Repairing a small, very old low-voltage UPS

I can’t even remember when or how this little board came into my possession, but it was clearly a long time ago. It sat around for years until I happened to come across it and decided, just out of curiosity, to figure it out and repair it. Read on to see how that went.

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 maybe write about something interesting too - link to the account

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

Apparently, this unit is sold both on its own and in a case with a battery. Since it provides 12 and 18 V outputs, it’s clearly intended for access control systems, where 12 V is used for the controller and 18 V for the electromechanical lock.
Three LEDs are connected by wires:
1. 220 V present
2. Output voltage present
3. Battery charging.


The components are packed in fairly tightly, which isn’t good for devices like this: heat causes component degradation, and power supplies like these usually run 24/7 and are expected to work for years.


Judging by its condition, the board has never been repaired and still has its original components. That’s exactly why I was interested in taking a closer look at it.

1. Model BBP-30, made by Nikton-S way back in 2005. The company is still in business and has a website.
2. The input has a mains filter, a thermistor, and a spot for a varistor, but no mains fuse. I assume they intended for it to be installed in the case itself, which, in my opinion, isn’t always convenient.
3. 47 µF, 400 V input capacitor
4. The PWM controller is a TOP234Y. I have to say, that was very advanced for 2005. I was using these chips extensively myself at the time, along with the earlier TOP22x models. I really like these controllers, and back then I was quite surprised to read that some people had problems with them.
5. The interwinding capacitor is an ordinary high-voltage type, which is of course wrong. Its markings even started coming off when I tried to clean the board a little with a cotton swab and alcohol. In short, never use capacitors like this in this circuit.
6. The transformer uses an E25 core; the rest of the markings aren’t particularly relevant here.


The output section is a mess. The board is visibly scorched, but judging by its appearance, this was caused by overheating rather than a fire.


There’s dust on the back. I only noticed later that the photo came out badly, and retaking it now wouldn’t feel as authentic :)
The MOSFET at the top left provides protection against overdischarge and reverse polarity. The diode in the middle is for the 18 V channel.


An interesting choice to mount the optocoupler upside down. The clearance between its pins is clearly too small, and you can also see the dust here. Add moisture and you get a breakdown.


The heatsink finish is simply amazing :)
By the way, one of the component mounting holes is oval, presumably to compensate for any misalignment.

But what I really want to point out is that the components are in ordinary, non-Fullpak packages, mounted with insulating pads. What’s more, the heatsink extends across both sides of the unit—the cold side and the hot side.


Yes, under these conditions, a short from input to output can happen very easily, so it’s better to use separate heatsinks. You can never have too much safety.


I couldn’t find the manual for this model, but I did find one for what was probably its successor. The heatsinks are separate there, and it also gives a date of 06.03.2008, so the 2005 date on the board seems quite plausible.
What I was interested in here, though, were the technical specifications.


When I tried powering it up, it made a few feeble attempts to start. This is clearly a typical problem: the capacitor in the PWM controller’s power supply circuit has failed. I replaced it with a new one, and the unit came back to life. In general, it’s best to use the highest-quality capacitors you can here.


The second problem: the transistor in the battery charging circuit had burned out. The charging current is regulated by a linear regulator, with a diode also connected in series with it. I consider this circuit design a dead end. I usually connect the current regulation loop to the power supply’s main feedback loop: less power loss, more stable charging current, and the battery charges faster toward the end of the cycle.

The wires to the LEDs had literally turned to plastic; the insulation was cracked.

Most likely, the battery failed, the regulator overheated, and that was the end of it.


When I was buying parts for other repairs, I also picked up a transistor for this power supply. In this case, it’s a BCP69


First, I shortened the wires to the LEDs and soldered them back as they were originally. It’s best to label the wires and take a photo before doing this.

1. Remove the old transistor. The copper pour it was soldered to comes off with it.
2. Remove the old diode as well. Now we’re left with a dead transistor, a scorched diode, and the copper pour.
3. Trying to clean the board didn’t help.
4. Take a milling bit and clean up the mounting area.
5. Coat the work area with protective varnish.
6. Solder in the new diode, feed its lead through the original hole, and bend it into an O shape to form a new contact pad. I used a regular 1N4937 diode, although a Schottky diode such as the 1N5819 would work better here. But I was restoring the circuit to its original state, and the original diode was a 1N4007.


Solder the new transistor in its original position.


Check that it works: the power and output-voltage indicator LEDs are lit.


I also replaced the battery terminals; the negative one was in really bad shape.


This photo isn’t directly related to the power supply, but it is related to the battery I used to test it. You can see that the battery holder cracked just from sitting on the table. I wouldn’t say the cells were packed in tightly; the plastic itself must just be lousy :(


I tested the charger using an electronic load in CV mode. This is important: you can’t test it properly with a load in CC mode. I’ve written about this more than once, but I’ll mention it again anyway.

1. At 12 V or below, the charging current is 340–370 mA.
2. At 12.8 V, the current drops to 270 mA, due to the relatively large voltage drop across the regulator.
3. At 13 V, the current drops to 220 mA, and the charging indicator LED almost goes out—which isn’t quite right, since the battery is still a long way from fully charged.
4. At 13.8 V, the charging current approaches zero.

Broadly speaking, the charging conditions are met: the battery voltage stays at or below 13.8 V. But the charging current at the end is veeeery low, so the battery will take a long time to charge. That’s why I use a different approach to charging-current regulation.


With the battery voltage at 11.5 V, the transistor heats up to about 90 degrees. So if the battery—or one of its cells—develops a short circuit, as sometimes happens, the transistor overheats.


This power supply is rated for a continuous output current of 2 A and 2.5 A for short periods, but unfortunately I could only get 1.75 A. At 1.77 A, the power supply went into protection. In practice, it turned out that it couldn’t sustain even 1.7 A for long; the protection trip current drops as the unit heats up.
The power supply has an 18 V rail, but it’s only loaded very briefly, so it has no effect on the thermal performance. By the way, in this design I’d probably use a DC-DC converter based on the 34063 chip, powered from the 18 V rail, to charge the battery—but that’s just my personal opinion.


I ran it for a while at 1.7 A, but the power supply started going into protection before it had even warmed up much.


In the end, I had to test it at 1.5 A. Here’s what happens to the temperatures.
1. Overall view. The thermistor on the right is getting very hot.
2. The power supply’s output diodes. There are two diodes connected cathode-to-cathode here: one from the power supply output and one from the battery. And here’s my question: why is the power supply’s main diode mounted on a large heatsink, while the diode used for isolation—which carries essentially the same current—is not only in a smaller package but has no heatsink at all? The result is a temperature of 102 degrees even at 1.5 A, with the board simply lying on the table rather than installed in a closed case.
3. The PWM chip also runs quite hot, at 103 degrees, which is somewhat strange.


The power supply also has a strong temperature dependence: its output voltage is 14.4 V when cold, then drops to 13.8 V and barely changes after that, eventually falling to 13.65 V.


At this point I wondered whether the power supply was shutting down early because of old capacitors. This is a common cause, especially since the unit is 18 years old.
1, 2. The original capacitors on the 12 V output and the power supply input: 2 × 1000 µF and 47 µF, respectively.
3, 4. New capacitors: one is the same type as the one installed, with the same 1000 µF, 35 V rating and from the same manufacturer; the other is a 22 µF high-voltage capacitor.

The output capacitors appear to be in perfectly good condition: the pair has an ESR of 81 mΩ, or 160 mΩ for one capacitor, while the new one measures 0.74 mΩ. But that’s not really all that critical. The input capacitor is another story: its ESR is 7 Ω, compared with 2 Ω for the new one.


But! I found a used capacitor from the same manufacturer, with the same specifications and the same type as the one installed in this power supply. Even that one had an ESR of 2 Ω, compared with 7 Ω for the original.
And yes, a reminder: ESR is often much more critical than capacitance!


For the experiment, I made two changes:
1. Connected a 1000 µF, 50 V Samwha RD-series capacitor in parallel with the output capacitors.
2. Replaced the input capacitor with an identical used one, but with noticeably better specs.

And the result was... it shut down at an even lower current. In other words, when cold, the power supply used to shut down at around 1.76–1.77 A; now it shut down at 1.72–1.74 A. It ran at 1.7 A, but still tended to shut down at a lower current as it warmed up.

So I got the exact opposite of the result I expected, and I think it has to do with the input capacitor.


Conclusions.
In my opinion, this repair didn’t work out. Or rather, that’s not quite right: I restored the power supply to its original condition, but I’m not happy with it.
In the end, we have a perfectly functional UPS: it charges the battery, powers a load and displays its status. But it has a number of shortcomings: safety isn’t great (I later replaced the interwinding capacitor), its power is below the rated value, the charging current is low at the end of the cycle, and the output isolation diode gets noticeably hot.

Still, I won’t blame the designers. First, this was a long time ago, and I don’t know why they chose this design. Second, it looks like they later fixed some of the shortcomings.

That’s all for today. I hope you found this useful, and as always, I’d be glad to hear your comments. As for this board, if anyone wants it, I can give it away—but you’ll have to pay for shipping.