60 V 300 W Power Supply, Version Two

As I mentioned recently, I needed to build a powerful charger for one of my custom electronic loads. I’ve already shown you the converter, so now it’s time for the power supply. I’ve reviewed a similar SMPS before, which makes this all the more interesting—we’ll compare them.

A reminder: to keep up with new posts on the blog, I recommend following me on Instagram, where I’ll post notifications about every new article and maybe 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



About three years ago, I reviewed an almost identical power supply, but that time one of my friends sent it to me. What started as a straightforward review turned into a modification—or, more accurately, getting the PSU into proper working order. But I’m getting ahead of myself; let’s take things in order.

This is actually the fifth similar power supply I’ve had my hands on. I just tested the first one, tested and modified the second, repaired and modified the third, and I modified the fourth but didn’t write about it. And now here’s the fifth.

I once built an electronic load with a built-in charger, but that version was limited to 30 V and 6 A. Around the same time, I thought, “That’s not going to be enough,” and bought a 20 A voltage converter along with a 300 W power supply instead of a 180 W one.
I didn’t have time, so I put it all on a shelf. Now I’m finally getting around to assembling it, which usually means I first need to check the components and sometimes bring them up to scratch.

I’ve already sorted out the converter—well, almost—and only just got to the power supply.
When I ordered it, the price was noticeably lower. What’s more, 60 V power supplies were in stock back then. They’re hard to find now: 24, 36 and 48 V models are available, but I can’t find a 60 V one. I have a feeling I’ll have to switch to Mean Well. They’re better and more powerful, but the prices are at least five times higher than the usual “Chinese” ones. I’m not sure it’s worth it, although there’s always the option of building a 600 W battery tester with a built-in charger in the same enclosure.
It’s a really strange situation: 60 V SMPS units are hard to find, and branded ones are almost impossible to find. The highest I can see is 54 V...



The power supply is an open-frame unit, exactly what I need. You’ve probably seen it plenty of times, though this one has some noticeable differences.


A closer look: the assembly is reasonably neat, though there are a few issues. The heatsink is bent here, a capacitor is dented there, and the transformer core is loose. In short, it’s a typical inexpensive Chinese SMPS.


On the input side, there’s a mains filter, a thermistor and a 220 µF 450 V input capacitor.


The terminal block is the same as before, but the fuse is in a plastic holder and rated for just 2 A, which seems odd, to say the least, for a PSU of this power.
The key difference—and something I rarely see in Chinese PSUs—is the pair of thermal switches mounted on the heatsinks. Both are rated for 85 °C, and it looks like some thermal compound has even been applied underneath them. The first disconnects the mains input; the second most likely cuts power to the PWM controller.


There are two high-voltage transistors connected in parallel. At first, the different shunt values puzzled me, but then I figured it out.
Initially, I thought each transistor had its own shunts. It turned out that wasn’t the case: the transistors are connected in parallel (except for the gates, which are isolated by resistors), and the shunts are all connected in parallel too. However, one pair has a longer PCB trace, so those shunts have lower resistance to compensate for the extra trace length.
The resistance seems a bit low. First, 0.17 Ω is an odd value; it’s not from the standard E24 series. Second, if we assume there are four resistors in parallel, the equivalent resistance is roughly 0.04–0.045 Ω. The previous version used 0.39 Ω shunts, also four of them, giving an equivalent of about 0.1 Ω. This one is twice as low, and in the primary circuit, too. A bit odd; essentially, this is short-circuit protection, not overload protection.




The control circuit is the same as in the previous versions, except for the first one, where it was on a separate board. In any case, just like before, the controller’s markings have been erased. Why? A mystery...


Since the power supply uses a single-ended forward topology, it has an energy-storage inductor.
There’s a Y capacitor nearby to reduce interference — more details
A pair of B20200G diodes are mounted on the heatsink.
There are three 470 µF, 63 V capacitors on the output; the one on the far left is noticeably deformed.


As before, I have no complaints about the board: the clearances are fine, there are protective slots, and it’s relatively clean.


And here’s the first “quirk”: for some reason, the manufacturer put a 1 kΩ resistor across the output. At first I thought, “No, that can’t be right—I must be seeing things,” but I checked, and it really is connected in parallel with the output.
To understand what’s wrong here: with a 1 kΩ resistor and 60 V at the output, the current is 60 mA, so the resistor dissipates 3.6 W. Not only is that an absurd amount of power, but the resistor is rated for just 1 W. Complete madness. Even if you switch the power supply on briefly, the resistor heats up to nearly 200 degrees, and if you leave it like that for longer, it could easily desolder itself.


The no-load power consumption is about 4.5 W, which, in this test, actually made me happy.


Let me explain why 4.5 W is good for this power supply. As I wrote, I got one of these power supplies three years ago; it looked exactly the same.


But that one “consumed” about 11 W with no load. In this case, knowing that the output resistor dissipates 3.6 W and the total is 4.5 W, that’s actually pretty good and easy to fix.


Last time, I had to replace the output diode assembly with a faster one. The original was too slow to turn off, which caused severe overheating. As you can imagine, replacing a resistor is easier and cheaper.


I decided to use a through-hole resistor; I simply don’t have an SMD one with the right value at home. While I was at it, I also decided to replace the capacitors.
Ideally, it would be good to use capacitors with the same capacitance but rated for 80 V—or at least 100 V. The problem is that 80 V ones are impossible to find, while 100 V capacitors are usually much larger. I needed capacitors no more than 13–14 mm in diameter.

First I bought three Samwha RD 680 µF, 63 V capacitors, but in the end I had to buy different ones: Panasonic low-impedance 470 µF, 63 V. If I can’t find anything with a higher voltage rating, then 63 V will have to do—but at least these are good quality. As usual, I bought them at the Kharkiv Cosmodrome.

Just for comparison, here are the capacitor dimensions. You can clearly see that the Panasonic ones are larger, despite having the same voltage rating and capacitance as the originals.


Testing showed that, first, all three original capacitors are working; second, the Samwha RD capacitors perform worse, although they may be more reliable; and third, since I wanted to improve things, the Panasonic capacitors turned out to be just the ticket.


I replaced the output resistor with a 3.3 kΩ one, reducing the power consumption from 4.4 W to 1.7 W.


But that’s not all. While I was at it, I changed one of the divider resistors from 750 Ω to 1.5 kΩ to lower the voltage from 60 V to 58 V. I decided to lower the voltage, first, to improve reliability, and second, because I can’t get the battery to charge at 48 V anyway, so there’s simply no point in having 60 V.
I also replaced the LED resistor: it was 20 kΩ, and I installed a 30 kΩ one. I could have used a higher value, but I didn’t have one handy. I soldered the original resistors I removed nearby so they wouldn’t get lost, in case I had to put things back the way they were.


As a result, no-load power consumption dropped to 1.4 W, and the voltage to 57.8 V. I should say a few words about the wattmeter. When mine burned out, I had to buy a new one, but unfortunately, although it looks similar, it’s much worse than the old one. The screen is dimmer, it measures slowly, and its resolution is poor: for example, it jumps straight from 1.4 to 1.7 W, with no values in between :(


A couple of words about the feedback circuit. It’s similar to the circuit in the first power supply I modified, which was also a 60 V unit, but there’s one difference: the optocoupler’s supply resistor here is 10 kΩ, not 2.4 kΩ.
In the power supply under review, I replaced one of the 750 Ω resistors with a 1.5 kΩ one.


Last time, I replaced the 750 Ω resistors with 1.5 kΩ ones. This raised the voltage and noticeably reduced its temperature drift.


But that’s not all. I hadn’t noticed before that the TL431 zener diode is connected directly to the optocoupler, even though it’s rated for a maximum of 35–40 V. So I connected an 18 V zener diode in series with the optocoupler. Ideally, I’d have used something like 24–27 V, but this is what I had, and it’s better than leaving it as it was.
The right thing to do would have been to reduce the resistor value a little, say to 5.1 kΩ, since the voltage has risen slightly this way. But in the end, I got exactly what I wanted.


And of course, a few tests. First, efficiency measurements and a load test.


I tested load currents from 0.5 to 5.5 A in 0.5 A increments.


The result was an efficiency of around 85–87%. I somehow expected a little more, at least around 90%, but it is what it is.
I’d planned from the outset for about 60 W of losses in the charging section at maximum power, and that’s pretty much what I’m getting: around 50 W lost in the power supply and another 10 W or so in the DC-DC converter (at a 12 V, 20 A output), for 60 W total. That’s the amount of heat I can “blow” out of the enclosure without resorting to any tricks.


I wouldn’t say I ran this power supply for very long, but it worked for about 10 minutes, without active cooling and at 100–110% of its rated power.
The temperatures seemed perfectly reasonable to me. The hottest components were the input thermistor and the output inductor: the thermistor was around 150°C, and the inductor about 100°C. By the way, for some reason the thermal camera started glitching, showing both -40 and +140 at the same spot. Maybe I just forgot to switch the range when reinstalling the software after an unsuccessful automatic update (which is why I like standalone thermal cameras).


Surprisingly, the transistors and diodes were all fine. Even in this mode, they only reached about 85°C. The snubber resistors were the only parts that got noticeably warm, but that’s what they’re supposed to do.


And of course, let’s look at the ripple. The oscilloscope probe was connected through a parallel C+C filter.


First, the high-frequency ripple, with no load and at 1.75, 3.5, and 5.25 A. I accidentally set the scope to 2 V per division for the first screenshot, but that doesn’t really matter; I’ll explain why later. Anyway, around 150 mV of ripple at maximum power and a 60 V output is genuinely great. Excellent, I’m happy with that. But the low ripple is mostly down to the topology itself. If this were a flyback, the ripple would be much higher, and I definitely wouldn’t risk using 63 V capacitors. That’s also why Mean Well uses 16 V capacitors for 12–13 V outputs. This topology also lets you reduce the output capacitance; the trade-off is an additional high-power inductor.


The low-frequency ripple makes it clear that the power supply runs in burst mode with no load. That’s not very good, but it’s the “achievement” of the so-called “greens” :(
In the other operating modes, ripple is up to 350 mV, and it increases specifically at maximum power; it’s noticeably lower in the other modes. Ideally, I’d increase the input capacitor’s capacitance, but since this is just a charger, it won’t be running in this mode around the clock, so it’s not critical. Besides, there’ll be a DC-DC converter after it.



Speaking of the increase in low-frequency ripple: the left image shows the amplitude at 4.3 A, and the right at 4.7 A. You can see that it starts increasing at around 4.5–4.7 A, i.e. close to the maximum.


Someone might say that the power supply runs in burst mode with no load because I increased the value of the output resistor.
No, the resistor has nothing to do with it. The left screenshot is at a load current of 350 mA, and the right at 370 mA. In other words, the load has to exceed 370 mA for the power supply to switch to normal operation. There is hysteresis: the “green” mode switches off at 370 mA and switches on again when the current drops to 340 mA.
At 58 V, 370 mA is around 20–25 W, which is far too much to dissipate in a ballast resistor, so I’ll just have to put up with it :(


List of changes made:
1. Replaced the T2A fuse with a T3.15A fuse. In my opinion, 2 A is cutting it a bit too close.
2. Added an 18 V zener diode in series with the optocoupler.
3. Reduced the output voltage from 60 to 58 V.
4. Replaced the 1 kΩ output ballast resistor with a 3.3 kΩ one.
5. Replaced the output capacitors. The capacitance and voltage rating are the same as before, but these are Panasonic.

This is the kit for my future battery tester charger: a 300 W power supply and a DC-DC converter rated for up to 20 A.



Conclusions.
As is usually the case with Chinese devices, this one has its share of flaws. They’re minor, but they’re all covered in the review, so I won’t repeat myself.
There are some good points, too. First, this power supply is clearly better than the previous ones. The output diode assemblies are faster now and don’t heat up under no load, and thermal switches have been added for overheat protection. Overall, the unit works pretty well. I’m especially pleased that the high-frequency ripple is low—though that’s after replacing the capacitors. The original ones had decent ESR, too, but their reliability is another question.

All in all, it’s a decent power supply. I can’t say how it is with other sellers, but it’s hard to tell, since I’ve already seen at least three versions of this PSU. Also, the output diode problem only affected the 60 V version, while the highest rating I see for sale is 48 V. On top of that, it’s hard to know which PSU the seller actually has, because even the store photos here show the old version without thermal switches.


That’s all for now. I hope you found the review useful. As usual, comments and links are very welcome.
And just to repeat: the link goes to the store where I bought it, but the price there is crazy right now. I do regret not buying at least five of these power supplies. Mean Well units are great, of course, but expensive :(