213 × 2 = 426 µF, though 680 µF is claimed. Once again, the stated capacitance is rounded up by one step in the E6 series: 426 is close to 470 µF, and the next value in the series is 680 (100, 150, 220, 330, 470, 680). Honestly, I don't understand why. Just state the actual capacitance—it's easy to check, and this immediately knocks the manufacturer's reputation down a notch.
2. The output capacitance is actually higher than claimed: 3 × 3300 = 9900 µF. No complaints there.


We take the board out of the case. This one is held in place by five screws, plus the component clamps. Note that
At first I thought the heat spreaders were riveted to the case, but they're not; they just have pressed-in pins to keep them from shifting.
This photo was taken after testing, as with the previous model. To avoid affecting the test conditions, I applied thermal paste between the power supply case and the plates after disassembly to improve heat transfer.



Overall, I have no complaints about the board: the clearances are good, the soldering is neat, and the board is clean. It's a pleasure to hold.


The purpose of the slot (marked in blue) is unclear. The clearance is already much larger there, while farther to the right there's a fairly long section where the primary- and secondary-side traces run much closer together (marked in red). I'd recommend increasing the distance between those traces.
The clearance under the transformer may be there in case dust gets in from above. The power supply has active cooling, after all, so dust could get under the transformer. In that case, I'd extend the slot upward.




The output voltage adjustment range isn't very wide, roughly 11 to 12.8 V.



Efficiency was measured over the 1–36 A range. At 37 A, the power supply went into protection. As a result, I got about 83–84%, slightly higher than with the previous model. I’d guess this is partly due to the lower winding resistance of the output inductor.



I measured the high-frequency ripple with no load and at currents of 10, 20, and 30 A. It was higher than on the 250 W model.
At maximum power, the peak-to-peak ripple was about 100 mV, which is a perfectly normal result, especially considering the power supply’s output power. In fact, ripple of this magnitude is found even in small 20–30 W power supplies, and that’s not even counting cheap Chinese “chargers.”


There’s also some 100 Hz low-frequency ripple, but I wouldn’t call it excessive; it’s within the normal range given the capacitance of the input capacitors. Increasing the capacitance would reduce the ripple roughly in proportion, and for most loads it isn’t particularly critical anyway.

Someone asked me the other day what high- and low-frequency ripple means.
High-frequency ripple is ripple in the output voltage at the inverter’s switching frequency. It “pollutes” the airwaves, heats up your capacitors, and so on. But there’s an important caveat: it’s relatively easy to suppress, even with the wires from the power supply to the load, especially if there’s a capacitor after those wires. Add an LC filter and you can reduce it almost to zero.
Low-frequency ripple consists of variations at the mains frequency. After the mains filter capacitor, the voltage changes in sync with twice the mains frequency. You can suppress this ripple either by increasing the filter capacitor’s capacitance or by using an inverter that can compensate for input voltage changes at that frequency. To some extent, you can reduce it by increasing the output filter capacitance, but that’s more complicated. It usually affects equipment that’s sensitive to power quality, such as audio amplifiers, where it shows up as 100 Hz hum.

So, in the end, the power supply output has switching-frequency ripple modulated at 100 Hz. Connect a long wire to the output and you’ve got a transmitter operating at 50–100 kHz (longwave), broadcasting a 100 Hz signal with amplitude modulation.



I tested this power supply not only at 50% and 100% load, but decided to expand the test a little and check it at 110% load as well (33 A).
1. With no load, the output was set to 12.029 V.
2. At 15 A, it was 12.024 V—excellent.
3. With the power supply warm and the current at 30 A, the voltage dropped to 11.993 V, also a very good result.
4. With the power supply hot and the current at 33 A, the voltage dropped to 11.951 V, which is also more than acceptable.
5. After it warmed up, the voltage at 33 A was 11.932 V.
6. After removing the load, I got 11.945 V.

In my opinion, these are perfectly normal results. The voltage depends somewhat on temperature and only very slightly on load current. I have no complaints.


So, how about the temperatures?
First, the results at 50% load.
Transformer winding: 48 degrees, excellent.
Transistors: 39.7, superb.
Diode assembly: about 60, normal.
Inductor: 40, excellent.



At 100% load:
Transformer winding: 108, a bit high, but tolerable.
Transistors: 72, excellent.
Diode assembly: 113, perfectly normal.
Inductor: 88, excellent.



At 110% load:
Transformer winding: 130 degrees, a bit high.
Transistors: 91, perfectly normal, with plenty of headroom.
Diode assembly: 127 degrees, tolerable given the 10% overload.
Inductor: 101 degrees, also normal as long as the power supply is running overloaded.

A quick note about the diode assembly temperature. In the thermal image, you can see areas right next to the diode assembly that are clearly hotter. I think the small jumpers and relatively narrow traces are getting very hot here. In other words, to some extent it’s not the diode assembly heating the board, but the board heating the diode assembly.

The temperature may also have been affected by the fact that the cover wasn’t installed in its proper place; it was basically just resting on top, which worsens cooling. I also noticed later that the terminal cover was raised and blocking the air intake at the end. I wouldn’t say this had a major effect, but either way, it could have raised the temperature by a few degrees on the output side of the power supply.





One more note, which I’d planned to include in the section on the 250 W model.
I specifically said the temperature was that of the transformer winding, not the transformer itself. They’re not the same thing. The winding is less sensitive to high temperatures. No, high temperatures aren’t good for it either: the insulation varnish can degrade, and copper resistance increases as it heats up. But it’s much more important to keep the core from overheating. The point is that ferrite (and other materials) stops acting as a magnetic core when its temperature rises above a certain value (the Curie point). At best, this will trigger the protection (if monitoring is done on the primary side); at worst, the power supply will fail (if current is monitored on the secondary side).

At 100% and 110% load, the magnetic core temperature in this power supply model reached 73 and 80 degrees, respectively, against a safe limit of 100. Of course, the inside of the core was clearly hotter, so both temperatures matter, but in my opinion the ferrite temperature is more important.
You could say there is even some headroom here, which is certainly good. But that headroom comes from the fan, which can get clogged with dust or stop working, so always make sure the cooling system is in good working order.




Conclusions.
I have no complaints about the build quality or neatness, especially considering the price. The soldering is tidy, the board is clean, and that goes for both models.
The ripple is also fine. I’d even say the high-frequency performance is excellent; the reduced-capacitance capacitors are what spoil the low-frequency results.
Speaking of capacitors, both power supplies use capacitors with a capacitance one step lower than the value stated on the case. If the capacitance matched the stated value, it would also be appropriate for the power supplies’ rated output. In other words, the markings look perfectly fine, but the actual capacitance is noticeably lower than specified.
As for power, in my opinion the 250 W model won’t be able to deliver that much continuously. I think its real continuous output is closer to 200–220 W. If you treat it as a 200 W unit, it’s excellent; for shorter periods (hours), it can also deliver 240–250 W, but I wouldn’t recommend exceeding 220 W.
The 360 W model does considerably better in terms of power, but that’s thanks to active cooling, not the electronics or components. Even so, I wouldn’t recommend running it above 330 W for extended periods. It’s always better not to run a power supply right at its limit, but to leave around 20% headroom.

That’s all for now. I hope you found this useful, and as usual, I’d be glad to hear your comments.
And thanks again to Vitaly for providing the power supplies for testing.