24V-20A Power Supply

I decided to write another simple, down-to-earth review of a 24 V, 20 A (480 W) power supply. The Chinese have used a slightly unusual layout.
I'll start with how I decided to buy this power supply. I saw it on AliExpress for $10. I didn't really need a power supply, so I figured I'd pass. I checked again a couple of days later—the price hadn't changed. So I decided to give the seller another chance to get it right.
In the end, after 2–3 days, the seller “got it right”: the price went up to $11.11. I waited a couple more days, then had to buy it. I couldn't pass up a price of $11.11 (including shipping)—I'd given the seller enough time to correct it. The power supply arrived relatively quickly via CDEK. It looks like our guys are pulling the wool over the Chinese sellers' eyes: the receipt listed the weight as 1 kg 450 g (which surprised me, since I know the standard weight). In reality, though, it weighs the usual 850 g, packaging included. The packaging was flimsy: just a padded mailer and the power supply in its own box. But it arrived undamaged. Some Moscow stores also list this same weight—1 kg 450 g (or 1 kg 300 g)—when calculating shipping.
The power supply itself is fairly standard and uses a KA7500B (TL494) chip. I didn't even turn it on; I opened it up right away.
Here's what we have: a fiberglass circuit board (good). There's a fuse and two 5D15 thermistors (connected in parallel, which is wrong, but it's a common Chinese shortcut and tolerable). A proper 14471 varistor is installed in the right place, after the fuse. There's an X capacitor (0.68 µF) and a 0.5 W, 680 kΩ resistor to discharge it. Next come a common-mode choke and two Y capacitors (222). So far, so good—standard stuff.
There's a spot marked for a second X capacitor, but it's not installed. There's room for a 0.22 µF one. The GBU808 bridge rectifier is standard, but this power supply is rated at almost 500 W. There's a spot for a heatsink for the bridge rectifier, and one should be installed. On these power supplies, including the ones with MOSFETs, the standard X-capacitor footprint is 22 mm between the holes. The same goes for the large red 400 V, 335J (3.3 µF) film capacitor.
So far, just a few minor flaws—tolerable. After the bridge rectifier, there's a 110 V/220 V selector switch. I desolder it right away (no need for any jumpers); it's not needed for 220 V. The hole for the switch is large, so I cover it from the inside with thermal tape to keep debris out of the power supply. Next, there are the famous Dongbaohe 250 V/680 µF capacitors, connected in series. Now they have gold lettering—but, as always, they're fakes, relabeled 470 µF/250 V capacitors.




What should you replace them with? In my opinion, good 820 µF/200 V (250 V) capacitors are ideal for these powerful TL494 (KA7500B) power supplies. The usual size is 30/35 mm. In fact, you can fit 30/40 mm capacitors in these power supplies. My favorite 150 Ω resistors (3 pcs.) and the high-voltage capacitor (blue, 103/1 kV) are installed reasonably well.
The transistors are standard D13009K. If they're blown, replace them with D13009K, MJE13009, D4515, D209L, or similar. Check the resistors. Kirich's review of the 36 V, 10 A power supply has an excellent repair guide.
The resistor values around the transistors are visible in the photo. R20 and R9 have slightly unusual values (using Kirich's schematic): 0.47 Ω. Usually they're 1 Ω, 1.5 Ω, or 2.2 Ω. If resistors R8 and R14 are damaged and you don't have 330 kΩ ones, you can use higher values, up to 510 kΩ. Replace R20 and R9, and R8 and R14, in pairs, using matching values (1 W resistors will fit). The larger diodes are FR157; the smaller ones are FR107. The two small transistors are now always C2655. The board has no cheap orange disc capacitors—all of them are film capacitors.
The usual set of small electrolytic capacitors: four 4.7 µF/50 V and one 1 µF/50 V. The PWM controller's supply capacitor is 47 µF/50 V, and the fan's is 100 µF/35 V. I replace them with name-brand 100 µF/50 V (for the PWM controller) and 220 µF/35 V capacitors. I also add SMD 1206 ceramic capacitors (0.1–0.47 µF). The board is marked 500 W. I'd give the soldering a 4 out of 5. Almost all the components are installed fairly neatly. Out of the box, the voltage adjustment range is about 20–27 V (I don't remember exactly; I tweaked it a little for my own use).


Now let's get to the drawbacks and what I didn't like.
There's no fan speed control: it always runs, with speed varying according to the load. Yet the board has footprints for all the components needed for speed control (using a small thermistor near the output choke). The Chinese keep increasing the power rating of these TL494 power supplies. The transformer hasn't grown by even 1 mm. The switching frequency is the same old standard. The capacitors are “old” too—in number, capacitance, and quality. There's a second chip and its associated components marked on the board. Once again, they're not installed, even though they made the holes. I looked around, and it seems they usually fit an LM358P.
On my particular unit, they forgot to trim the leads on one component, and left some very long leads in a couple of places. There are “solder balls” on the board that they didn't clean off. It's rated for 20 A, but they installed just one MBR30200PT output diode assembly—not enough. They also forgot to install the output jumpers, even though the holes are there. They need to be fitted because the current is high.
Now for the beloved output choke. I once asked Kirich, “What if I use just one core from the standard choke (the one with two cores) and wind copper wire around it?”
The approximate answer was: “It depends on the frequency and so on. You shouldn't just do that.” This choke is covered in the reviews of the 36 V, 10 A and 12 V, 30 A power supplies.
And now the Chinese have given us the answer. Apparently, you can simply take that one core from the old choke (the one with two cores), wind 1.1 mm aluminum wire around it (30 turns, two strands in total), and claim a power rating of almost 500 W. Easy as that—well done, guys!))).
The single-core choke first started appearing in power supplies with MOSFETs. Later, they began gradually using it in TL494 (KA7500B) power supplies too. But the thing is, originally this ring was Sendust. Sendust can withstand higher temperatures and barely loses its properties. Then the Chinese figured they had plenty of ordinary cores, so why use two cores for the inductor? They could use just one, but a basic one. Saves money, after all. And they could raise the power supply’s rating to 480 W. But they tested these power supplies: with the fan running at full speed, the power supply just about manages to deliver its rated power. I’m not going to blame any particular manufacturer for this inductor. Because of the second chip, they’ve painted themselves into a corner. The transformer has shifted closer to the terminals, leaving no room for a standard inductor. Nobody promised us copper. The transformer is probably wound with aluminum, too. I opened up a 48 V, 8.3 A power supply and found aluminum wire on the transformer. So, just for the challenge, I had to practice winding a transformer with copper wire (for a power supply using a TL494). They’ve started attaching the power supply board to the case with just three screws. If there are screw holes, we fix that. I use motherboard standoffs as spacers for the screws.
They’ve also started using mica pads on the power components quite often. You can buy them here, too (they’re cheap), and they’re available on AliExpress. They “breed well in captivity.” You can get 2–3 pads from a single sheet.




What to do with the power supply.
Leave everything before the bridge rectifier as it is. You can keep the original bridge rectifier, but it’s a good idea to fit a heatsink. I just liked the KBU1010 bridge rectifiers for these power supplies. If there’s room, add a second X capacitor. Because the KBU1010 is so bulky, I fitted a small X capacitor (15 mm), mostly for show. Remove the 110–220 V switch. If you’re going to use the power supply
under a heavy load, replace the original 250 V/680 µF capacitors with decent ones, or add a 100 µF/400 V capacitor after the bridge rectifier. Replace the output capacitors in any case. Fit name-brand 3300 µF/35 V capacitors (16 mm). I removed the KA7500B chip, fitted a socket (a turned-pin socket, which solders really well), and installed a TL494 from the shop. At this point, I’m just being extra cautious. I ran into a faulty KA7500B in a power supply from the manufacturer featured in Kirich’s review of the 36 V, 10 A model. The fault mostly showed up under load. The manufacturer’s label is shown in Kirich’s review. Simply replacing the KA7500B with a TL494 fixed the problem. Several power supplies in our crews had problematic KA7500B chips. His power supplies are probably among the best-selling on AliExpress (with shipping from Russia).
A note about the manufacturer featured in Kirich’s review of the 36 V, 10 A model. The seller on AliExpress is CLICK. Plenty of his power supplies are sold by other sellers, too. The 48 V, 5 A model is a good one. He’s caught the wave with 36 V and 48 V power supplies (8–10 A). He often uses the cheapest components. If he doesn’t have the right parts, he may just use whatever is in stock. The 48 V, 10 A model may have a 606 bridge rectifier fitted. He cleverly selected the voltage-sensing resistors (34–52 V).
Buy a 48 V, 10 A power supply and it outputs 48 V. Buy a 36 V, 10 A power supply and it outputs 36 V. The components haven’t been changed at all. He immediately started fitting 63 V capacitors in both the 36 V and 48 V power supplies. The output diode assembly is the same in both. An interesting question is what voltage the transformer was wound for.
Almost all the small capacitors near the chip are cheap orange ceramic disc capacitors, not film capacitors.
What he’s doing now: all the components are still the same, but he’s removed the KA7500B and is fitting a TL494. The main change is that he suddenly replaced the large red 400 V-335J (3.3 µF) film capacitor with a 225J (400 V) one—2.2 µF. It looks like he’s still using Rubicong.
The power supply will work, but the standard value is 400 V-335J (3.3 µF). Even Kirich probably couldn’t say exactly where the limit is—the point below which you shouldn’t reduce the capacitor’s capacitance. Kirich’s answer to what happens if this capacitor’s capacitance is insufficient or reduced is: “Sooner or later, the circuit will switch to current regulation mode.”
Back to our power supply. If you want fan control, there’s a small component marked as some kind of “transistor” near the fan connector. It isn’t installed; there’s a jumper in its place. Remove the jumper and fit a KSD9700 rated for 45°C. But since the jumper is quite far from the inductor (the KSD9700 wires won’t quite reach), first solder on
a couple of short wires, then solder the KSD9700 to them. Insulate the wire joints with heat-shrink tubing. Place the KSD9700 against the output choke and tape it to the transformer with thermal tape.
If possible, rewind the output choke with 1 mm copper wire. Wind at least two strands in parallel, preferably three. It had about 30 turns; the Chinese wind 24 to 30 turns (the exact number isn’t critical).
Then coat the choke with PLASTIK-71(70) varnish. You can do the common-mode choke as well. If the core is loose on the common-mode choke or small transformer, apply some of this varnish around the mounting points and let it dry.
What’s the best way to secure this choke? Neutral-cure sealant is best. If you don’t have any, use “liquid nails” (I’ve tried it several times). I’ve also tried clear Moment silicone sealant. Its temperature range is suitable (-40/+150). If you use “liquid nails,” apply it, wait 15–25 minutes, and only then adjust the sealant. If you adjust it right away, the “liquid nails” will stick to your hands instead of the choke. Always degrease the surfaces before applying anything (with alcohol, etc.). It’s best not to use hot glue on the choke. You can use it on the capacitors (they don’t get very hot if the fan is running).
The output has one MBR30200PT diode assembly rated at 200 V, 30 A. That’s not enough (the power supply’s output current is 20 A), so fit a second one from the same manufacturer. You can use a 150 V part, but those are less common and may cost more. Remove the short factory jumper between the diode assemblies. Fit the second assembly and solder the two marked jumpers. The mounting plate will have to be made from scratch; the old one only fits a single diode assembly.
Next, install the unpopulated jumpers on the power supply output (on both the negative and positive sides). If you like, install a Y capacitor between negative and ground (two are marked). A new Y capacitor is also marked on the board
between the negative terminal of the bridge rectifier and ground (five Y capacitors in total).
I liked the placement of the load resistor, well away from the capacitors (2 W/680 Ω). Two resistors are marked, in parallel; I installed two 1.5 kΩ resistors (2 W).









To wrap up: if you think you've done everything you wanted to do to the power supply, grab a toothbrush. Apply alcohol or similar to the solder side of the board and give it a final clean. Wipe it down with an alcohol-soaked cloth. If you like, you can also coat the entire solder side of the board with PLASTIK-71(70) lacquer. Send the family out for a walk while you spray it. A note on the cleaning cloths:
Take an old pillowcase—yours or your wife's (there's usually a spare). Cut off a couple of pieces: one for cleaning flux off the board, the other for wiping off, say, thermal paste. To clear solder from holes, use wooden toothpicks or sharpened matches.
I wrote about where to get decent solder in Russia: ZUBR at DNS/Technopoint, MECHANIC at Profi, and green-spark. Buy thermal tape at green-spark too; it's cheaper than on AliExpress. They also sell heat-shrink tubing by the meter, in a choice of colors. Be sure to get FKS-P flux with a brush (20 ml). The brush will come in handy.
I tested the power supply under a load of around 320 W for an extended period. Briefly, I pushed it a little over 400 W. Everything was fine with the fan running. Be sure to lubricate the fan and keep an eye on it. This was a test of the modified power supply.
Looks like they made a big batch; this power supply is now available from many sellers. When choosing a power supply, keep in mind that the TL494 version and the MOSFET version (at the same price) can differ significantly in ripple. If ripple is critical, go for the TL494 version. The out-of-the-box ripple on the TL494 power supply is lower, often much lower.
The spot for a 45°C KSD9700 on the standard, older-version power supply—CLICK. Chinese manufacturers often put one lead of a 2 W resistor (26–30 Ω) in its proper place and solder the other lead where the KSD9700 should go.
That leaves the resistor acting as a jumper. Put the resistor in its proper place and the KSD9700 in its own spot. You can increase this resistor to 100 Ω (2 W). Choose the value based on your power supply, load, and fan speed. Photo from Kirich's review (marked with a red arrow and circled)

If you ever want to take the transformer apart, the best way is to boil it in water on the stove. You can use a large tin can as a container. Don't put it into boiling water; heat it gradually. Don't touch the hot transformer with a cold, wet cloth—it may crack. As you unwind the wire, write down in a notebook which terminals you unwound it from, the number of turns, and the winding order. I made a mistake: I started counting first, but the wire was aluminum and kept breaking. I gave up and just unwound it. Then I put it on the table and noted the order and terminals on a scrap of paper. I thought I'd take the transformer apart at my leisure the next evening and count the turns with it disassembled. That didn't work out: someone had tidied up, and the wire and notes ended up in the trash chute. The transformer's plastic parts were poor quality—the bushings were cast asymmetrically, thick on one side and thin, with holes, on the other.
Don't go out of your way to get varistors for the power supply. If it's plugged in at home through an extension cord or power strip, there's usually a varistor in there already. If you're installing the power supply in an enclosure, you can get an additional filter (some come with a socket).
Search AliExpress for "Power EMI filter".
220 V test lamp — CLICK.
How to add an extra 100 µF/400 V capacitor — CLICK.
Video about capacitors — CLICK.
That's about it. Good luck, everyone.