Mean Well LRS-200-12 review and modification

This review—or article, whichever you prefer—is a continuation of my write-up on modifying a power supply to turn it into a low-voltage UPS. Last time, I covered a 150 W power supply and complained that the LRS-150 turned out not to be a very good fit for this purpose. So, let’s see how its bigger brother fares.

A reminder: to keep up with new topics on the blog, I recommend following me on Instagram, where I’ll post notifications about all new topics 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

Like the LRS-150, this power supply has a fanless design. That was partly why I wanted to modify this particular model, but things turned out to be much more interesting.


This series includes no fewer than eight models with different output voltages. Most use the standard range of 3.3, 5, 12, 15, 24, 36, and 48 V, but there’s also an unusual 4.2 V model. To be fair, the 15 V power supply doesn’t quite fit the standard 12-24-36-48 V range either, but it’s fairly common. The 4.2 V model, apparently intended to match the voltage of a single lithium cell, is much rarer.
Data from the original datasheet.


There’s a terminal block on one end for the input and output. The output has three terminals for each polarity. Next to them are a trimmer and an LED indicating that the power supply is operating. Everything is pretty familiar.
It’s the same thickness as the LRS-150. One side has an input-voltage selector switch, but this time there’s an opening for it; on the 150 W version, it was under the cover.


Power supply dimensions.


Let’s see what’s inside. First, we’ll remove the perforated cover.


At this point I said, “Wait, this is an LRS-350!” And sure enough, it really does look exactly like the LRS-350 I reviewed a while ago—literally identical.


1. The input filter is a bit skimpy, to be honest. The one in the LRS-150 looked much more substantial.
2. The input filter also uses two capacitors, each with a varistor connected in parallel. The varistors are larger than those in the LRS-150, but that model also had a varistor on the input; this one doesn’t.
3. A pair of thermistors and a pair of varistors.
4. The transistors are pressed against the case, which also serves as a heatsink. The current-sense shunt resistors are 0.39 + 0.39 Ω, or about 0.2 Ω total. I think the 350 W version had two 0.22 Ω resistors.


1. Unlike the LRS-150, this power supply, like the 350 W version, uses a forward-converter topology. You can tell by the large output inductor next to the transformer. There are three output diode assemblies.
2. Bus bars are visible next to the diode assemblies, reinforcing the traces on the circuit board.
3. The output capacitors are the same as in the 150 W version, but there are three of them here. Don’t be surprised: this is due to the power supply’s topology. In a forward converter, the output capacitance can be significantly lower.
4. And here’s the explanation: the board is marked LRS-350, so this is a shared design. The board also specifies the fuse rating, and the 200 W version is listed there too.

In simple terms, it’s the same LRS-350, but without the fan and its control circuit, and with some component values changed.


Another difference between a single-ended forward converter and a flyback converter is that the transformer is smaller. For example, the transformers in the 200 W and 150 W versions are about the same size. This is because power supplies with a forward topology have no air gap in the core, so all else being equal, they can deliver more power..


The output voltage was slightly high to begin with. The full adjustment range is 9.7–14.1 V. The adjustment seemed smoother than on the LRS-150, although the range is about the same. Perhaps that was just a coincidence.


Regulation across the 0–20 A load range (0–115%) is excellent.


The overload capability is about 120%. I ran two tests, and the difference was within the measurement error.


Next, I ran my usual thermal soak test and monitored how the output voltage drifted with temperature. I didn't stick to any precise intervals, but ran it up to a sustained 20 A load. The test took about 1 hour 40 minutes in total.
1. No load; the power supply is only slightly warm.
2. After testing first at 6 A, then at 12 A, for about 50 minutes total.
3. After testing at 17 A and then 20 A, for the rest of the time.
4. The power supply is hot, immediately after removing the load.
5. 10 minutes after removing the load; the power supply was still connected to mains.
6. Another 40 minutes later, with the power supply switched off.

Unlike the ideal load regulation, the voltage here increases with temperature. The change is small, around 20 mV, and not at all critical, but it's there.


By the way, the datasheet has separate graphs for power supplies with 3.3–5 V outputs, because they have lower efficiency and consequently run hotter.


Temperatures at load currents of 12, 17 and 20 A


Individual components at a 20 A load
1. The output inductor gets quite noticeably warm.
2. The transformer is fine, with some headroom for a higher ambient temperature.
3. The inverter transistors barely got warm, I would say, unlike the overload protection resistors.
The thermistors also get quite hot, around 110–115 °C.


I deliberately left the board in the case until after the thermal soak test, but now I can take it out.
I'll note right away that the diode assemblies and transistors have thermal paste on them, both outside the cap/insulator and inside it.


It's the same board as in the 350 W model, as I mentioned earlier.


The hot side is identical.


The only changes to the output section are because the 350 W power supply was rated for 48 V, while this one is 12 V, so the values of some components have been adjusted accordingly.


In my opinion, the output section is a little easier to modify than the one in the 150 W model.


Input bridge rectifier, transistors, diode assemblies.
The input capacitors here are 330 µF; the 350 W model had 560 µF capacitors, which makes sense. Since they have the same diameter, this power supply still has some room to spare.


For obvious reasons, I didn't redraw the schematic; I don't see the point, since it mostly repeats the schematic for the 350 W version.


Now let's move on to the modification. This time I made a few changes to the schematic and the circuit board, and made two boards. One is designed for a fuse in the standard holder, but I'll say in advance that small SMD fuses are a better choice: their resistance is much lower. I showed this earlier using a Mean Well power supply as an example.


Also, since the output current is higher here, I used a different relay. I figured the power supply would be about one and a half times more powerful than the previous one, so I designed for 12 A of output current and 3 A of charging current, 15 A total. At 13.8 V, that gives 207 W.
A reminder: the power supply is rated for 17 A, but since we're increasing the voltage, we need to reduce the output current proportionally. That gives us a 15 A limit, which we then split between output and charging current. For example, we could set it to 10 A output and 5 A charging, or 1 A charging, in which case the output current would be 14 A.

The relay turned out to be the limiting factor. The ones I have at home are rated for 10 A maximum, so I had to look for the highest-rated one available in this size. I found the HLS8L-DC12V-S-A, and it’s inexpensive, around 50–60 cents.

The relay was specified for 15 A. There’s no datasheet at the link above, but there is one in the description for the HLS8L-DC12V-S-C, which has a different contact configuration.
According to the datasheet, this relay has one group of normally open contacts and can handle up to 12 A continuously. Other versions are rated for only 10 A, while the peak current is the same: 15 A.
But the maximum current printed on the relay case is 10 A, which is a bit odd. Does that mean the datasheet contradicts reality, or the other way around? In my view, the datasheet takes precedence.

In any case, I couldn’t find any other relays with this current rating in this form factor. There was one that was a little taller, but it wouldn’t have fit properly in the case. I remember this from modifying the 150 W version. The other relays had completely different form factors.


I made a few minor additions to the circuit. It’s important to install the VD8 protective Zener diode, rated for 12–15 V, because voltage spikes can damage the TL431. I’m not entirely happy with this section of the circuit and think I’ll change it in the future.
I didn’t populate the auto-start circuit (marked in green).


After assembly, I ended up with two small boards. Almost all the components are soldered on the copper side; only the relay, capacitor, and a couple of TL431s are on top. The “TLs” might move to the other side too.


The boards in Sprint Layout. The second board got noticeably wider because of the fuse, but as I mentioned, I eventually installed a small 1206 SMD fuse instead.


There’s clearly more room here. I only had to move one capacitor on the top side, and on the adapter board I moved one of the TL431s to the back, because it was in the way of the capacitor between ground and the power supply’s common wire.


A reminder: all the work involved in installing the board, cutting traces, and so on is done solely to make the device functionally complete. If this seems complicated or unnecessary, it’s much easier to connect the board directly to the power supply output. In that case, you only need to connect one additional wire to the power supply: the current feedback line. There’s no need to cut or modify anything.
But I find that option inconvenient and inelegant.


To mount the adapter board, drill a 2.3 mm hole and tap it M3. If you do it the way I did, make sure to drill in exactly the right spot so everything fits.


You also need to connect the power wires. The spots marked in red must be drilled in any case; for the green ones, you can either use the hole or solder to the power bus. I eventually went with the second option.


On the underside, strip the protective coating from the contacts and then tin them.


I made a mistake in one spot—and made it twice.
I needed to isolate the terminal block’s positive contact from the rest of the circuit, but I didn’t notice that a resistor was connected to it. I had to reconnect it with a jumper afterward. I could have left one trace partially uncut, which would have avoided the jumper, or done it as shown in the photo; it makes no difference. The important thing is that this resistor is connected as shown in the photo. It’s part of the power supply’s voltage feedback and must not be cut.


Solder the current feedback wire to one of the optocouplers.


On the schematic, this is the pin marked with an arrow. When I checked the board with a multimeter, I found a resistor between the optocoupler and the TL431, but that doesn’t matter at all: the current feedback wire goes to this optocoupler pin.


Next, take a 15 mm standoff, a washer about 0.5 mm thick (the relay is slightly taller than 15 mm), and a screw. I also stuck a piece of double-sided tape onto the relay.
Install the standoff, and at the same time solder a wire or a component lead to the bus. It’s best to use something with a large cross-section, since it carries some of the heat away from the adapter board’s shunt resistors. If you do it the way I did, the adapter board will simply slide onto this lead, and of course you’ll need to solder it afterward :)


The power wires ended up connected in a mirror-image arrangement. It would have been better to install the board with the copper side facing down, but then it simply wouldn’t have fit.
You can use 1.5 mm² wire; I used 2.5 mm² to be on the safe side.


The power wires connect to the underside of the board.


Although the design looks complicated, it actually has just 5 connection points, and I managed to fit it within the power supply's overall thickness, with a little room to spare.


Now we can move on to setup and testing.
First, set the output voltage. This must be done at the output terminals, which are now terminals 2 and 3 (positive) and 4 and 5 (negative). Adjust it with no battery connected, setting it to 13.80 V. The voltage also adjusted without any problems here, by the way, though it dropped slightly afterward—not enough to matter.
Once adjusted, the battery can be connected to terminals 1 (positive) and 6 (negative). The battery was almost fully charged, and after power was applied, the balancer LEDs even lit up, although, as I wrote in my review of it, for some reason they refused to light up back then.


A few tests. First, let's check the voltage drop across the different sections of the circuit.


Current: 8 A. Voltage drop across:
Positive input-to-output
Negative input-to-output
Shunt
Relay
Fuse


The same test, but at 10 A.


And the maximum, 12 A.


Results in a table, compared with MeanWell PSC-160A-C
At 10 A, the modified LRS-200 has a total voltage drop of 0.178+0.095=0.273 V. The modified PSC-160A-C comes in at 0.227 V, slightly lower. I think that's because the PSC-160A-C has no wires—the connections are all directly on the board—or perhaps it has a higher-quality relay.
By the way, it's better for the relay to switch on under load rather than with no load—that is, it should switch on first, and then current should start flowing.


Board temperature at 10 and 12 A. Unfortunately, I couldn't keep it running at 12 A for long because the transistors on the protection board heated up quickly. Even after relatively short tests, they were at around 155 °C.


The battery was almost fully charged, so I had to discharge it to measure the charging current. I set it to discharge at 8 A, which also let me check the current setting accuracy. I was even surprised: the set current still matches the actual current, several years after calibration :)
After discharging it, I turned on the power and checked the charging current. It was about 3.3 A, which is roughly what the calculations predict.


And now for a long-duration test at maximum output: 12 A load current, with the battery charging at the same time, for a total power of over 200 W.


The test started at around 2 p.m., so here are thermal images taken after 15, 27, and 37 minutes. The power supply kept running like this for another three hours or so, but the charging current had dropped and the temperature was barely changing, so I didn't take any more measurements.


End of the thermal test. Just out of curiosity, I connected the charge-complete indicator LED. It lights up while charging and goes out when charging is complete—or, more precisely, when the current drops to about 0.55 A. It doesn't go out all at once, but gradually, over a range of about 0.58–0.54 A. I could make it switch off abruptly, but I don't see the point.
Once warmed up and fully charged, the output voltage was 13.813 V. I could lower it a little, but I don't see the point.


To protect the circuit board from shorting against the case, I covered it with a piece of plastic from some blister pack.


Both power supplies have been modified.


And now, a quick summary.
I definitely recommend this power supply, just like the previous LRS-150. I have no complaints about either one.
The LRS-200 is better suited to this modification, and it looks like the LRS-350 should work too. The only thing to check is whether the fan control circuit will get in the way. Otherwise, everything fits fairly easily. There is one issue, though: the relay won't handle a high current—unless we're talking about the 24 V version of the power supply, or perhaps a high charging current is needed.

That's all for today. I hope this was useful. If you have any questions or corrections, ask away and I'll do my best to answer. Suggestions and advice are also welcome.