Repairing an SVEN SB-550 power supply by replacing the PWM controller

Another power supply has come my way, this time from an SVEN SB-550 active speaker system. The repair is fairly routine, but still interesting in its own way, and involves replacing the PWM controller with a non-original part.

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

To get ahead of things, in this case the power supply could have been replaced with any commonly available unit, but that would have required some modifications, so restoring the original was easier.

So, the power supply is installed in an SVEN SB-550 active speaker system. It’s a 2.1 system consisting of a soundbar and a subwoofer. The soundbar has its own external 18 V, 3 A power supply, while the subwoofer has one built in. The subwoofer connects over a 2.4 GHz radio link.


According to the person who brought me this power supply, something happened at some point: the subwoofer made an unexpectedly loud noise and shut down.
The same person found that the power supply had failed, checked the components, and replaced the burnt-out shunt and PWM controller, but that didn’t help—the power supply wouldn’t start, although diagnostics suggested it was making attempts. More on that later.


An online search showed that this is a common “ailment” with these power supplies.


The power supply looks rather modest from the outside, and even familiar to anyone who repairs household appliances.


The PWM controller is a non-original part, and there are also signs that the fuse has blown.


There are signs of a repair attempt on the underside: a temporary shunt resistor. The board is clean, though—thanks again for that.


I connect the power supply through an incandescent bulb installed in place of the fuse. It uses a standard power connector, so I’m using an equally standard power cable.
The voltage on the PWM controller’s supply pins is fluctuating, just as I was told: it cycles between 10 and 16 V, which usually indicates that the controller is trying to start. The process should look like this:
1. The controller’s supply capacitor charges from the input capacitor through a resistor chain.
2. When the voltage reaches a certain level, the PWM controller should start.
3. Once started, the controller begins switching the power MOSFET.
4. Once the power MOSFET starts switching, voltage appears on the auxiliary winding, and the controller is then powered from it.
5. If the PWM supply capacitor doesn’t have enough capacitance and the controller shuts down, the cycle repeats.

Most often, a power supply fails to reach normal operation for one of several reasons:
1. The auxiliary supply circuit fails. There’s often a small 10–20 Ω resistor there, and it burns out.
2. A diode in this circuit fails short, though this is rare.
3. The PWM supply capacitor dries out, so the energy stored in it isn’t enough to transition to normal operation.
4. Problems in the secondary circuit: a short circuit or a failed output diode.



1. Let’s see what’s happening at the PWM controller’s output. Nothing. Of course, checking with a multimeter isn’t entirely conclusive, but even so, there should be some activity or spikes. Here, there was absolutely nothing.
2, 3. Measuring resistance with the power supply switched off showed an almost short circuit in the control circuit, on the PWM controller side: the resistance at the gate was higher by the value of the gate resistor.
4. At first I thought the controller’s output was shorted, but no. If you thoroughly discharge both capacitors (the input capacitor and the PWM supply capacitor), the resistance rises sharply. That means the PWM controller itself is holding the output low. That’s better, although it doesn’t prove the controller is working properly.


The original controller in this power supply was a JT3027; it was in a little bag with the power supply.


But the power supply came with an SG6849 controller installed, and there’s a small catch here. The JT3027’s direct equivalent is the SG6848, although the difference between the 6848 and 6849 comes down to pin 5. The JT3027 and SG6848 connect an external resistor to this pin, while the SG6849 marks it as NC, i.e. as if it were disconnected.
Formally, it shouldn't affect anything with the SG6849, but just in case, I disconnect it.
Unfortunately, that doesn't help. What's more, when I checked the voltage on the optocoupler transistor, I saw that it wasn't rising, although it fluctuated slightly. It seemed like the optocoupler had failed, but no: even if I remove it completely, nothing changes, and the optocoupler itself is fine.
So the controller isn't entering normal operating mode, and the voltage on the FB (FeedBack) pin isn't rising. Not good.


One last test: I power the PWM controller from an adjustable power supply, with the current limit set to 100 mA. The controller definitely can't draw any more than that.
I gradually raised the voltage while watching the current. Everything looked fine—the current was close to zero. But once the voltage exceeded 16–17 V and the controller started up, it immediately became clear that it was drawing considerably more current than expected.
The current draw from the supply was 90 mA, and it was clear that the power supply hadn't entered CC mode; the controller was simply drawing that much.

Measuring the controller's temperature put an end to any further investigation: it started heating up quickly, which most likely means the high-side transistor in the output stage has failed.


The shops are closed, and it doesn't make sense to look for a single chip in local stores—the delivery cost would ruin everything. Besides, they often have a minimum order value, and I had nothing else to add to the order. So I decided to replace it with something similar.
I rummaged through my parts stash and found plenty of PWM controllers, but they all had significantly different pin layouts, and the board would have ended up looking like a squid.

Then I suddenly remembered I had a power supply lying around https://www.kirich.blog/obzory/bloki-pitaniya/904-para-blokov-pitaniya-moschnostyu-60vt-na-napryazhenie-12-i-24-volta.html that had already donated its transformer to my soldering station.


The PWM controller is still there, and all the other components are too—well, except for the transformer. That's good news.


This one uses an NCP1251, which has roughly the same surrounding circuitry—in other words, it barely needs any external components at all :)
Its operating frequency is also about the same: 65 kHz, compared with 60 kHz for the JT3027.


Even better, it's essentially pin-to-pin compatible with the SG6849, which is itself equivalent to the SG6848, an alternative to the original JT3027. So in this case, we're using an equivalent of an equivalent :)

One thing to clarify: it's pin-compatible with the six-pin SG6849 package; the eight-pin version has a different pin numbering.
NCP1251 on the left, SG6849 on the right


But the pin numbering isn't different for no reason. The PWM controller in this package and with this pinout is actually fully compatible with the DIP package too. Why? Take a look at the first and second photos. The only difference is that if you try to install it on top of the board, you have to put it in upside down. But it fits normally on the underside, and you could even make a universal board.
1, 2. Pin compatibility.
3. Cut one trace—the board wasn't designed for this package, after all.
4. Solder the two outer lower pins where the original resistor was, and connect the left and middle upper pins to the cut trace so they land on either side of the cut. I had to solder jumper wires to the middle lower and upper right pins, but I don't think that's a big deal.


Unfortunately, the replacement isn't quite as straightforward as it might seem (who would've guessed?). The NCP1251 has a pin for the OPP/OVP network, which provides overload and overvoltage protection.
And there are some oddities here. The OPP circuit isn't really OPP: the output power limit works whether or not the circuit is installed. But with the circuit in place, you can make the power supply start up with one power limit and then reduce it after startup. The difference is set by the divider and can be up to about 33%.
Put simply, the original OPP protection trips when the voltage across the shunt exceeds 0.8 V. Adding a divider lets us lower this threshold by introducing a negative offset—in effect, subtracting a set value, say 200 mV, from the original 800 mV.

To be honest, I never figured out why. The only thing I can think of is handling “heavy” capacitive loads: for example, the limit could be 4 A at startup and drop to 3 A once the supply is running. The circuit would keep the PWM controller from tripping its protection—or make it trip less often—during startup. A sort of “soft start” in reverse. But honestly, I didn't try too hard to understand how this circuit works; it was enough for me to know I could do without it here. Overall, it looks like this circuit is only needed during startup and has no effect afterward.

The second function, OVP, is genuinely useful: it makes the PWM controller shut down if the input supply voltage is too high. But even then, it has to at least attempt to start before it can detect the problem. Proper protection should block operation completely.


I removed the input capacitor and replaced it with a smaller one. I temporarily connected a 22 kΩ resistor from the OPP/OVP pin to ground, just to keep the pin from floating, and tried starting it up.


It started without any problems, so I put the input capacitor back, replaced the current-sense shunt resistors, and connected the OPP/OVP pin to ground through a 1 kΩ resistor. That value is shown in the datasheet examples, although it really doesn't matter in this configuration.
I also changed the transistor's gate resistor from 47 Ω to 22 Ω and added a 4148 diode in parallel to speed up transistor turn-off.
I also added an integrating network to the current-sensing circuit: a 1 kΩ resistor and a 200 pF capacitor.

I left the original startup resistors in place, with a total resistance of 1.5 MΩ. I replaced the PWM controller's supply capacitor with a 47 µF one. Together, these changes resulted in a long initial startup. According to the datasheet, the original startup resistor should be 200 kΩ, so you could replace it, or just leave it as is.


And here's the power supply in operation. The main output is 25 V, and the auxiliary output has a 5 V linear regulator. That's why this power supply can easily be replaced with any other 24 V unit, with a linear regulator added as well.


Replacing a PWM controller, especially with a non-original part, is a significant modification to the circuit, so it needs to be load-tested.
I ran the test in stages, for about 15–20 minutes per stage. The power supply's output voltage doesn't sag; the difference you see is because the multimeter was connected with short wires, but alongside the load.
1. No load
2. 1.0 A
3. 1.5 A
4. 2.0 A
5. 2.5 A
6. 3.0 A

As you can see, the power supply can deliver up to 3 A, although I only ran the longer tests up to 2.5 A. I'll explain why below.


Board temperature. There's no point showing the results at 1 A, so I'll show 1.5, 2.0, and 2.5 A. Judging by the output diode temperatures, the power supply has reached its maximum power and continues to operate correctly. The transformer and high-voltage transistor temperatures are still within the normal range.


At 3 A, the power supply ran for a while before the protection kicked in. A subsequent test, with the current increased gradually, showed that the protection trips at 3.4 A. But apparently this depends on the duration, temperature, or presence of brief current spikes.
In any case, everything looks fine to me so far.


After all the work, I coated the area around the PWM controller with varnish—not so much for insulation as to secure it mechanically to the board and protect it from damage.
The arrows mark the areas I worked on.
1. Large yellow arrows: replaced components
2. Large red arrows: added components
3. Thin red arrows: places where I cut the traces.


The power supply looks a little unusual now. You might think it doesn't have a PWM controller :)


And, of course, some conclusions.
Naturally, this kind of replacement only makes sense as a last resort. Although, no—the last resort, in my view, would be replacing most of the high-voltage section with something like a TOP244Y. In that case, almost everything has to be removed, and substantial circuit changes are needed too. For example, the optocoupler has to be connected not to ground but to the auxiliary supply circuit. I've done that kind of repair a few times, though, and it worked.

In any case, there are plenty of situations where you can replace one PWM controller with another, even if they seem completely different at first. The main thing is to have at least a rough idea of what you're doing.

That's all for now. I hope you found this information useful.