UPS-1228-12 low-voltage universal UPS board, updated 15.05.24

One of my readers saw my comment about a low-voltage UPS board and suggested I review it, so they sent me this board.
Given the current situation, these devices have become relevant again, so of course I agreed—especially since I was very interested in it myself.

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 too - link to account

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

I think everyone knows that we have a deranged neighbor nearby, along with his dictator boss, who decided it was urgent to liberate Ukraine from power plants, substations, and power lines—and recently, from television as well. For example, this is what one of the largest combined heat and power plants near Kharkiv looks like.


And here, I guess, there was a military factory too, or maybe it was full of observers? By the way, TV broadcasting was fully restored a week later.
I'm just curious: do Russians really believe the nonsense they're fed by all those Putins, Skabeyevas, Solovyovs, and other criminals?


And this is what Kharkiv looked like on March 22–23: yes, two days without electricity, the internet barely crawling, and making a phone call was almost impossible. I guess this is also “liberation”—including liberation from the people who can't even call an ambulance.


To be fair, though, based on last year's experience, Kharkiv residents at least couldn't care less about Putler and his cronies, or about all of Rashastan in general, because:
1. Stores operated without any problems on generators, even large supermarkets—and they even lit up the surrounding area.
2. Unlike last year, when communications and the internet went down completely, this time both worked, more or less. It was very hard to get through on the phone, though, and the internet crawled like it was the 2000s.
3. All the invincibility centers were open. There were quite a lot of people there, but still.
4. Personally, I got through the two days just fine and used about 25% of my battery capacity. By the way, while the power was out, I received an electronic load I'd ordered and assembled it. Then, once the power came back, I calibrated and tested it (I could have done that without waiting, but this was more convenient). So for now, the Russians aren't doing much to stop Kirich from working.
5. The military don't just not give a damn about this—they even hit Belgorod Oblast afterward.

So here's the question: who exactly is Putler fighting? These people? What a tough guy....
Back in February–March 2022, our garbage was collected regularly, and someone thinks they can break these people by cutting off their electricity and TV? They haven't got what it takes.



Yes, I've heard the story many times that they're shelling businesses that work for the defense industry. But you'd have to be an idiot to shell them in the third year of the war, hoping it will somehow hurt the military.
Let me explain: unless we're talking about smelting metal on an industrial scale, generators can do the job just fine. For example, a 1.8 MW gas turbine generator isn't all that big. Someone might ask how many of them you'd need. Well, however many you need, that's how many there are. What's more, if even our supermarket has what looks to me like at least a 1 MW generator, how hard can it be to install something more powerful for the military?
Besides, the grid is being restored—and guess who'll get power first? So here's the question: who is Russia fighting if the military don't care about these power outages?



And that brings us to the subject of this review. Backup power is great, but if your devices use low-voltage power—which includes all kinds of mini PCs, routers, access points, switches, optical terminals, and storage devices—converting 12-230-12 V is completely impractical. It's much better to power all of them directly from a battery.
But as is often the case, “there's one catch.” First, it would be nice if the whole thing worked on its own—that is, charged the battery, monitored it, regulated the voltage, and so on. Second, batteries usually output something like 10-14 V, or even worse, 10-16.8 V, and not every device will like that.

The device under review solves most of these problems, although it does have some drawbacks, and the price isn't great either. Still, it's definitely worth a look.

It's a small board.


The following specifications are listed, translated from the product page:
Input voltage range: 10-28 V
Maximum input current (input fuse rating): 15 A
Standby current: 95%
Output voltage regulation specifications
Output voltage: 2-28 V, continuously adjustable.
Output current: 8.3 A
Output power: 100 W
Conversion efficiency: > 95%
Output ripple: <80 mV (peak-to-peak)
Regulation accuracy: 2%




The board layout is moderately dense, that is, The layout is dense, but not excessively so.


The battery and power supply terminals are on one side, and the load terminal is on the opposite side.


Nothing interesting on the underside.


Let's see what this thing is built around.
1. The charge controller chip. You can already tell this isn't a primitive circuit, since it has a full-fledged controller, though its markings have been rubbed off. It's a PWM charger with synchronous rectification.
2. Some kind of protection diode, apparently for reverse polarity on the power supply input; it's connected in parallel with the supply.
3. The input power switching section. The two transistors at the top control power from the power supply input; the four below are connected back-to-back in the battery circuit. And there's a 15 A fuse.
4. All the electrolytic capacitors on the board are identical: 470 µF, 35 V.


There's not much point in describing how it works; it's as simple as the brains of those who came to our country.
There's a battery, a power supply input, and a charger. The battery and power supply are combined through the MOSFETs' built-in body diodes. To reduce heat, the transistors turn on when needed, so the output is either the battery voltage or the power supply voltage.
This “dirty” power then goes to the input of a universal converter, which provides the voltage we set and also protects against overloads.
So there is no such thing as “switching time,” and there can't be. And since the converter is universal, the output voltage remains stable regardless of what's at the input, even if it's below the required level.

Basically, this is roughly the same solution I proposed for my UPS.


A tiny Tiny24A handles all the processes, including voltage thresholds, timers, and other service-related stuff.


The universal converter is the classic LTC3780, which I've already written about.
This converter can operate in both step-down and step-up modes. The next step in this design approach is a reversible universal converter, which can not only step voltage up and down, but do so in both directions. These are used in modern power banks.


As you can see, the board has a fairly large heatsink. Let's see what's underneath.
First, there's some thermal paste underneath it. It's quite runny; I later washed it off and replaced it with my own.


Under the heatsink is a bridge made up of four AOD4184A transistors: 40 V, 50 A, 7 mΩ.


Current and voltage are adjusted analogically, using no fewer than five trimmers.
There's one at the top for adjusting the output voltage. Then, from left to right:
Adjusting the battery low-voltage cutoff threshold
Adjusting the battery voltage turn-on threshold
Setting the charging voltage
Setting the charging current.

It's all pretty straightforward, except for setting the cutoff and turn-on voltage thresholds. More on that later.


In short, the board has two test points, TP3 and TP4, which are used to set the respective thresholds and are located next to the corresponding trimmers.
The principle is very simple: the cutoff voltage is VUVP=VTP4×4 (V), and the automatic turn-on voltage is VSTART=VTP3×4 (V).
So, if we want to set the cutoff threshold to 10 V, TP4 needs to be at 10/4=2.5 V. To cut off at 9 V, it needs to be 9/4=2.25 V.
The procedure for turn-on is similar: to turn on at 12 V, set TP3 to 12/4=3 V.


To make things easier, the product page has two tables. The first shows the charging termination voltage to set for each battery configuration, and which power supply to use, since the charger only steps voltage down.


The second table shows the over-discharge cutoff voltages and the corresponding turn-on voltages. Very handy.


The board also has several jumpers for setting its operating mode. In fact, jumpers are installed in only 5 of the 6 positions. The one closest to the trimmer resistor is the output for controlling the motherboard, if needed; the other five set the operating mode.


The explanation is rather confusing, especially since the jumpers are labeled J4, J5, J6, and J7, which would suggest there are four, while the board has six. That's because jumper J6 consists of three positions, and different delay times can be set by combining the jumper positions.

In plain English, here's what I figured out:
J4, also known as MBSW: toward the edge of the board is -, toward the inside is +. It provides a 5 V logic signal, but the signal is momentary and active low.
J5, the board's power switch: closed means the board is on; open means it goes into sleep mode.
J7, auto-start: if the jumper is installed, the board will start again if the voltage rises above the threshold after a low-voltage shutdown. If it's not installed, the board will wait until you toggle the power button connected to J5.
And here's the problem, a really basic one. You essentially have two options: either the board turns itself back on after shutting down, then turns off again, then turns back on again, or you have to go and press the power button every time it shuts down automatically. Apparently, adding a feature that shuts the board down after a low-voltage event and keeps it off until power is supplied by the power supply is too complicated for the Chinese... honestly, it's idiotic...
J6 sets a two-stage delay. Let me explain with a specific example:
If you set it to 5/30 sec (position 1), it works like this:
The voltage drops below 12 V (the turn-on threshold); after 5 sec, an off command is sent via the MBSW output.
After 30 sec, the board's output is switched off.
When the voltage rises above 13 V, the MBSW output sends a command again, but this time it's a turn-on command.
So, the battery discharges, and once it drops below the threshold, the board tells the computer to shut down. Then, after a set time, it cuts power to it. As soon as the voltage rises again, a turn-on command is sent. But given that the board can start up by itself (see above), I don't think this feature is much use.
The table of motherboard power-control modes is on the product page. There's also a fan connector next to the jumpers. I don't know if anyone needs it, but it's connected in parallel with the board's output.



Let's move on to the tests. First, I simply applied power to the board's input. At 12 V, only the output LED and the LED in the center of the board glowed faintly. By the way, I still haven't figured out what the center LED does. It would sometimes blink, sometimes light up, but was off most of the time.
The board draws about 5 mA, provided there's voltage at the output, which is pretty good.
I raised the voltage from 12 to 20 V. The blue LED lit up, and the current rose to 40 mA.
I lowered it back to 12 V, and the green LED started blinking. The board's current draw changed as well.

As for the LEDs, there are five on the board: one in the center, whose purpose is unclear; one by the fan connector, which indicates that the board's output is powered; and three between the trimmer resistors.
These three LEDs indicate:
Blue — voltage is present at the board's input.
Red, in the center — the battery is charged.
Green — the battery is charging.

This wasn't much of a test, really. At the initial stage, I just had to figure out what the board actually does and how much power it draws. But one thing I definitely forgot to measure was the current drawn from the battery after the low-voltage protection kicked in....


For the initial tests, I used a battery from an old cordless drill. It was convenient, had enough capacity, and already had a protection board.


Next, set up the charger for the battery configuration. First, set the voltage to around 16.8 V, or a little less.
Then connect a multimeter in current-measurement mode in series with the discharged battery, and set the charging current.
In both cases, the voltage at the board's input must be comfortably higher than the charging voltage, by at least 3–5 V, and the battery must be discharged.

If I remember correctly, turning these trimmers clockwise decreases the setting, while turning them counterclockwise increases it.


By default, the output is set to 12 V. That makes sense; it's the most commonly used voltage in these systems.


But the board lets you set the output voltage anywhere from 1.6 to 28 V. And remember, this voltage doesn't depend on your battery's voltage or the power supply's voltage. The power supply's voltage only needs to be matched to the battery voltage, and the battery voltage can range from 6 to 21 V. Yes, and even then, the power supply voltage needs to be a few volts higher than the battery’s end-of-charge voltage. But there’s nothing stopping you from using a 24 V power supply with a 7.4 V battery.


By the way, I checked that charging works correctly.
First, it charges at roughly the set current (which drops slightly toward the end of charging). Then, for the usual reasons, the current gradually falls. Once it drops below a certain threshold (about 1/10 of the set value), the board stops charging and turns on the red LED. The battery is now fully charged and disconnected from the charger.


One important detail: when running from the battery, the board does not feed power back to the power supply input. This matters because power supplies often have a load resistor across their output to ensure stable operation. When running from a battery, that resistor can drain it. That’s not a problem here.


Since the board doesn’t care whether it’s powered by a power supply or a battery, I ran the load test and efficiency measurements from a power supply with a wide adjustable voltage range. Just to be safe, though, I connected it to the battery input, since that’s where efficiency matters more—and this also disables the charger.


First, I checked how the output voltage varied with load current. The output was initially set to 12 V, powered from a 4S battery.
At a low load current, the voltage sagged slightly, but after that it stayed very stable.


And with a 24 V power supply, there was a small initial spike to 12.13 V for some reason. I don’t remember why, but after that it behaved just like it did on battery power: the voltage sagged and was almost independent of load current. So it may make sense to set the output a little higher to begin with, or adjust the voltage with at least a small load connected.



I measured efficiency at three load levels over an input voltage range of 6–21 V. That is, 6 V corresponds to a fully discharged battery in the minimum configuration, while 21 V is a fully charged battery in the maximum configuration. The output was set to 12 V.
The load current was 4, 8, and 10 A, so I even exceeded the specified 8.3 A.

A couple of caveats: at 8 A, the board could operate down to 7 V, while at 10 A it could go no lower than 8 V. But that’s to be expected: for example, 12 V at 10 A is 120 W, so at an input voltage of 8 V the current would be 15 A, even without accounting for efficiency.
Second, at a load current of 8 A and an input voltage of 8 V, the output voltage dropped slightly. The same thing happened at a load current of 10 A, but the drop started with an input voltage of 10 V.
Overall, the board’s efficiency is decent, in my opinion, especially considering the losses in the fuse, switching section, and wires to the power supply.



This is what the output voltage drop under overload looked like: 8 A on the left, 10 A on the right.


I checked the thermal performance when powered by a 4S battery. Since it doesn’t have a very large capacity, the test lasted about 10 minutes.
1. 50 W load
2, 3. 100 W load; top view and the converter’s outermost transistor.

Looks great to me.


After that, I ran a much more interesting test: the load was about 100 W, the input voltage was 9 V, and the test lasted over half an hour.
This test shows what happens when the board is powered by an almost discharged 3S battery while running at maximum load. I ended the test myself; the board didn’t stop it :)


1. Temperature near the inductor: about 106 °C
2. Transistor at the edge of the board: 93 °C.
3. Transistor closer to the center of the board: 91 °C.

That seems perfectly fine to me too, but I wouldn’t use the board like this if it were installed in an enclosure. Otherwise, you’d need to ensure at least some airflow.


To test the charger, I had to use a larger battery, though with the same voltage.
I set the charging current to 5 A. Near the end of charging, with the battery at 16.7 V, the input current was about 3.55 A, for a power draw of 85 W.
Again, no problems: the charger handles this current easily. That’s down to its good efficiency. The battery was receiving 16.7 × 5 = 83.5 W, while the input power was 85.3 W, so charging efficiency was 98%.


I measured ripple in six operating modes. On the left, at 4 A; on the right, at 8 A. I also tested three input voltages—9, 12, and 16 V—with the output set to 12 V, so The converter worked as both a step-up and a step-down converter.
The maximum peak-to-peak ripple was 240 mV, but only because of the spikes; the main ripple was no more than 50–100 mV, which is perfectly normal.


A couple more words about adjusting the board’s cutoff and turn-on voltages. I found the photos later, so I didn’t insert this in the middle of the review.
As I wrote, I used a 4S battery, so I set the undervoltage protection cutoff threshold to 12 V and the turn-on threshold to 12.8 V. The corresponding values at the test points were 3 and 3.2 V. As a reminder, set the required voltage at the test points divided by 4.
The actual cutoff occurred at 12.03 V, so the error when setting the voltage this way is quite small.
You can also get an indirect estimate of the board’s power consumption after cutoff here (last photo). The description claimed around 200 µA. At the moment, I can neither confirm nor disprove that, since the 2 mA shown by the power supply can’t be accurate. Subjectively, though, the board’s consumption in this mode seems closer to 1 mA.


Update.
I bought a couple of boards for one of my orders. I only needed one, but two worked out cheaper. I placed the order on May 5 and received them on May 14, just 9 days later. I picked them up from a Nova Poshta parcel locker.
The packaging gets 5 out of 5: it was wrapped thoroughly, so one board couldn’t damage the other, and the box was also lined with foam.


The boards are the same as the one in the review and were bought from the same link. As far as I can tell, this is the lowest price for them on AliExpress.


But I wouldn’t have added this update if it weren’t for one small detail. The charging chip on one board hadn’t been sanded off properly, and some of its marking was still visible. After experimenting with alcohol, I managed to make out part of the marking, so I started looking up the chip. It turned out to be a BQ2461. There are actually two chips in the series, the BQ24610 and BQ24617. I didn’t look into the differences; that’s a minor detail.


What’s more interesting is that this is essentially a full-fledged UPS. The chip is responsible for charging the battery correctly, switching power to the load, and connecting the power supply.
So it turns out that this chip and the universal converter chip together make up the UPS, while the microcontroller most likely monitors the battery voltage and handles the shutdown timer.



Conclusion.
Personally, I think this board deserves a solid 4+. I’ve taken off a little for two issues:
1. The price. Almost $40 is still quite a lot.
2. There’s no way to disable auto-start while still allowing the board to start automatically when power is supplied by the power supply.

Otherwise, the board is simply excellent: high efficiency, rated power output—and actually even more—lots of adjustments, a capable full-featured charger, the ability to control a PC motherboard, and a compact size.
Honestly, if it were a little cheaper, say $25–30, and didn’t have the auto-start issue, I’d say buy it without hesitation.

As for the drawbacks, it’s a matter of perspective: price is relative, and auto-start only comes into play if the battery is completely drained. Plus, there’s a workaround: set the auto-start threshold higher than recommended. For example, with a 12 V battery, you can easily set it to 12/13.2 instead of 12/12.8. Then, when mains power is applied, the voltage will quickly rise to that level and the board will start properly, but it won’t start by itself after a power outage. It also depends a lot on your battery. If it has a large capacity, it will discharge at a relatively low current (compared with its capacity), so the voltage sag will be smaller than with a weaker battery.

That’s all from me. I hope you found the review useful. As always, I’ll be glad to hear your comments, and thanks for providing the board for testing.