Review and conversion of a Mean Well LRS-150-12 power supply into a UPS

As I wrote earlier, the next step will be putting the PSU-UPS adapter board to use. Along the way, today I'll also review the Mean Well LRS-150 power supply.

As a reminder, to keep up with new posts on the blog, I recommend following me on Instagram. I'll post notifications about all new posts there, and might also write about something interesting from time to time: link to my account

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

This is far from my first review of Mean Well products. Before we get started, I wanted to list all the Mean Well power supplies that have passed across my workbench, just in case. I've even drawn schematics for some of them, which might come in handy for someone.
RPS-120-27
SDR-120-24
EPP-100-12
EPS-65S-24
HDR-60-12
HDR-100-12N
IRM-60-24ST
LRS-350-24
SP-320-5
PSC-160A-C part 1
PSC-160A-C part 2
LRS-100-48
LRS-350-48
HDR-15-24
EPP-500-54
AD-55A

While putting together the links, I was surprised myself by how many of these I've reviewed. Today's review is already my sixteenth Mean Well power supply, and it won't be the last, since I have two more.
Yes, I picked up a couple of power supplies, rated at 150 and 200 W, specifically to convert into UPS units. But mostly, I wanted to find out which one would be easier and more practical to modify. I bought them locally; the AliExpress link is just for reference, although the price there is pretty good too. I haven't come across any counterfeit Mean Well units.


Today we'll look at the smaller 150 W version, as indicated by the model number.
The voltage is also indicated: 12 V. This series includes five models: 12, 15, 24, 36, and 48 V. Interestingly, unlike most other product lines, there's no 5 or 9 V model, so 12 V is the lowest voltage available.
Here and below, I'll quote excerpts from the datasheet.


On the outside, it's a standard power supply in a perforated metal case.


Perhaps the only thing that sets the LRS series apart visually is its slim profile. And believe me, that can sometimes be critical.


Speaking of dimensions...


It connects via a screw terminal block with no cover. There's an output voltage adjustment and an LED indicator showing when the output is on.
One slightly unusual feature is that this isn't a wide-range power supply, so it has a 115/230 V selector switch, partially hidden behind the case grille.


But there's not much to see on the outside, so let's remove the cover and take a look inside.


The layout is fairly compact, which is a drawback for my modification, since it'll be difficult to fit an additional board in here.


A few close-ups of the components.
1. The input EMI filter is excellent. Not only does it have a fuse, several varistors, X capacitors, and two thermistors, but it's also generously fitted with Y capacitors—before and after the inductor, and between the secondary and primary sides. There's also a spark gap.
2. The input has a 5 A fuse and a 470 V varistor.
3. Unusually, there's an interwinding capacitor on both the negative and positive output rails: 3.3 nF on the negative rail and 10 nF on the positive.
4. The spark gap flashes when there are mains voltage spikes, for example when power is applied and arcing occurs.


1. Two thermistors are located after the bridge rectifier, one on the positive side and the other on the negative side of the high-voltage circuit.
2. The capacitors are connected in series, with a 240 V varistor in parallel with each one.
3, 4. Lelon input capacitors, 330 µF, 200 V. I think the narrow input range and the switch are a necessary trade-off for compactness: a wide input range would require a higher-capacitance input capacitor. As it is, 165 µF is enough for 150 W of output power.


1. A thermal sensor for overheat protection is pressed against the transformer. Next to it are two optocouplers: one for the main circuit and one for output overvoltage protection.
2. Hot side: TVS diode, snubber diode, transistor.
3. Output diode assemblies, with the case serving as a heatsink. Classic.
4. There are no fewer than 8 output capacitors: Rubycon 1000 µF, rated for 16 V, though 25 V ones could have been used.


An insulator is laid under the board, and the board itself is held in place by the case's molded supports, one screw, and a couple of clamps for the power components.


The underside is nice and tidy, and the board has protective slots. Looks good.


1. High-voltage transistor 60R290P made by MagnaChip. I don't think I've come across one of these before.
2. I remember someone complaining in a video review of a cheap Chinese inverter that it had transistors from different batches. Well, here's a branded power supply, and on the output we have a pair of diode assemblies connected in parallel, clearly from completely different batches—probably even different factories :)


There's plenty of stuff on the underside too:
1. Input X-capacitor discharge circuit. A dedicated chip was originally planned, but they decided to use plain resistors instead.
2. The PWM controller chip is marked MW03A. I've seen it before in the LRS-150, but couldn't find a datasheet.
3. The output circuit is a classic design, with feedback based on a TL431. The board is also designed to accommodate a synchronous rectifier, which is usually fitted in the 3.3–5 V models—the lower-output models in the range. But this one is a 12 V model, the lowest-voltage version, and it doesn't have a synchronous rectifier.
4. Output overvoltage protection circuit, used to protect against failure of the main feedback circuit. It's set to about 15–16 V, with a 14 V Zener diode.


There was a block diagram in the datasheet, but there's nothing interesting here.


And now for a few tests.
1. It was initially set to 12.15 V.
2, 3. The full adjustment range is 9.6–14.1 V. Excellent: not only does it fully meet my needs, but the OVP protection is also set to a higher threshold.
4. The adjustment is a bit coarse, but setting it to 12 V was no problem.


Voltage regulation under varying load is excellent. Across the 0–14 A current range (with a maximum rated current of 12.5 A), the change doesn't even show up in the last digit of my multimeter.


Overload protection trips at around 15.6 A: 15.66 A on the first run and 15.62 A on the second. Since current is measured on the primary side, the threshold will drop as the output voltage increases.


A short thermal test:
1. Test start, no load; the power supply is slightly warm.
2. After 20 minutes at 4.2 A load (33%)
3. After 20 minutes at 8.4 A load (66%)
4. After 20 minutes at 12.5 A load (100%)
5. After 30 minutes at 14 A load
6. After a total of 1 hour 30 minutes of testing, once the load was removed.

There's a clear positive correlation between voltage and temperature: we started at 12.008 V and finished at 12.034 V.


After the test, I left the power supply on with no load for half an hour, but the voltage only dropped to 12.020 V.
Then I switched it off to cool down for another hour and checked again. The voltage was 12.013 V. So the voltage is drifting, presumably because the wiper of the variable resistor shifts with temperature, though I can't say for sure.


The power supply was tested at room temperature, with a 230 V mains supply, and in a horizontal position. According to the datasheet, it can be loaded to 100% without any problem, but I pushed it even harder.


I checked the temperature in every mode, but I doubt anyone cares what it looks like at 33% load, so here are thermal images at 66%, 100%, and 110% load.


And a closer look at a 14 A load.
1. Input thermistors. At 66–100% load, their temperature is noticeably lower.
2. Output transistor: cool, but the snubber diode gets hot.
3. Interestingly, the output diode assemblies aren't the main source of heat; it's the board next to them. On the underside, there's only one RC network and the traces.

The transformer, on the other hand, reached just 81 degrees even at 14 A, so there's still some power headroom—or, more precisely, some headroom in terms of ambient temperature, which is certainly good.


Now let's turn the power supply into a UPS. For this, I used the board from the previous article.


The only place it fits is where the input voltage selector switch goes. Since I don't need the switch, I had already desoldered it to test-fit the board.
I also had to desolder the varistors connected in parallel with the filter capacitors and remove a jumper that was in the way of the relay.
It would of course have been nice to keep the varistors, but unfortunately, that wasn't possible.



I had to move the jumper to the other side of the board. One interesting detail: the thermistors are wired so that the startup circuit is powered from before the thermistor, while the Y capacitors are connected after it. The corresponding traces were added for this purpose.



1. I only noticed the thermistor's marking after desoldering it, so here's a photo just in case.
2. Test-fitting the board, drilling holes, and moving the thermistor.
3. I had to shift one of the thermistors slightly. It wasn't really in the way, but since it gets hot, I decided to move it a little farther away. The other thermistor had been installed right between the bridge rectifier and the switch, with virtually no clearance.
4. Drill one hole, scrape the trace, then cut the trace nearby and solder a jumper to the original installation point.


This power supply isn't well suited to modification. I had to cut traces and find places to drill holes without anything getting in the way.
Three holes need to be drilled; the fourth uses the existing hole for mounting the busbar.


It was especially important not to hit the terminal block or any of the capacitors.


The board is mounted on plastic standoffs. It was originally designed with one mounting point, but for safety and reliability, I decided to use two.
I messed up a little during the process: I drilled one hole correctly at 2.3 mm, but made the second one 3 mm. I had originally planned to tap it for an M3 thread so the standoff could screw directly into the board, saving the nuts and reducing the thickness of the hardware underneath.

I soldered the CC circuit wire on right away.



Next, I prepared four pieces of 2.5 mm² wire. 1.5 mm² would have worked too, but 2.5 mm² is better, and I had some on hand. Naturally, the shorter the wires, the better.


The board is connected. One convenient thing is that the battery negative trace runs to both edges of the board. I also installed the missing fuse right away: I used a 20 A one because I didn't have a smaller suitable fuse. Ideally, use a 10–15 A fuse. While I was at it, I replaced the standard 150 Ω resistor in the lower leg of the current-sense divider with a precision 100 Ω resistor, and added another shunt resistor (soldered on top), reducing the resistance to 10 mΩ.

Reducing the divider resistor value cuts the charging current by a factor of 1.5, while reducing the shunt resistance increases it by a factor of 1.33, so the two effects roughly cancel each other out—but there's a “catch.” The planned continuous current is 8 A. At that current, the voltage drop when running from the battery will be 0.08 V, and the power dissipation will be 0.64 W. The board should work fine at 10 A too: the drop will be 0.1 V and the dissipation 1 W. That's nothing to worry about, since the power wires also act as a heat sink. That's the optimization I came up with :)


The underside—it's barely noticeable :)


The installation is finished. And, again, to improve safety, I put insulation between the new board and the filter capacitors.


The first—and practically the only—thing to do is set the output voltage to 13.80 V. That's when I noticed something odd, and it wasn't caused by my board: the output voltage slowly drifts within a range of ±3 mV. I also found that the adjustment is rather coarse. It would be good to either remove this resistor altogether or modify it to make the adjustment smoother.


Now let's test it with an 8 A load. Everything works perfectly.


I wanted to measure the voltage drops in the key circuits at a working current of 8 A while running from the battery.
1. Battery negative to output negative: 107 mV
2. Battery positive to output positive: 155 mV
3. Current measurement circuit: 81.5 mV, which is in line with the calculations, taking the measurement error into account.
4. Relay and fuse contact circuit: 123 mV
5. Voltage drop across the relay contacts is about 100–110 mV; it didn’t fit in the photo.

The total voltage drop across the unit at 8 A is 107+155=262 mV, which is excellent in my opinion. Most of the drop is across the relay contacts, with the shunt in second place; the rest can be ignored.

Oh, I almost forgot: the power supply now has four output terminals. The outer ones connect to the battery, and the inner ones to the load.


There were no problems with the charging current either: a steady 2.1 A. After a while, as the battery charged, the current began to drop. The PSU output voltage was then 13.796 V, which is the maximum output voltage because this PSU has no “overshoot.” Even if it did (if connected with the control loop ground interrupted), it would be only 20–25 mV. So it works fine both with and without overshoot, though the first option is technically simpler.


In the comments on the previous review, people brought up possible problems. I think they could be caused by a TL431 failure in the battery voltage measurement circuit. In my opinion, you could try improving the reliability of this section by adding a VD8 Zener diode in parallel with the TL431; a 12–18 V rating should be plenty. I’ll have to add one to the PSU I’m reviewing.

A note about the board’s auto-start circuit and the battery’s automatic unlock circuit: I haven’t tested them yet, but I can say it’s better not to use both at the same time. The battery auto-unlock circuit could trigger the board’s auto-start with both circuits installed, disabling reverse-polarity protection.


And that’s the result. Here are a few words about the load conditions.
The PSU was originally rated for 150 W at 12 V, or a 12.5 A load. I raised the voltage to 13.8 V, which at 150 W gives us the required reduction in output current to about 10.8 A. Of that 10.8 A, 2.1 A goes to charging the battery, leaving 8.7 A for the user’s load continuously.
The current can easily reach 10–12 A for short periods. Even if the PSU goes into protection temporarily, the battery will remain connected, and once the load drops, the PSU will switch back to mains operation.

You might ask: can the charging current be increased? Of course. It’s easy to set it to 5 A or even 10 A, but there’s a “catch.” The charging current is a trade-off, because you’re effectively reserving that much of the total output current for charging.
Roughly speaking, if you have a 10 A PSU, you can set the charging current to 1 A and have 9 A left for the load, or set the charging current to 9 A, leaving just 1 A for the load. The PSU itself is limited to 10 A, and you can’t exceed that. The worst part is that once the battery is charged, the unused current goes nowhere.

A couple of words about the thermistors heating up. Yes, they’re close to the board, but they heat up when the board isn’t hot, and the board heats up when the thermistors aren’t hot :)
Also, the PSU will normally operate at no more than 70–80% load, since the current will drop once the battery is charged.


Just to be on the safe side, I also stuck some insulating tape to the underside of the cover near the battery’s positive wire, so even if the cover gets pressed in, nothing will short out.




That’s about it for now. Here’s what I can say:
1. This PSU is great. The filtering is excellent, and I have no complaints about the temperature either. There’s really nothing to nitpick. I liked it and definitely recommend it.
2. It’s really not suitable for conversion into a UPS. There’s too much mechanical work involved, two varistors have to be removed, and everything is packed in extremely tightly.

I’ll most likely put this UPS up for sale at a reasonable price, but first I’ll ask the person who sent me another UPS whether the one I built in today’s article will suit their needs.
The second 200 W UPS will most likely be for sale too, but first I need to finish it. Also, maybe someone can suggest what would be better: a 10 A output and 5 A charging current, or a 13 A output and 2 A charging current, for example?

That’s all for today. Good luck, and I’m off to tinker with the second PSU.