SCP-35-12, a very strange UPS from Meanwell
To stay up to date with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about every new article 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
Anyway, I told him to send it over. Usually there are two possibilities: either the battery isn’t charging, or there’s a problem with the voltage monitoring circuit. If I’d known then how bad things were, I… well, I’d have told him not to send it, because there was no point—and then there wouldn’t have been an article. So there was a point after all.
He also asked if I could increase the charging current, and add a button for forcing it to start when replacing the battery.
Sure, the battery might have less actual capacity than stated, and the load might draw a lot of current, but let’s take things in order.
So, I received a little box containing a UPS, model SCP-35-12. I’ve known the AD-55 and AD-155 models for a long time, as well as their relatively recent successors, but this is the first time I’ve seen this one.
I’d guess this was a predecessor to the AD series. After that, they released the PSC-160A-C, which is a genuinely interesting UPS. They also make UPS units that connect to a standard power supply and essentially sit between the power supply, the load, and the battery—and there’s even a DIN-rail version.

There are two versions of this UPS, rated for 12 and 24 V. I have the 12 V version here.

The connections are straightforward: a terminal block for the mains input, one for the load, a pair of wires for the battery, and three wires with a connector carrying the power output and a mains-failure signal. In practice, it’s just the same 12 V, but without battery backup.
There’s also a trimmer and a connector for the temperature sensor.
I’ll point out the idiotic cover design right away: the capacitor gets in the way when you try to remove the cover, and of course it gets in the way when you put it back on, too. You can see its insulation is already damaged.

The temperature sensor is a neat idea. You connect a 100 kΩ thermistor, which adjusts the charging voltage according to the temperature—quite useful.

The board from different angles. It’s basically a typical Mean Well power supply; I’d even say there’s nothing unusual about it—it’s as standard as they come.



On the input side, there’s a mains filter, a small fuse, and no varistor. Otherwise, everything looks fine. There’s a thermistor, but its markings aren’t visible. The diode bridge leads have ferrite sleeves fitted over them; in fact, ferrite sleeves are used on quite a few components here.
The input capacitor is 82 µF, which is plenty for a 35 W power supply.

1. High-voltage transistor P9NK70
2. PWM controller supply capacitor
3. Two optocouplers: one for voltage feedback, the other clearly for output overvoltage protection.
4. X12SL output diode assembly
Again, all classic Mean Well power supply design.

The board is held in place by a single screw. That’s actually convenient: I like that it isn’t screwed to the case, which means less messing around with thermal paste.

The board is single-sided, which is another indication that this power supply is clearly based on an older design.

The PWM controller is marked 1203P60, which is none other than the NCP1203 from OnSemi. For some reason, it was potted with sealant.


There’s a 0.5 Ω shunt on the primary side. I think this will be useful for anyone repairing this power supply.
There’s also a small thyristor on the hot side—a trigger-type overvoltage protection circuit, a standard Mean Well design.

The output has three capacitors in parallel: 1800+470+470 µF, all rated for 25 V, all name-brand. Classic.

The manufacturer clearly put a lot of effort into the filtering: the capacitance is generous, and they didn’t skimp on the inductors either. The resulting chain looks like this:
1800 µF - inductor - 470 µF - dual-winding inductor - 470 µF.

I switch it on: 13.86 V at the output. A little high, but nothing to worry about.

Next, I decided to look at the block diagram first, and realized I was looking at some nonsense. I checked the circuit board, and sure enough, it really is nonsense.
The thing is, this power supply is essentially just a charger, because the load is always connected to the battery. Well, except that the battery itself is connected through a fuse.
No, it’s a simple, crude solution, but it’s awfully amateurish.
In practice, we have a CC/CV charger whose output is connected through a diode (the feedback is actually taken after the diode; the schematic is wrong) to the battery, with the load connected to the battery.
In other words, there’s no over-discharge protection at all. If the load shuts itself off at 12 V, everything is fine; but if you connect a load that can run at both 5 V and 12 V (many routers, for example), you’re in trouble.
In fact, there’s nothing here besides the charger, a protection diode (to prevent the battery from discharging back into the power supply circuitry), and a fuse.
Well...

The specified output current is 2.6 A. Let’s check.
The electronic load was set to CV mode (important when testing devices like this), and I simulated the battery voltage over a range of 12–14 V.
At 9 V, the current is about 4.5 A. In practice, if there’s no load, that will be the battery charging current. If you have a typical 7 Ah battery, that’s really bad for it. If you’re not sure why, check the charging current specified on the battery.
As the voltage rises—or, in other words, as the battery charges—the current drops. Essentially, the power supply regulates power rather than current, because the shunt is on the primary side.
At around 12.5–13 V, which covers most of the charging process, the current is about 3.7 A, also clearly more than 2.6 A.
At 13.7 V, the current started to drop noticeably. I’d say that’s pretty good, since it stayed at a decent level almost all the way to the upper limit, even though the load was connected with relatively long wires.
The current drops further, and at 13.9 V it’s almost zero. Excellent.

Just out of curiosity, I decided to test how hot this power supply gets in different operating modes, since the current is almost one and a half times the rated value.
For this, I used one of the loads waiting for a new owner :)
Same operating mode: CV.

I checked at random intervals, about 40 minutes each, but at different output voltages—12, 12.8, 13.5, and 13.7 V—to simulate the battery charging process. The load power was 45–50 W in the first three modes and 33 W in the last one.

I have to say, the power supply handled the heat quite well.
In the first test, at 12 V—let’s say the battery is just starting to charge—the hottest component was the output protection diode at 107 degrees. The next hottest was the output diode, and third was the inverter transistor at just 72 degrees.

At 12.8 V, the situation was almost unchanged.

But at 13.5 V, the power supply managed to surprise me. The power was about 10% higher than in the first two tests, but the transistor temperature was lower. I suspect the power supply was operating in a more optimal range. The diode temperatures probably dropped because the output current fell slightly.

At 13.7 V, temperatures predictably dropped, since the power was down to 33 W. Let’s say the hypothetical battery is hypothetically charged.

The circuit works, in principle. The only thing to keep in mind is that the load’s current draw comes out of the charging current—in other words, the higher the load current, the lower the battery charging current. For example, if your load draws 2 A, the charging current will be about 1.7 A, since the current at the midpoint of the adjustment range is 3.7 A.
But what about the lack of automatic cutoff? The simplest solution is to use a lithium iron phosphate battery and a BMS board. That solves all the problems at once. The battery will also last longer under light loads, since these batteries handle high charging currents better.
What if the battery is lead-acid?
There’s a solution for that too: a similar protection board.

Or perhaps even this one—it might be better, since it has an adjustable cutoff threshold.

The difference between the boards is that the first has a “start” button, while the second doesn’t.
With the first one, after the protection trips, you’ll have to press the button once power is restored so the relay can reconnect the power.
I’ve drawn both options. In the first case, the relay and control circuit can be powered from the circuit’s output. When the protection trips, the circuit is completely de-energized, but it can’t restart on its own, so you have to press the button briefly.
With the second option, there’s no need to press a button, but the circuit may turn back on after shutting down if the battery voltage rises above the trip threshold.

In my opinion, the first option is clearly worse, especially if the UPS is somewhere far away or hard to reach. But what about automatic restart?
That’s very simple: we can connect the input of the second circuit to the UPS and its output to the battery. Here’s what happens: the voltage drops, the relay disconnects, and power to the circuit is cut at the same time, since it was essentially powered through those contacts. Once the power supply starts working again, 13.8 V appears at this input, the relay switches on again and connects the battery, and the circuit is once again fully ready for charging and discharging.
The board with the button can be installed on either side of the UPS—between the UPS and the load, or between the UPS and the battery—but you’ll have to press the button.
The load connection shown is just illustrative. In practice, the only difference with a proper connection to the UPS terminals is that there’ll also be a fuse, but the principle is the same. What matters is where the board is connected and which way its input and output are oriented.

That’s about it. Despite its fairly good performance, reasonably decent build quality, and good components, the UPS circuitry is utterly primitive. In its original configuration with a lead-acid battery, I don’t recommend using it. You should either add a protection board or use a lithium iron phosphate battery with a BMS board, which is essentially the same thing as a protection board.
Yeah, I really wasn’t expecting that.
As for the original problem, I have a few theories, assuming it used to work normally. Here they are in order of likelihood, in my opinion.
1. The load has increased, and the UPS can’t keep up with charging the battery while it’s also discharging faster.
2. The battery’s capacity has dropped, especially if there’s no voltage monitoring and it has been over-discharged—particularly if that’s happened more than once.
3. The battery itself has failed. For example, one cell may have shorted, but there are two batteries connected in parallel, so I think that’s unlikely.
That’s all for now. I hope this was useful, and as always, I’d be glad to hear your comments.
scp-35-12 как подключить, scp-35-12, scp-3430 hp блок питания, 11, scp-35-12 mean well подключение
Questions and answers
What charging current did the Meanwell SCP-35-12 deliver?
At a battery voltage of about 12.5–13 V, the current was around 3.7 A, above the stated 2.6 A. It dropped almost to zero at 13.9 V.
Does the Meanwell SCP-35-12 protect its battery from over-discharge?
No. The load is permanently connected to the battery, and the author found no built-in low-voltage protection. He recommends adding a protection board or using a LiFePO4 battery with a BMS.
Why might the Meanwell SCP-35-12 shut down after about an hour and a half?
The author suggests three possible causes: increased load, reduced battery capacity after over-discharge, or a faulty battery.
The figures from this review are in the measurement database: Chargers
Discounted products
You may be interested in
Discounted products
All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!
Comments3
Only registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.
Log inRegister




Comments are published in their original language.