48 V - 12.5 A Power Supply

Review / thoughts on the 48V-12.5A power supply.
I decided to write a short, user-focused review of the 48V-12A power supply.
There are three main sellers on AliExpress. If it’s in stock in Russia, delivery is fast. They’ve also started shipping from Ukraine. Seller VELIKA - CLICK.
The second seller, YiYuan - CLICK.
The third seller, Sun-Lite - CLICK.
I’d had this power supply bookmarked for a long time. When I saw it in a photo posted by "IGOR IVANOVICH" (in the comments on the 36V-10A power supply), I decided to buy one and see how good it was.
I don’t have a power supply with MOSFETs at home yet. The seller promises copper in the output inductor and transformer. When ordering (if it ships from China), specify the mains voltage: "AC85-145V" or "AC185-265V". Delivery to Russia is handled by "IML Express" and brought to your door by courier (10 business days from China). For once, I have no complaints about "IML Express". Usually they call and say they’re short-staffed and buried in parcels, so I should come pick mine up myself.
No complaints about the packaging: it was done right, with foam and "air pillows". I usually never switch on a power supply without inspecting it first. But I switched this one on without an inspection, through a 220V test lamp (a regular old Soviet 100W bulb). Everything was fine; the output was set to 48.5V (adjustment range: 43-55V). I got the photo of the power supply out of the box from "IGOR IVANOVICH". All the components and their ratings are exactly the same, except for the output capacitors.
"IGOR IVANOVICH" has the 36V/10A power supply, with 50V capacitors (1000uF/50V). Mine is the 48V/12.5A version, with 63V capacitors (1000uF/63V).
Photos -



Here are my photos right away






A quick note for the manufacturer. How can the 360W and 600W power supplies have exactly the same components? Is only the transformer different? And is that transformer actually different at all?
The component side of the board is coated with cleaning fluid. The solder side, however, is clean. In the end, I couldn’t get it completely clean; the stains had soaked into the board.
Starting with the components: the first one is the varistor. It was planned for the wrong place and wasn’t installed (it goes before the fuse). The fuse is a small 10A glass one. It would be better to replace it with a larger one. Two 5D15 thermistors are connected in parallel. That’s the wrong way to connect them (see KIRICH’s video). Next is the standard filter: two X capacitors (0.47uF and 0.22uF), two Y capacitors, and a common-mode choke. The common-mode choke is wound with aluminum, and the wire gauge is smaller than in KIRICH’s 36V-10A power supply. The GBU810 bridge rectifier has a small heatsink. Next is the standard 110V-220V selector/jumper.
Why they ask you to choose and emphasize the input voltage when ordering is a mystery. I desolder this switch right away (it’s very easy to remove on all these power supplies). Next are two high-voltage capacitors rated at 250V/820uF. The ones labeled Nippon Chemi-Con are fake; they measure 650uF and 670uF. They’re relabeled standard 250V/680uF capacitors. Power supplies using a TL494 (KA7500B) almost always have 680uF/250V capacitors (Dongbaohe), which are actually standard 470uF/250V capacitors. The capacitors are connected in series. They often have dents, so inspect them.
Now for my notes and the changes I made to the power supply’s "input section".
I have major complaints about whoever designed this board. Why did you squeeze the space at the input, only to leave a huge empty area after the bridge rectifier? Even with the 110V-220V switch, there’s more room than in a standard power supply. In the end, I still couldn’t find space for a varistor (14471). I replaced the small fuse with a large 8A one, using the existing hole for the varistor and the original fuse hole. I installed a 5D20 thermistor: remove the two old ones, take a 1.5mm drill bit and make one new hole. Use one of the old thermistor holes for the other lead. I wanted to install larger X capacitors, but there’s no room at all widthwise, so I gave up on that. I installed a KBU1010 bridge rectifier with a different heatsink. I checked the old bridge rectifier; it gets fairly hot, but nothing critical. The KBU bridge rectifier body is wider than the GBU, so I had to install it at an angle. I replaced the input capacitors with 1000uF/200V ones.
While I remember, a note about power supply cases, fans, and capacitor dimensions. The cases, holes, and so on are usually standard across these power supplies. In other words, you can take the case cover from one power supply and put it on another. That’s handy if, for example, you have a DPS module mounted on the cover using standoffs. You can swap power supplies quickly without modifying anything.
As for the fan, this one runs on 12V. There’s no speed control; it runs all the time. There’s an auxiliary winding on the output inductor (7.5 turns). Most importantly, the fans in these Chinese power supplies don’t have a rubber plug. Don’t turn the fan over, or the lubricant will run out.
If you do need to turn it over, fit a plug from a computer fan. Check the lubricant; this particular power supply had none.
As for the capacitors, regardless of the power supply’s voltage and power rating, the standard size that will fit at the output is 16mm/40mm. At the input, 30 mm/40 mm. A little over 40 mm tall.
On power supplies using the TL494, the Chinese save money on the thermal switch (they leave it out). The standard part is a KSD9700 rated at 45°C (check AliExpress). Note that the KSD9700 must be the normally open type (there are different versions). All the boards have a spot for a KSD9700. On high-power TL494 supplies, it’s usually behind the transformer, next to the fan connector and the 2 W resistor. The Chinese use a 26–30 Ω resistor here. By increasing its resistance, you can adjust when the fan switches on and how fast it runs. Buy a few different resistors up to 100 Ω and experiment until you get the result you want (without buying a KSD9700).
Back to our power supply.
The PWM controller is an NCP1216. It uses hard-to-find (not available here) fast MUR4100 diodes, a pair of small 2N2222 (NPN) transistors, and a PC817 optocoupler. The two IRFP460A transistors are rated for only 500 V. The current-sense resistors are two 2 W resistors in parallel: 0.22 Ω and 0.18 Ω.
In my power supply, the transistors are from different manufacturers and look used; one has a chip in the case. The transformer measures 46.5 mm × 42.5 mm × 15 mm. What it’s wound with (copper or aluminum) is an open question. Output voltage feedback is handled by a TL431. The output rectifier assemblies are two MUR2040PTs. I thought that was a bit light, so I installed MUR3040PTs (I’d had them lying around for ages; bought them almost by chance).
The output choke is a black ring, about 38 mm in diameter and 14 mm wide. It’s wound with 1.4–1.5 mm aluminum wire (made to look like copper), in two strands. The seller insists it’s copper (and says the same about the transformer).
Let’s assume I’m the only one who got a power supply with aluminum wire. I should send him some wire and ask him to try stripping the enamel and soldering it with ordinary flux. But aluminum wire is used in all the inexpensive power supplies from AliExpress. On TL494 supplies (where the choke uses two cores), you can tell right away from the weight of a choke rewound with copper.
All these power supplies with aluminum wire work fine. Pay attention to the number of strands on the choke: the higher the supply’s current, the more strands it should have. The rough Chinese standard for TL494 supplies is one strand per 10 A. So, on a 12 V, 30 A supply, the output choke will be wound with three strands of aluminum wire about 1.4–1.5 mm in diameter.
As for the fan on this power supply, it’s inaudible with no load. But apply more than 1 A and it immediately starts roaring. You can shorten the auxiliary winding on the choke. If your choke is wound with aluminum wire and you’re going to rewind it with copper, use 1 mm, 1.2 mm, or 1.4 mm wire. The core is relatively small, so copper is easy to wind. I experimented with the copper wire gauge, winding 1 mm wire in two strands; the fan voltage seemed to be 11.9 V. Then I wound it with 1.2 mm wire and suddenly got about 19 V. I didn’t understand why, but I connected the fan and the voltage went back to normal.
I didn’t like the fan speed and noise out of the box, so I took some 0.4 mm wire and started tuning it to my liking. I reduced the number of turns on the auxiliary winding. I settled on a version that gives about 5–6 V at the connector with the fan connected and no load, and about 10.6–11 V under load.
I tested the power supply under a load of up to 350 W (using KIRICH loads) for about 30–40 minutes. I couldn’t detect any critical temperatures by touch, so I didn’t even bother measuring them. The output capacitors were completely cool (thanks to the fan). The choke was fine. The new bridge rectifier runs cooler than the old one at the same load, both the bridge itself and its heatsink. The transformer was fine. The transistors and rectifier assemblies were fine. But this was only a partial load, nowhere near the claimed 600 W (even briefly).
Now for my thoughts and comments on the remaining components.
I don’t have an opinion on the transformer yet. As for the MUR2040PT rectifier assemblies, I think there aren’t enough of them. My complaint about the output choke is the claim that it’s copper (mainly because they promised copper). It wasn’t secured with sealant. On inspection, one of the four ends had what looked like a cold solder joint.
The solder joint had cracked or come loose, and the wire end moves around in the hole. The load resistor is 2 W, 2.2 kΩ. I replaced it with a 3.3 kΩ resistor.
I think the output capacitors can stay. They look like proper factory-made low-ESR parts. I installed my own (which I had on hand): two 1000 µF/100 V and two 2200 µF/63 V capacitors, arranged diagonally. Why 1000 µF/100 V? A little trick. Our local shop had brought in 1000 µF/63 V capacitors, and they were snapped up for these power supplies. But nobody looks at the higher-voltage ones. The 80 V and 100 V capacitors (16 mm) fit perfectly, with even a little clearance in height. I’d ordered them ages ago, and they cost almost the same as the 1000 µF/63 V ones. I put a 100 V film capacitor (104) between the capacitors. I also added a 250 V SMD ceramic capacitor in parallel. All the output jumpers are thick (that’s not always the case).
The power components have proper thermal pads, but they’ve been cut in half to save money.
One questionable detail: the RC snubbers across the rectifier assemblies use 200 Ω resistors and high-voltage ceramic capacitors (221, 1 kV). These power supplies usually use 10–20 Ω resistors here. KIRICH needs to settle this and clear things up. My guesses: 1) The Chinese got the resistor value wrong when copying a branded power supply. 2) That’s how it’s supposed to be; it’s fine. 3) Ignore these resistors: if the power supply works, don’t mess with it. On TL494 supplies, the Chinese have made mistakes with the load resistor, mixing up the red and orange bands.
Next, between negative and ground there’s an ordinary capacitor. It should be a Y capacitor (here, it’s connected through other Y capacitors). There’s also an ordinary capacitor on the power supply output.
It should be a Y capacitor, installed between the power supply case and the positive output (connected to the board’s mounting screw in the case).

Bottom line: the power supply works, but some modifications are recommended. Like all Chinese power supplies from AliExpress, it has questionable capacitors and power components.
I can just about believe that, at different times, the choke and transformer may have been wound with copper. Judging by photos taken at different times, the manufacturer uses whatever parts are available. Sometimes the PWM circuitry has film capacitors (red), sometimes ceramic ones, sometimes high-voltage ones (blue). The production line can’t be stopped. Maybe wind the inductor on the 36V-10A unit with three strands (just in case) — as “IGOR IVANOVICH” suggests.
You can take advantage of the huge amount of space after the diode bridge: it’s very easy to install an extra 100uF/400V capacitor to increase the total capacitance of the original capacitors. This is covered in KIRICH’s videos.
The manufacturer’s website is link
People sometimes ask how to tell from a seller’s photos which power supplies they’re selling: newer ones with MOSFETs, or older ones based on the TL494.
There used to be an easy way to tell: new power supplies had fiberglass circuit boards (blue-green), while the old ones had “cardboard” boards. But now they also make TL494 power supplies on fiberglass boards.
The main way to tell a MOSFET power supply from a TL494 one in photos is to look at the customer review photos (sort the reviews by date). You need a photo showing the output terminals. Look at the trimmer resistor. If you see a round blue capacitor next to it (or behind it), then it’s a “new” power supply with MOSFETs.
MOSFET power supplies come in two variants: one based on the LD7575 and this one based on the NCP1216. The difference is minor. For example, an LD7575 power supply might be marked T-500-48: 500 is the power rating, 48 is the voltage. There were problems with the LD7575 power supplies this summer. They fitted only 2 output capacitors instead of 4, and used a 470uF/250V input capacitor. They seem to be fine now. The load resistor is usually placed less conveniently, where it heats up the capacitors. In the NCP1216 power supplies, the load resistor is placed farther away from the capacitors.
Tip: don’t pay attention to the seller’s photos unless the seller is the manufacturer. For example, the seller of the power supply featured in KIRICH’s 36V-10A review shows a new MOSFET model in the listing photos, but ships a TL494 power supply. Check the customer review photos.
To wrap up, here are a few notes about some components and what you might want to do to the power supply.
TL494 power supply boards often have crooked, half-inserted green film capacitors and small orange ceramic disc capacitors near the IC. Why use different types? The orange ceramic discs are almost always rated 104. If you use film capacitors instead (the green ones), they’re much larger. The boards also often have crooked, high-power resistors (usually 3 × 150 Ω). I always replace the orange ceramic capacitors with red 5 mm film capacitors, and replace those 150 Ω resistors. It’s not about them being crooked; they’re almost always rejects. If you compare them side by side with identical Chinese resistors and capacitors from a shop, you can immediately spot the rejects by the lead thickness and overall finish.
The holes for these capacitors on the power supply boards are sized for “red film” capacitors (5 mm). Search for “CBB Polypropylene film capacitor”.
When you get the power supply, it’s best to switch it on for the first time through a safety lamp. Write down all the components in a notebook—you’ll start learning to recognize them. Then check that the power components are secure and apply thermal paste. Before testing, make sure the high-voltage capacitors have discharged (check with a multimeter). You can discharge them with the same safety lamp (carefully). Be sure to check the solder joints on the energy-storage inductor; it’s usually the first thing damaged in transit. Check the load resistor’s value—there have been mistakes.
If the power supply has a high current rating, it’s a good idea to tin its output with solder, preferably using copper wire. Keep in mind that some of the capacitors may be counterfeit. If you like, reflow any suspicious joints. I reflow every component. That way, you can’t blame the Chinese if something goes wrong—you’ll have to blame yourself. I always replace the small electrolytic capacitors in the PWM controller’s power supply and the fan’s power supply.
I also connect 0.1uF–0.47uF SMD ceramic capacitors in parallel with them. Carefully scrape the traces clean and solder on the SMD capacitors (usually SMD1206). Then coat them with PLASTIK-71(70) lacquer. You can use the same lacquer on the inductors. It’s also a good idea to lubricate the fan (check that it’s lubricated; you really should).
One more thing I remembered)) — in KIRICH’s reviews of TL494 power supplies, you can read that if there’s only 1 diode assembly, you can install a second one to improve efficiency and reliability. This mainly applies to high-power supplies with high output voltages (36V, 48V, etc.). Low-voltage models usually come with 2 assemblies already.
A common mistake: people install a second diode assembly, solder the indicated jumpers, switch on the power supply, and it doesn’t work. Be sure to remove the short factory jumper between the diode assemblies (it’s shown in the 36V-10A review). That jumper is only for a single assembly. Add the jumpers as shown in the 12V-30A review. Use identical diode assemblies (same manufacturer, preferably the same batch. That’s KIRICH’s advice).
A little digression.
If you’ve started reading KIRICH’s reviews and got interested, like I did back then, but don’t have a decent soldering iron and so on, here’s what experience has taught me: go straight for a soldering iron based on an STM32 board from AliExpress. Tell yourself (and your inner miser), your wife, or your girlfriend: enough with the “shaving foam” gifts for the holidays—let’s buy a soldering iron. With a coupon on 11.11, you could get the board, soldering iron handle, wires, and connectors (without the power supply, case, or tip) on AliExpress for 900 RUB (about $13). Buying a Bakon 950 or another soldering iron will just be an interim step, and you’ll still have spent the money. For solder, if you can’t get our local solder, go with the tried-and-tested MECHANIC HX-T100. 0.8 mm is a good all-purpose diameter; 0.4–0.6 mm is for small jobs.
In Russia, you can find our local solder and MECHANIC in Profi and green-spark, even in smaller towns. If you can’t find it anywhere, go to DNS/Technopoint and buy ZUBR solder (it gets good reviews).
Good luck, everyone. I’ve probably forgotten something as usual, and maybe made a mistake somewhere.

P.S. While I was writing this “essay,” I got a message from the seller-manufacturer. He asked me to send photos and specify in more detail which capacitors, transistors, and inductors I’d told him about.
I’ve already sent him the modified power supply. The verdict: he agreed about the transistors and capacitors (but said nothing), but got “offended” about the copper/aluminum. He marked the storage inductor, transformer, and common-mode choke with arrows, and pointed out (very insistently) that they use copper.
At the end, he added: "We saw that you changed the fuse. You did the right thing (professionally). We approve of your decision."
Something like that.