Minor restoration of three LRS-350-12 power supplies and converting them into chargers

Given the current situation—or, more precisely, the lack of "power"—charging batteries has become a serious problem, preferably with fast charging. On top of that, I got the idea of building a small UPS for the refrigerator, which also needs a charger.
I bought several LRS-350-12 power supplies for this whole project, but as you can imagine, things aren't quite as simple as I'd hoped :)

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 maybe write about something interesting too - link to my account

You can also subscribe to updates and new articles on the Telegram channel - https://t.me/KirichBlog

As you can imagine, I wanted to keep costs down as much as possible, so I checked OLX, where I found a listing: someone was selling three power supplies he'd tried to modify for 500 UAH. But that's no big deal, right?
Unfortunately, the seller insisted on pickup only and wouldn't budge on the price, even a little. The funniest part was that he'd used my own circuit for the modification and was offering it to anyone who bought a power supply from him :)
Of course, driving to Kyiv to pick up the power supplies was out of the question, but a huge thanks to a friend who, as it turned out, lives near the seller. Anyway, Kirill met the seller, bought the three power supplies, and sent them to me by Nova Poshta, paying for shipping himself. Thanks again!

I got the power supplies and started figuring out what was what.
Their condition was pretty sad, though not beyond repair. One had been heavily butchered, while the other two were more or less okay. The parts and screws were in a separate bag, and one power supply also had its fan grille cut out and the fan reinstalled backwards.


The power supplies had basically been taken apart, and the boards weren't screwed in, which actually made the job a little easier.


There were jumpers on the boards and components had been desoldered, so I first had to put everything back the way it was. Naturally, I started with the two units in better shape.


After that, I gave them a quick clean on top as well.


The first power supply started right up, but the second had a few surprises in store.
First, it had a jumper across the zener diode in the PWM controller's power supply circuit, and some of the resistors in that circuit had been shorted out. I restored the circuit, but found that several resistors had failed. They measured fine in-circuit, but didn't work when soldered in place—a case of mechanical damage. I replaced them.
Then it turned out the PWM controller was dead. I took one from the most heavily butchered power supply, and it came back to life. It turned out someone had replaced the trimmer with a fixed resistor, so I quickly desoldered a trimmer from that same butchered unit. Then it worked perfectly.


The third power supply had taken the worst beating. The seller had tried to modify it to run from 100 V DC. In theory, this can be done, but he chose the wrong power supply. I'll try to explain why.
This is a forward-converter power supply, so it doesn't have much headroom on the input voltage. A flyback power supply, on the other hand, can operate over a wide range.



The transformer has two primary windings connected in series. This improves coupling between the windings, with the secondary winding placed between the two halves of the primary.
The person who modified it did the right thing in principle by connecting the windings in parallel, but there's one "catch": nobody guarantees that the windings are identical. One could have 56 turns and the other 55, so connecting them in parallel would give you a mess.
Anyway, I reconnected the windings the way they were originally.

If a power supply is rated for an input range starting at 180 V AC, or 240 V DC, then with the windings connected in parallel it should theoretically work from 120 V DC. It would be right on the limit, but should work, possibly with reduced output power.


The seller had also changed the current-sense shunts, desoldered the thermistors—one of which had gone missing—removed a zener diode, and changed the rectifier diodes. A real party.


The board itself looked pretty rough too, so I had to clean everything up and replace a few other small components. I can't even remember anymore.


Since the first two power supplies worked, and the third needed components both for the repair itself and to replace the ones I'd removed from it as donor parts (the PWM controller and trimmer resistor), I moved on to modifying the first two.

I happened to have three boards left over from another project for adding CC mode to a standard power supply, so that's what I'll use.


The adapter circuit is very simple: a TL431 voltage reference and two op-amps. One handles current feedback, and the other drives the indicator. The inputs are fed from voltage dividers connected to the reference and the current-sense shunt, respectively.


The connection to the power supply is highlighted in red; it's all very straightforward here too. When the current exceeds the set value, the board sends a current-feedback signal and the power supply starts regulating the current.


On the power supply board, you only need to cut one copper pour and drill four holes. Then install the current-sense resistors.


The board was originally designed for a different power supply, but it just so happened to fit the LRS-350 series almost perfectly.


The locations for drilling the holes and connecting the board's terminals. The hole for the +PSU connection may already be there, though.


And here's what the board looks like installed. I soldered the LED in later; in this configuration, it shines through the gap between the terminal block and the case.


Power it up and check it. For the test, I used one of the electronic loads I build.


These power supplies are rated for 12 V, but since we need to charge a 12 V battery, we set the output to the standard 13.8 V. Don't worry that this is higher than the voltage specified on the power supply: they can normally be adjusted up to 14.1 V. These power supplies also have an advantage: OVP protection. So even if something goes wrong with the stock feedback circuit, the power supply won't put out a high voltage; it'll simply shut down into protection mode.


Unfortunately, I hadn't accounted for the fact that we're not measuring the voltage directly across the shunt: there's also a voltage drop across the contacts and the copper pour itself. As a result, the actual current was quite different from the calculated 25–26 A.
The load was operating in CV mode; the regulation voltage is shown at the top. I'll explain later why the current drops like that.


To adjust the current, I reduced the value of the divider's upper resistor by adding a 39 kΩ resistor in parallel with the stock 10 kΩ one.


The current went up to 24.7 A. You can see that at a CV voltage of 13.2 V, the current is 24.7 A, but at 13.3 V it drops to 23.98 A. The catch is that this is the voltage at the load terminals; the voltage at the power supply output is noticeably higher. So the power supply actually holds 24.7 A all the way up to an output voltage of 13.8 V.
The charging indicator turns off at 4.9 A and turns on at 5 A.


On the second power supply, I also reinforced the copper pours around the current-sense resistors. This raised the current a little, but didn't help much; the only difference was that I had to use a 56 kΩ resistor instead of 39 kΩ.
The current came out to the same 24.7 A here.

To answer the obvious question: why am I setting it to around 25 A instead of the 29 A stated on the power supply? It's simple: the power supply is rated for 350 W. At 13.8 V and 25 A, the power is 13.8 × 25 = 345 W, so I'm leaving a little headroom even for the very last stage of charging. The power will be a bit lower the rest of the time. The point is not to exceed the rated 350 W and to leave a little margin.



While I was modifying the pair of power supplies, the parts for the third one arrived. I ordered them from Kosmodrom, and the order included:
1. Panasonic EEUFR1C102B capacitors - link
2. STF18N60M2 transistors - link
3. NCP1252ADR2G PWM controller - link
4. Silicone caps for TO-220 - link
5. SF38 diodes - link

The transistors, PWM controller, diodes, and trimmer caps were bought specifically for the repair; they cost about 100 UAH. I bought the capacitors for the modification and possibly to replace the originals.



The PWM controller has the ADR suffix. The transistors are in fully insulated packages. I got the diodes wrong: I bought decent, fast ones, but for some reason in a larger package.


Buying the PWM controller turned out to be a problem: only versions with the A and C suffixes were available, while the original has the D suffix. According to the datasheet table, version C won't work because its maximum duty cycle is 65%. Version A has a 50% maximum, which is of course much closer to the 47.5% of versions D and E, so I had to buy that one.


I bought the capacitors both as possible replacements for the originals, since the power supplies are used, and for the modification—especially since there is even space for them on the circuit board. I chose ones with roughly the same specs as the originals.


But it turned out they didn't need replacing. The measurements of the new ones at 1 and 7.8 kHz are shown above; below are the old ones from the two power supplies. The old capacitors are actually still fine, so there's no point replacing them. I'll write a little more about the modification below.


I installed the transistors, which were missing to begin with, and replaced the shunts; the originals were in the included bag.
I found a trimmer in my parts stash, though it's reverse-acting, so now turning it left increases the setting and turning it right decreases it. I also installed an input thermistor; there were jumpers there originally, and the kit included only one of the two. I had to find a replacement: a single, more powerful one with suitable resistance and current rating.


First, I power it up through a light bulb, but everything works right away.


Just to be safe, I checked the full output-voltage range. The adjustment works without any issues, too.


While restoring them, I noticed something interesting.
First, all three power supplies had different circuit board design dates: 2016, 2018, and 2019. It turned out that some of the components differed as well.
1. 2019: no original transistors; I installed 18N60M2 in plastic packages. Gate resistors: 8.2 Ω
2. 2018: 6R280P6 transistors in plastic packages, with 22 Ω gate resistors
3. 2016: 18N60M2 transistors in metal packages, with 51 Ω resistors.

The resistors are original. At first I thought someone had replaced them after blowing the transistors, but no—that's how they came from the factory.



While soldering the transistors, I noticed a messy solder joint on one of the capacitors. They'd piled on solder with a very generous hand—and it was clearly done at the factory.


There was one more difference: the 2019 version (and probably all newer ones) has an extra spot for a capacitor. In the second photo, you can make out that there isn't room for this capacitor underneath the circuit board.


Now for the additional capacitor.
The idea is that a capacitor after the shunt helps improve the stability of CC mode. Since all the capacitor positions on the circuit board end up before the shunt, I had to cut a trace and run a lead to the right place.

On the left: on the older versions, you just need to trim the copper pour a little and run a lead to the shunt (sorry, the photos are at different angles). The capacitor is at the bottom; run the lead up and to the left.
On the right: the newer version. The added capacitor makes cutting more difficult, but the idea is the same: connect to the capacitor's negative terminal at the top right, then run the lead left and down.

You don't really have to do this; it works fine as is. In fact, I tested the first two power supplies without these capacitors, but it's still better to add them.


Power it up—it works, but the current is like the second unit's, around 20 A. So, just as before, add a 56 kΩ resistor in parallel with the upper resistor in the divider. That gives us roughly the same 24.6–24.7 A output.
So you can see that the modification is highly repeatable. In principle, it's easier to use a lower value than 10 kΩ for the upper resistor, but for 25 A you need about 8.45 kΩ, and the E24 series only offers 8.2 or 9.1 kΩ. Accordingly, it will be either higher or lower, or the lower resistor in the divider will need to be changed as well.


The LED stays off at 5 A, flickers at 5.1 A, and lights steadily at 5.2 A, which is fine in principle.


Next, I removed all the boards, washed and cleaned everything, replaced the bad silicone caps, lubricated the fans, put back the one that had been installed the wrong way around as it was from the factory, and added a mesh screen. Then I reassembled everything and put it through a test run.


In the end, I got three tested chargers delivering around 25 A, built around genuine power supplies, even if they are used.

As for the power supply shown above, that's a separate story, and I'll tell it another time. In short, I wasn't happy with the charging algorithm when the charger is operating in buffer mode, so I decided to modify it. I used an LRS-200 power supply as the experimental unit because it's essentially the same as the LRS-350, but with passive cooling, which means the adapter board should be more or less universal.


By the way, about the fans: remember that I recently posted a review of counterfeit Sunon fans? I wrote to the seller then, too. To be fair, he replied right away, even though it was already night, but the conversation soon went off the rails.
I sent him a link to the article and explained my complaint. In response, he complained that apparently I was just being picky, and that the fans were great—“everyone buys them and nobody complains.”
So, who do you think is in the wrong here: me, for expecting the product to match its advertised specifications, or the seller, who insists that a sleeve-bearing fan works just as well as one with ball bearings? Especially since the product page clearly listed both the manufacturer and the bearing type.

There was actually more to the exchange. The seller claimed the sleeve bearing was rated for 20,000 hours, and also asked, “What should I put in the listing as the name, Sony?” I replied that he could simply write it the way Chinese sellers do, something like “For Sunon,” or say that it isn't genuine.




So, I bought three power supplies for 500 UAH, and a friend paid for shipping. The repair components cost around 100–120 UAH, plus I had some parts at home. The conversion boards are relatively inexpensive, so let's round their cost up to 50 UAH each. All told, not counting the time spent, each charger came to about 250–300 UAH. Pretty good, if you ask me—and they're Mean Well units, after all.
There is one downside, though: I bought four fans to replace the originals. One small one went into the RD6024, so we won't count that, but I still spent around 400 UAH for nothing. Really, that amount should be added to the cost :(


That's all for now. I have three articles planned for the near future: modifying the LRS-200, a review of the RD6030, which I received the day before yesterday, and a review of an 800 W power supply.

I hope this was useful. As always, I'd be very glad to hear your comments, and thanks again to Kirill for helping me buy the power supplies!