24 V 1.2 (2) A Power Supplies

This is the third and, for a while, the last review in the series about used power supplies for various DIY projects. Today I’ll talk about how a power supply rated for one power level can sometimes turn out, when tested, to be an “overclocked” version of a simpler PSU. I hope this will be interesting—or at least useful.
Tests, schematics, and the usual other stuff are after the jump.


In the first review, I mentioned that there would be three reviews of used power supplies in total. These are power supplies that were once used in some piece of equipment, but were later removed from their enclosures and sold separately as circuit boards. Surprisingly, these PSUs can turn out to be better quality than products from today’s small Chinese basement companies.
If you’re interested in power supplies with other output voltages or currents, let me know. I’ll try to get some and test them.

As with the previous two, I ordered a lot of three, both to get a larger sample and just in case one died during testing.


The seller lists this power supply as 24 V, 2 A, or 48 W. Those were the specs I needed. I planned to use it to power a TS100 soldering iron.


They look like perfectly ordinary power supplies: a transformer, capacitors, heatsinks, and other unnecessary bits.
Build quality gets a solid B: fairly neat and even, with the heatsinks firmly attached to the boards rather than hanging off the components. The only catch is that the heatsinks are steel, not aluminum—they did cut costs after all.


The power supply measures 93×49×25 mm.


Here’s a comparison photo to give you an idea of the relative sizes. 1–2: PSUs from the previous two reviews; 3: the one under review; 4: the well-known “people’s” PSU (links one and two). Below are various little bits and pieces; the camel is just standing guard.


The PSU from different angles.


1. The input EMI filter consists of a common-mode choke and a 0.22 µF Class X capacitor. I talked about what this filter does and how the different types compare last time.
2. The input has a 14 mm varistor rated for 560 V (peak), which seems a bit high to me. But what surprised me more was the presence of two fuses—not one on the live wire and one on neutral, but one before and one after the varistor. They also have different ratings: 5 A before the varistor and 2 A between the varistor and the PSU itself.
3. The bridge rectifier is installed near the heatsink, but doesn’t touch it.
4. The input capacitor is rated at 68 µF, which is enough for normal operation on “our” mains voltage at the stated 48 W output—but more on that later.


1. The high-voltage transistor, marked 4N80L, is mounted on a heatsink. The heatsink itself is well away from the secondary circuits, and the transistor has an insulated case with thermal paste applied.
2. There are even two interwinding EMI-suppression capacitors, connected in series, which is good for the safety of both the load and the user.
3. The output diode assembly is mounted on a separate, small heatsink. It’s marked MBR10150 and rated for 10 A and 150 V—more than enough headroom.
There are three output capacitors: two rated at 470 µF before the filter choke and one rated at 330 µF after it. There’s also a second output choke on the board, this time a common-mode choke, followed by another capacitor rated at 10 µF.
All the capacitors except the last one are rated for 35 V; the last one (10 µF) is rated for 50 V.
Overall, the PSU’s output stage is quite well designed, both in terms of the components used and the circuit design. All the capacitors are made by Taicon. It’s not Rubicon or Nichicon, of course, but it turns out to be a fairly large Chinese company too, and it even has its own website. All the capacitors meet their stated capacitance ratings.


Another photo of the PSU’s output stage. Note the protective Zener diode; the output common-mode choke is also easier to see here. I’ll definitely come back to the Zener diode later.


Two of the three PSUs were made in 2014 (on the left), and one in 2015 (on the right). The only difference is the marking on the transformer.


I had no particular complaints about the circuit board. The soldering is reasonably good, the board is clean, and it has protective slots. The only thing I can nitpick is the lack of slots under the Y capacitors.
The second thing I noticed during a closer inspection was that there’s no feedback optocoupler; regulation is based on the voltage on the PWM controller’s supply winding.


I also noticed a rather unusual design choice during the inspection. There are four jumpers under the input inductor.
At first I wondered why there were jumpers under the inductor, and why they were there at all when there’s already an inductor. But on closer inspection, I understood.
They form an improvised spark gap, essentially acting as a spark arrester. There are four jumpers, but they aren’t connected to each other, and the spacing between them creates a zone where an arc can form. For safety, slots have been cut into the board at this point.
The inductor windings are marked in blue, and the spark gap itself in green.


An iW1710 PWM controller is installed on the “hot” side. At the power supply output, there’s a 28 V TVS diode to supplement the Zener diode.


There’s also a label stating that the manufacturer of this power supply is Jentec Technology, along with its specifications: 21 V, 1.2 A.


Wait, what do you mean, 21 V and 1.2 A? The store page clearly says 24 V, 2 A.


It turns out the power supply has been “overclocked” in both voltage, from 21 to 24 V, and current, from 1.2 to 2 A. That nearly doubles its power, from 25 W to 48 W.
In the photo, you can see an extra resistor in parallel with the current-sense resistor on the left, and another in parallel with the feedback divider on the right.
I don’t want to blame the store here, because the chain of events was probably something like this: the manufacturer made a power supply rated at 21 V, 1.2 A, and did a decent job. Then someone on a forum probably realized it had plenty of headroom, “modified” it, and doubled its power. After that, some enterprising people took notice of the modification, modified a batch of power supplies in the same way, and sold them to the store. The store’s staff are unlikely to be experts on this, especially since the power supply works.
So we end up with a 48 W power supply for sale.


I’ve marked the added components in red on the schematic. Overall, the power supply’s circuit design is pretty good. It could perhaps use a thermistor for overtemperature protection, especially since the controller supports this function (pin 4).
But let’s get back to the Zener diode. I mentioned above that a 24 V Zener diode (ZD2) is installed at the output. Although it’s made by Philips, according to the marking, using this setup is still risky: when the output voltage equals the Zener voltage, sudden load changes can cause voltage spikes that may destroy it. I’d recommend removing it altogether and leaving only the TVS diode. It would be just right for a 21 V output.


The power supply is built around an iW1710 quasi-resonant controller made by iWatt, with an operating frequency of up to 130 kHz.


Controller block diagram.


The typical application circuit also uses the PWM controller’s supply winding to regulate the output voltage, so in this case we’re not seeing a cost-cutting measure by the manufacturer, but rather a design that follows the controller’s datasheet.


Now let’s move on to the tests. There’ll be quite a few, and they’ll cover different things.

First, I swept the load current from 0 to 2.5 A while simultaneously capturing output ripple waveforms.
You can see that the voltage remains very stable from 0.5 to 2.0 A, with a difference of just 30–35 mV.


The high-frequency ripple was also very low. Remember, we’re testing a power supply running at twice its rated load.


Here’s the test graph. You can see how the voltage sagged when I raised the current to 2.5 A.


The 100 Hz low-frequency ripple is somewhat worse, though not bad enough to call it excessive.


After that, I tested its load capacity. The voltage remains stable up to 2.3 A and only starts to sag at 2.4 A and above. When I tried to load the power supply with even more current or trigger a short circuit, it went into protection.
But since the current-sense resistor is rated at half its original value, I wouldn't be 100% confident in the power supply's reliability in this mode.


As an additional test, I checked for conducted interference on the input. I also included the circuit in my previous review, so here are just the photos.
Unfortunately, I haven't found a fluorescent lamp ballast yet, so I repeated the experiment using the same 60 W transformer as last time.


Again, there were no problems. The interference barely got through, and when it did, the maximum amplitude was about 0.35 V, or roughly 1.5% of the output voltage.


One of the main tests is a long-duration run under load. The procedure was the same as before: several steps at load currents of 0.5, 1.0, 1.5, and 2.0 A, with 20 minutes between tests.
I covered the power supply to simulate an enclosed space.


The test took 1 hour and 25 minutes in total, including the measurements between steps. The voltage stayed within normal limits the whole time.


Unfortunately, I don't have good news about the temperature. The high-voltage transistor is clearly overheating. It reads 116 degrees in the photo, but the actual temperature was higher: I took the photo with the "case" open, and the transistor had time to cool down a little.


While I was at it, I also measured efficiency. The photo shows the power supply's output power and its power draw from the mains. The last result isn't included in the summary table because it was measured briefly at a load current of 2.2 A.


Here's the summary table, which lets us draw a few conclusions.
First, the temperature: there is overheating. The inverter's output transistor reached almost 125 degrees, due to the power exceeding its design rating and the steel heatsink.
The other components, however, are at perfectly acceptable temperatures. Improve the transistor's cooling, and the power supply should be perfectly capable of operating at this load current, even though it's outside its specified operating conditions.
As it warms up, the output voltage rises slightly, effectively compensating for the drop under load. I have no complaints about it in this respect.
Efficiency is at a normal level and increases with load. This points to a well-designed power supply circuit.


Someone in the comments hinted that these power supplies aren't new, so what if they're radioactive? To answer that: they're fine; their radiation level is only slightly above background. For comparison, I measured my shoes, and they were about the same, maybe even a little lower.
My dosimeter is homemade, of course, and pretty old, but it's more than adequate for this test. The display shows 1; in reality, the counter registered noticeably more events, but there's a divider on the board. I don't remember the division ratio, maybe around 40. Each measurement takes 40 seconds. The circuit is from Radio magazine.



That's it for the tests. Time to sum things up.
First, let's separate fact from fiction. If I were testing a power supply rated according to the specs printed on the bottom of the circuit board—that is, 21 V, 1.2 A—it would have passed all the tests without any problems, and I think it would even have had some headroom.
But if we go by the specs listed in the store—24 V, 2 A—it would still pass the tests, but fail the thermal test.
Yes, I can say that the power supply itself is quite well made, especially considering its price. But I wouldn't run it continuously at its advertised rating of 24 V, 2 A, particularly in an enclosed case. Otherwise, it has very low ripple, good output voltage stability, and working protection. And all this despite the fact that it isn't running at the 25 W claimed by the manufacturer, but at 50 W—in an "overclocked" mode—which is pretty impressive for a Chinese product.

One downside is that it's relatively noisy, making a characteristic "buzzing" sound while running. You'll definitely hear it in a quiet room; I could hear it even with the electronic load's fan running. But as soon as I covered it, the sound disappeared.

That's all. As usual, I'm looking forward to your questions, suggestions, and comments. I hope you found the review useful. If you'd like to see reviews of other power supplies, let me know and I'll do my best to make them.