A Compact Uninterruptible Power Supply for Access Control Systems

Many computer users are used to uninterruptible power supplies being fairly bulky things that usually sit somewhere under the desk and occasionally beep when the power goes out.
But small devices also need backup power, and these little UPS units are made for them.
Of course, you can use a regular "computer" UPS, but there are a few things to keep in mind:
1. Not all of them can handle low-power loads
2. The efficiency of this kind of setup is usually about as good as a steam locomotive.
3. The battery sits next to a hot transformer, which can significantly shorten its service life.


By the way, regular UPS units can sometimes cause problems for computers that need to run all the time, such as those in a video surveillance system. This isn't directly related to the review, but I'll show you a situation where a UPS can actually do more harm than good.
1. Set the computer to start automatically when power is restored and shut down properly when the battery runs low.
2. The power is on, everything works
3. The power goes out, and the computer runs on battery
4. The battery runs out, and the UPS sends a shutdown command to the computer.
5. The automatic shutdown process starts, and then the power comes back on.
6. The computer shuts down automatically, but the power is on.
7. That's it, we're sunk.


For almost the entire time I was dealing with this product, I kept thinking, "Well, they can do it when they want to." But that thought went away when I drew up the device's circuit diagram.

But I'll take things in order. First, let's have a look.
My item arrived in a bag with a zip closure, containing a cardboard box and a bag with the remote controls.


The package includes:
1. Uninterruptible power supply
2. Four remote controls
3. Receiver board for the remote signals
4. Manual.

The device is sold in three configurations:
1. With one remote control
2. With two remote controls
3. With four remote controls

I figured that if I was going to order one, I might as well get the full package. And if you're buying one, this option seems like the better deal to me, especially since remotes have a way of breaking or getting lost.


The manual is in Chinese and English. In addition to the description, it shows wiring diagrams for different configurations.


The receiver module. A small board with a spring antenna. There's not much to say about it, except that it works :)
You can set your own code on the board. If you change the code, you need to set the same code on the remotes, too.
The code is set using solder bridges. There's usually no need to change it, but if pressing a button on your remote opens your neighbor's electric gate or makes them suddenly come home, you'd better change it :)))


I think many of you will recognize the remote's design, though this one has just a single button—and it's red, just like in the movies.
It has a telescoping antenna, but it actually works pretty well with the antenna retracted, too.
All in all, it's an ordinary remote, nothing unusual.


The remote is very easy to open: remove three small self-tapping screws and you're inside. It runs on a standard 12 V battery, the kind I think car alarm users know very well. Though lately, I see thin lithium batteries more often.


There are also solder bridges on the remote's board. For the codes to match, the bridge configurations on the remote and receiver must match as well.

Personally, I think using a remote control to operate doors is a pretty bad idea—or, to be more precise, it creates an unnecessary security vulnerability. So it's best to use this only at non-critical sites.


The remote and the other accessories aren't particularly interesting, so I'll wrap up their description and move on to the part that actually caught my attention.
I'll say right away that when I picked it up, it felt like a fairly high-quality product, and the workmanship is very neat.



Besides the familiar photo with a matchbox for scale, I'll include a picture from the store to give you a better idea of the size.


The unit consists of an L-shaped aluminum chassis that also serves as a heatsink, plus a transparent cover that protects most of the components.


A long terminal block takes up almost the entire front side. I'll explain what each terminal is for below.
There's a trimmer on the left, which adjusts the relay switching delay.


1. On the left is the power input connector. A short cable with the matching connector is included. I used Wago connectors for the tests, but I think a screw terminal block would have been a better choice here.
2. On the right is the connector for the radio module.
3, 4. Installing the module is straightforward, except that it won't seat fully in the connector: the trimmer capacitor and inductor will hit the relay housing. This won't affect functionality, but the design clearly doesn't leave enough clearance.


Now I'll move on to how to connect it, and explain what this device can do.
Essentially, the whole assembly consists of a power supply, a battery charger module, and a circuit for controlling an electric lock.
Since I'm describing the connections, I'll focus on the control board.
Basically, it's just a command hold timer. In other words, it has:
1. An input from the remote control
2. Inputs for external commands (Push1 Push2), which are driven either by 12 V or by 0 V, depending on the input (active-low or active-high).
3. An optically isolated input, which accepts a 3–12 V signal, for example from an Arduino or simply a battery.

The output is a changeover relay. With the right wiring, it can either supply 12 V to the load or cut off its power.

There's also a trimmer on the left for setting the hold time. In other words, you can send a brief command and the relay will stay on for a set time. The adjustment range is about 0.5–15 seconds.
Keep in mind that the circuit doesn't work as a monostable timer. For example, if it's set to 10 seconds and the input is closed for one second, the relay will turn on for 10 seconds. But if it's set to 5 seconds and the input is held for 10 seconds, the relay will stay on for 10 seconds too.

For an electromechanical lock or latch, set the minimum time and configure the output to supply power to the load when triggered.
For an electromagnetic lock, do the opposite: set the time to 10–15 seconds and configure the relay to cut off the power.

Warning: this circuit isn't designed to switch high voltages, and it's assumed that all power comes from this same power supply. The relay's common contact is connected to the common wire (the circuit's negative), and there are also protection diodes connected in parallel with the relay contacts.

The board also has a 12 V output for powering the controller and separate terminals for connecting a battery.

The diagram is very simplified. It looks like it was drawn by the seller, and it would be better to draw a separate diagram for each wiring option. If you need this information, I can provide a couple of examples.


The power and battery cables are included, but the battery itself isn't.


The housing is covered with a protective film. I recommend removing it for better cooling. I didn't do that right away, so I ran half the tests with the film still on.


The protective cover is held in place by two M3 screws. You shouldn't need to remove it in normal use, since all the connectors are accessible, and the only fuse is soldered in.


Before inspecting the board, though, I decided to take a quick detour and make some preliminary measurements. As it turned out, that was a good idea.
The 12 V output looks fine, but the tester showed over 16 V at the battery connection output.


Next, I ran a few tests with a battery and found a few things worth noting.
1. The board draws about 50 mA with no load.
2, 3. The charging current is about 100 mA when the battery voltage is 12.5 V.
4. Since I found that the charger has no idea what voltage limiting is and puts out up to 16.5 V with no load, I ran an experiment.
I picked a pair of resistors that would bring the voltage close to the end-of-charge voltage (actually, it should have been 13.8 V) and calculated the current that would flow through the battery at that voltage.
I got a load resistance of 208 Ω (300 + 680 Ω in parallel). At 14.05 V, the current would be 67 mA (14.05/208 = 0.067).

On the one hand, the charging current is very low, so it won't damage the battery. On the other hand, it's not high enough to charge it in a reasonable amount of time.
So while you might be able to live with the lack of automatic charge cutoff, charging a 7 Ah battery for more than three days is a bit much. And if you use a small battery, say 1 Ah, the lack of charge cutoff becomes a problem.


The high-voltage PWM controller IC and output diode assembly are screwed to the heatsink. The heatsink is in turn connected to the ground wire of the input connector, so it must be grounded for safety.


Unscrew the board from the heatsink and remove it. A thick protective film is applied to the underside—a plus for the manufacturer.


The input filter is almost fully equipped.


Almost textbook design:
1. Class X and Y capacitors at the input—not only those connected to the ground wire, but also across the transformer windings.

2. What's more, there's a varistor, which is extremely rare.
3. There's also a thermistor to limit the filter capacitor's charging current. I tried to make out the bridge diode markings, but it looks like their leads were deliberately bent so the markings can't be seen.
4. The capacitor itself has a capacitance of just 33 µF, compared with the required 56–68 µF, but it's a genuine branded part.


And the fact that they used a decent input capacitor wasn't a fluke. The photo on the store page shows a capacitor from a different manufacturer (Samwha), but of comparable quality.
The same goes for the output capacitors.


Almost as surprising as the filter was the use of a powerful PWM controller, KA5L0380R, which is also a genuine part made by Fairchild. If I'm not mistaken, this is the controller used in Samsung satellite receivers. The KA prefix in the part number is a reference to Samsung.
This controller is designed for power supplies rated up to 75 W and includes a wide range of protections against overheating, overload, and overvoltage and undervoltage.

But they cut costs on the output diode assembly: they used a YG902C2, rated for 10 A (I'm not sure whether that's 1 × 10 A or 2 × 10 A). Worse, it's not a Schottky diode.


The power supply has a maximum output power of 60 W, and the transformer (E30/15/11) is generously oversized. According to my calculations, it can handle 80–100 W in this circuit.
One downside: the IC and diode assembly were screwed down without any thermal paste!


A few words about the other components:
1. The feedback circuit is textbook: an optocoupler, TL431, and a compensation and output-voltage adjustment network (the adjustment is very smooth). The power supply runs silently across its entire power range.
2. The output has not only genuine capacitors, but also a substantial inductor.
3. A relay for switching the battery, and diodes to isolate the power supply and charger.
4. A timer circuit and an optocoupler for galvanic isolation of the control signal.
5. A load-control relay and a timer trimmer.
6. The layout is very dense. Not only is the board double-sided, but there are components even under the transformer. That's not ideal for an electrolytic capacitor, though.


Just in case, here are a couple of overall shots of the board from two angles.




Although the board is double-sided, components are mounted on only one side. Interestingly, they're all through-hole parts, not SMD. It feels a little unusual—reminds me of some branded devices.


Another plus: the PCB has all the necessary safety slots—not only between the board's “hot” and “cold” sides, but also between the input connector pins. A big plus for the manufacturer.


As you may have guessed, I decided to draw up a schematic for this device. I was especially interested in how the battery charging and monitoring circuit was implemented. It wasn't exactly easy work, but curiosity got the better of me :)
It helped that some component values were marked on the board itself, but unfortunately, not all the components have reference designators.


For clarity, I divided the schematic into functional blocks, shown here:
Red — high-voltage section of the power supply
Blue — low-voltage section of the power supply
Green — charger and battery relay
Orange — relay delay timer circuit.


I was interested in the battery charging and switching circuit, so I isolated it and removed the components unrelated to its operation.
Red — charging section
Blue — relay control.
Black — main power supply.
It's pretty clear that a student designed it, since charging is done through a resistor. In other words, the circuit is: an auxiliary transformer winding, a rectifier, and a resistor.
I was just as surprised to find that the battery is switched by a relay. I once showed how to build a small UPS yourself. It had no switching, but it did have protection against over-discharge, charge-current regulation, and protection against connecting the battery the wrong way. There was one drawback, though: in normal operation, the output was 14 V rather than 12.

Here, everything is the other way around. The only advantage of this design (apart from its simplicity) is that the output is 12 V when powered from the mains, and only rises to 12–13.5 V when it switches to battery power.

Speaking of switching, I think it’s better if the device either never loses power at all, or loses it for longer than the time it takes the relay contacts to switch. Some video cameras or controllers may freeze.

The circuit definitely needs some work, but not this time. As a little preview, I have another similar power supply waiting to be reviewed. For that review, I’m planning to come up with (or, more likely, I’ve already come up with) a small add-on board to fix most of the shortcomings. If you think it’s worth doing, I’ll post the schematic and drawings.


While writing this, inspecting the assembly, and drawing the schematic, I kept feeling that something about it was familiar. It feels as if several people designed the power supply: some parts are done exceptionally well, while others are downright poor.
It’s like a perfectly good car with one wheel out of three taken from a bicycle.

Or here’s another analogy :)


I’ve inspected it, drawn the schematic, praised it a little and criticized it too. Now it’s time to run some tests.
I put together a classic test bench consisting of:
1. Electronic load
2. Multimeter
3. Oscilloscope
4. Thermal imager
5. Thermometer
6. A pen and paper.


Next, I tested the power supply under load, stepping from no load up to maximum power.
1. No load.
2. 15 W (1.25 A)


1. 30 W (2.5 A)
2. 45 W (3.75 A)


Ripple only became visible on the oscilloscope at maximum power, which is an excellent result.
Just to be safe, I ran an additional test at 110% of the rated load. It got quite hot, but the power supply behaved absolutely stably.


But I didn’t stop there: I also measured the efficiency. The result was affected by the fact that one of the relays also switches on when power is applied, though I don’t think that matters much at high power.
For this test, I used a different electronic load and set it to constant-power mode, which makes efficiency measurements more convenient.


1. No-load mode with the relay switched on.
2. Battery charging.
3. Output power: 15 W (25%)
4. 30 W (50%)
5. 45 W (75%)
6. 60 W (100%)

The efficiency was:
25% load — 74.25%
50% load — 78.53%
75% load — 78.67%
100% load — 78.84%


I compiled all the measured data into a table.
A few notes about the table: I increased the load in steps, with 20-minute intervals between them. The final test lasted 15 minutes, bringing the total test time to 95 minutes.
The temperature of the diode bridge (first column) is approximate, since I couldn’t get the pyrometer into position and it also picked up the temperature of a large resistor, which was hotter. The transformer temperature shown is that of its core, as this is the most critical part.


Having measured the device’s efficiency during testing and taken a thermal image, I can now say where the excess energy goes.
The thermal image shows that the transformer windings reach the highest temperature. I suspect that although the core was chosen with plenty of headroom, they skimped a little on the wire gauge.
Unfortunately, fixing this would be quite difficult. And if you want to gain a few percent by replacing the output diode assembly with a Schottky diode assembly, that's entirely doable.
That said, I wouldn't say the power supply reached critical temperatures. The output capacitors are probably the only components worth keeping an eye on; the others are still well below their temperature limits, especially considering that the test was run at maximum power.
This power supply can be operated safely at currents of up to 4 A.


That's it. Time to sum things up.
Advantages
Quality components
Genuine rated output power
A wide range of control inputs
Remote control
Well-thought-out circuit design (for the most part)
All remotes come with batteries
Very low ripple

Disadvantages
Incorrect charging circuit
No battery over-discharge protection
Mains/battery switching.
The input capacitor has too low a capacitance

My opinion: Honestly, if you look at this device as a power supply and a lock switch, it's excellent. There's really not much to criticize. Quality components, a decent circuit design, and neat construction.
But when it comes to the device's main selling point—uninterruptible power—it's the complete opposite. The charger needs work, or more precisely, a redesign, and over-discharge protection needs to be added as well.

Overall, you can see they made an effort, but they handed part of the design work to someone who didn't know what they were doing.
I'm now planning a review of a second, similar device. This time, I intend to figure out how to do the whole thing properly.

I'll wrap up here for now. I hope you found this interesting and useful.