Panasonic air conditioner power supply repair, review added

About four years ago, I wrote an article about repairing an air conditioner control board. Now someone with exactly the same board and exactly the same fault got in touch with me. We initially talked about the quirks of service centers, but one thing led to another, and the conversation turned to whether I could repair the board. Well, why not? I said I could, got the board, and here’s the report.

A reminder: to keep up with new topics on the blog, I recommend following me on Instagram. I’ll post notifications about new articles there, and maybe write about something interesting too — link to my account

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

So, this is the article I wrote back in 2019. To be honest, I’d already forgotten about it.


If you read the article, or even just look at the photos, you can see that yes, it’s the very same board. The symptom: it won’t turn on.


Interestingly, just like last time, the varistor and fuse are both intact, while the high-voltage transistor has failed. I know why this happens, but an intact varistor usually means there wasn’t a major overvoltage at the input. Then again, this one is rated for 510 V peak.
The board’s owner had clipped out the transistor.


There are two resistors on the board that run hot all the time; the board has even darkened around them. The boards in the previous article had exactly the same problem.


The underside, rotated to make the component designators easier to read. You can see that the transistor has clearly overheated.


The problem area. You can see the burned components, including a 1 Ω resistor (presumably) that burned out before the fuse. The fuse is shared by the entire system and is rated at 3.15 A.


Since this is the third identical board that’s come my way, I decided to trace out the circuit and mark the components that had burned out.
I also measured the ceramic capacitors. The one value I’m not sure about is the current-sense resistor; based on my experience with similar power supplies, it should be around 1 Ω. The values of the other components were reconstructed from the photos in the previous article.
The VT1 transistor is unidentified. It’s marked WG, which points to a “digital” transistor that’s unlikely to be used here. But an internet search also turned up the KTC3876 transistor, which has the same WG code marking.

I redrew the schematic. Now I’m waiting for it to show up on some random website in a few days, with a caption like “Panasonic air conditioner power supply schematic” or “download air conditioner schematic for free” — or like this: the schematic, a chunk of text, and that’s it. Just a reminder: I’m fairly relaxed about people copying material from my site, but there’s at least one non-negotiable condition: include an active link to the article.



I desoldered the faulty components, removed the excess solder, and wiped down the board.


A pile of burned-out components and their replacements. By the way, the small diode in the previous power supplies was still good; I still need to check the Zener diode.
I used a BC817 transistor.


I soldered in the components and cleaned the board.


The transistor is a trickier matter, and it’s one reason I don’t like power supplies built around self-oscillating circuits. The transistor’s gate capacitance can make a noticeable difference here, so you can’t just install a “more powerful” one. The original 2SK2865 had a gate capacitance of 380 pF, a voltage rating of 600 V, and an on-resistance of 4.2 Ω. Previously, I used an IRFU420 as a replacement: 360 pF, 500 V, and 3 Ω.
First, I found an STK0260 at home that I’d desoldered from something: 290 pF, 600 V, 4.7 Ω. It’ll do for testing; if it burns out, at least I won’t mind.


But the power supply started up without any problems and delivered the required 5 V and 12 V.


Last time, the failure of the high-voltage section caused the Zener diode on the 12 V rail to fail. That’s the second reason I don’t like self-oscillating circuits. Unlike circuits with PWM controllers, there’s a very real risk of the load failing because the power supply can run away, sending the output voltage higher and higher. A typical example is the standby power supply in cheap ATX units: dried-out capacitors on the 5 V rail can cause the PWM controller, powered from a different winding, to fail.

Interestingly, the Zener diode turned out to be rated for 15 V, even though the board clearly specifies 12 V. There is also a 3.9 V zener diode in the high-voltage section. The third one, rated at 4.7 V, is connected in parallel with the TL431. Naturally, I checked those too. The circuit design is rather odd in places, especially the feedback circuits.



Since the power supply started up, I replaced the transistor with a more suitable one—in this case, an SSS4N60A, which I had at home. This transistor has a higher capacitance, 450–710 pF, and a 600 V rating, but its resistance is lower, at 2.2 Ω. It also comes in a larger package than the original, so it’s less likely to overheat.


I connected a load: 10 Ω on the 5 V output and 20 Ω on the 12 V output, for a total load of about 10 W, and... the power supply wouldn’t start. I removed the load from the 5 V output, but it still wouldn’t start. More precisely, it tried to start, but the output voltages were much too low.
And that’s the third reason I don’t like self-oscillating circuits: they’re more “dumb.”


Next, I tried to work through it logically.
The power supply clearly isn’t very powerful, judging by the small output capacitors. Still, I checked them just in case. They turned out to be in perfect condition, with ESR and capacitance matching new Samwha RD-series capacitors.
I rechecked the primary-side components several times and tried replacing the high-voltage transistor, but it made no difference.


Ideally, of course, it would be good to have a working board to see whether this is normal behavior.
Then I remembered that just a week earlier I’d repaired a board from an air conditioner made by the same company. When I looked at the photos, it turned out the boards were almost identical—even the overheating resistors were there :)

But unlike the old board, this one has a PWM controller, so I decided to estimate the power supply’s output power indirectly.


I found the datasheet for the PWM controller used here. Its maximum power is only 7 W. Since these are usually rated with some headroom, the actual power of this supply is no more than 4–5 W, possibly less. And since the control boards seem to be similar, the old board’s power rating is probably about the same.


After that, I found that the startup problem occurs when a resistive load is connected to the output. If you connect it after startup, the power supply works normally.
I connected a 30 Ω load. The output voltage was about 11.1 V, so the current was 0.37 A and the power was 4.1 W. I ran it like this for about 15 minutes and found no abnormal heating. The transistor only reached 60 °C, which I consider normal.
Finally, I washed the board again, coated it with varnish, and called it done.


Of course, someone might say, “If you’re going to criticize self-oscillating circuits, then, as they say, don’t just criticize—offer a solution.” So I’ll suggest one modification I sometimes use when repairing power supplies with self-oscillating circuits. Yes, it doesn’t always help, but it works in about 80% of cases, and the modification is usually not very complicated. If this were my air conditioner, this is exactly what I’d try first.

So, I used the circuit shown earlier as a starting point. I didn’t even move the components: I simply removed the unnecessary ones, added a couple of new ones, and replaced a couple more.
Remove almost all the components, leaving the supply diode. Add a capacitor before the optocoupler, replace the capacitor connected in parallel with the small transistor’s base with an electrolytic capacitor and a resistor, and replace the high-voltage transistor with a Power Integrations controller, such as a TOP242 or TOP243.

The result gives us all the “bells and whistles” of a modern power supply: overtemperature and overvoltage protection, stable operation, and higher efficiency.


In the end, the air conditioner is working again and the owner is happy, especially since it’s already gotten quite hot here in Kharkiv.
As background and feedback, the owner provided a video. I hope it’ll be useful.
By the way, there are several videos on the channel about lenses. From what I understand, that’s what the person does.


The same person also filmed the process of installing the board in the air conditioner. They were doing it for the first time, but everything went well.
I’ve embedded the video here because it’s long.



Conclusions.
It’s too early to draw any conclusions. I sent the board to the owner today so they can install it in the air conditioner and test it. Still, I’d like to mention a few things.
1. It’s unclear what caused the failure, and that matters a lot to me. In my view, you can only consider a repair complete once you’ve fixed the cause—or at least figured out what it was.
2. What we have: a working varistor and fuse, working capacitors, and a working output section, but a shorted switching transistor and burned-out components on the board.
3. We have an oddity with startup under a resistive load, though maybe that's normal here.
4. I want to say once again that I hate self-oscillating circuits because they often behave the way they want to, not the way I want them to. They’re also generally unstable and less reliable, and unlike modern power supplies, they can actually cause the load to fail. That’s where the stories about a failed power supply killing the load came from; the chances of a modern power supply killing the load are close to zero.

Anyway, I’m waiting for a report on installing the board in the air conditioner. Once I get it, I’ll update the review. That’s all for now—I hope you found it useful.