How to Change a Power Supply’s Output Voltage

I’m often asked about various things, and questions about changing a power supply’s output voltage are probably among the most common. Today I’d like to explain the subject a little; hopefully it will help someone.

To keep up with new topics on the blog, I recommend following me on Instagram, where I’ll post notifications about new articles 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

A friend brought me a couple of laptop power supplies. One needed to be modified to provide a different voltage, and the other repaired, if possible.


The power supply needed to be modified so it could work with a charger that refuses to operate at voltages above 18 V.


Originally, this was a Fujitsu laptop power supply, although it was actually made by the equally well-known company Delta.


Fortunately, the power supplies had already been taken apart, so I didn’t have to do the mechanical work of removing this one from its case.


I removed the shield and plastic cover to see what was inside. Inside was a pretty decent power supply, although it was clearly a relatively old design, with a component density well below what’s possible today. It also appears to have active power factor correction, but since I had a different task, I didn’t look into that in any detail.



The board doesn’t have anything particularly noteworthy about it, except perhaps for the large number of insulating stickers.


As I mentioned, this power supply works; all I need to do is change its output voltage. By the way, I’ve noticed many times that although these supplies are rated at 19 V, their actual output is usually around 19.5 V.


To change the output voltage, we need to look at the output stage.
There’s a catch, though. I’m often asked how much the output voltage can be changed. A change of 10–20% is usually safe; you can go further, but you’ll run into a few problems:
1. If you lower it too much, the PWM controller may not get enough voltage to operate properly, and the power supply will start restarting. The fix is to add 1–2 turns to the PWM controller’s supply winding.
2. If you raise it too much, the high-voltage transistor, output diode, and snubber circuits will operate outside their intended range, and may fail. It’s easier to buy a power supply with the voltage you need.

Some power supplies have overvoltage protection that shuts them down even when they’re still operating within their normal range. I don’t recommend disabling this protection, but you’ll need to adjust its trip threshold.


In this case, we just need to lower the voltage by about 2 V. So we need to slightly reduce the resistance of the upper resistor in the divider, or increase the resistance of the lower resistor. Reducing it is easier than increasing it.

How do you find the resistor that needs changing? If you have a simple Chinese power supply, there’s no problem at all: you’ll probably spot the divider right away.


Even in more complex power supplies, finding the divider is usually easy, especially when the feedback circuit is based on a TL431. It’s also easier to find if the power supply has an output filter inductor on the power rail: the divider is always connected after it.


But what if the power supply is more complex, with lots of components, and it’s not at all clear where to probe?


First, rule out the primary side. We don’t need it at all, so we’ll focus on the secondary side.


First, use the continuity tester to trace the positive output rail and see what’s connected to it. I’ve marked the section before the filter inductor in green and the section after it in red, because continuity testing will show that they’re connected.


Mark in blue all the resistors connected to the positive rail.


Next, check the connections of the resistors marked in blue one by one. You’ll find that only one is connected to another resistor, whose other lead is in turn connected to the output ground. And again, all you need is a multimeter in continuity mode.


What if the power supply is even more complex, with no TL431 and some obscure IC instead, plus a dual-winding inductor on the output?
The principle is exactly the same, but there are a few nuances.
First, in this case the feedback is always taken before the dual-winding inductor.
Second, you need to ignore whatever chips are on the board—a small three-pin package or a 64-pin microcontroller. It makes no difference. At this stage, all you need to find is the divider itself.


Take the board we're investigating, turn it around a bit, and start looking.


Here's another tip that can make the search easier.
Most of the time (especially in branded power supplies), the divider resistors are precision parts, so they'll have either a four-digit or an alphanumeric marking. Resistors in other circuits are usually less precise and have three-digit markings.
Of course, Chinese manufacturers may use ordinary resistors throughout, and a branded power supply may use precision resistors throughout. Still, keeping this in mind can make the search easier.
Also, divider resistors are often connected in parallel, or there may be a footprint next to them for a second resistor to be installed in parallel. The two resistor values are usually different, though there are exceptions.

Ordinary resistor on the left, precision resistor on the right.


Next, set the multimeter to continuity mode. Put one probe on the power supply's positive terminal and touch the board with the other.
In my case, the multimeter beeped when I touched these resistor contacts.


That shows the circuit they're connected to. This is how it's usually done, by the way: there's a thick power trace and a thin feedback trace running from a point closer to the output terminals. This is intentional, so the voltage drop along the power trace has less effect on the feedback signal.


Several resistors were connected here. Two of them are low-value, 100 Ω and 2 Ω, so we can rule them out right away. They aren't part of the feedback divider, since the divider's upper resistor is usually around 10–100 kΩ.
That leaves these resistors: the two on the left are connected in parallel, and the one on the right is on its own. We'll check both, though—maybe the robot's hand slipped and it installed an ordinary resistor instead of a precision one, or vice versa.


Get the multimeter out again and switch to continuity mode. This time, put one probe on the other contact of the resistors we've found and probe around the board with the other.
We found two corresponding resistors, each connected at one end to the ones identified earlier.


Now put one multimeter probe on the power supply's output ground and use the other to check the second contacts of the resistors we found in the photo above.
The first one is connected to ground (the multimeter beeps); the second shows some relatively high resistance. The exact value doesn't matter—the important thing is that it isn't shorted to ground.


So, these are the divider resistors: the upper ones are at the top, and the lower one is at the bottom, with a footprint next to it for a second resistor that can be installed in parallel.
As a reminder, the easiest way to lower a resistor's value is to install a new resistor in parallel with the existing one:
to lower the voltage, reduce the value of the divider's upper resistor
to raise the voltage, reduce the value of the divider's lower resistor.


The upper resistors are marked 7322, which means 73,200 Ω, or 73.2 kΩ. They're connected in parallel, so their combined value is 36.6 kΩ.
The lower resistor is a little trickier. It's marked 71B, so it's best to check a table of alphanumeric resistor markings.
71 = 536, and the B multiplier is 10, giving us 5,360 Ω, or 5.36 kΩ. These are precision resistors too, as mentioned earlier.



You shouldn't change the value too much. It's best to calculate it—for example, using a voltage divider calculator—or adjust it by trial and error. I needed to lower the voltage by about 10%, so the resistor installed in parallel can be about 10 times the original value. To lower it by 20%, choose a new resistor with a value about 5 times higher. This is a bit crude, of course, but hopefully clearer.
I had a 270 kΩ resistor on hand, so I wired it in parallel, just hanging off the board.


The voltage drops from 19.5 V to 17.5 V. In principle, I could have left it like that, but leaving a resistor dangling there is simply inelegant and not always convenient.


I rummaged through my parts and found a pair of 470 kΩ resistors. They don't have to be particularly precise here, since this is an additional resistor, and even if its value shifts a little as it heats up, it won't have much effect on the output voltage.
Two 470 kΩ resistors in parallel give us 235 kΩ, close to the 270 kΩ we tried earlier.

Solder them on top of the original resistors.


Let's check.
As expected, the voltage dropped a little, but that was perfectly fine with me.


Put it back together, connect it to the iMac, and everything works :)



The second power supply was simply faulty, but it also arrived disassembled. Thanks again to the person who brought it in.
I won't change the output voltage, although after the repair I found out that it could have been changed here too. Too bad I learned that too late; there goes another example...

I hadn't really noticed before, but now I see that the current is given in two formats on this one and the previous unit, with a period and a comma as the decimal separator. I don't know why.



Inside is the same kind of shield, which also has to be removed first. This shield is soldered at one or two points to a lead or a copper pour on the circuit board, so be careful.


There's a plastic cover under the shield, and that's where I noticed something odd.


There are signs of heat damage on the plastic. That wouldn't be unusual, except the marks are both above and below the input bridge rectifier, indicating it overheated.


This power supply is simpler and has no power factor correction, although its specs are comparable to the first one. I had already desoldered the bridge rectifier before I thought of writing this article, but it was there.



A couple of photos for the collection.



There are signs of severe overheating on the circuit board around the bridge rectifier, even though it was mounted on a heatsink. It's very strange, since the power supply is rated at only about 70 W. Let's say that, accounting for efficiency and everything else, the current through the bridge rectifier was 0.5 A. Even at 1.5 A, it would dissipate maybe 2 W, which shouldn't have heated it enough to start melting the plastic.


A non-original fuse was installed, and there was a short circuit downstream of it. But the high-voltage transistor was intact, and there was no short across the input capacitor either, which suggests the power supply is more alive than dead.
After desoldering the bridge rectifier, the first thing I checked was whether the short circuit was gone.
Then I measured the capacitance and ESR of the input capacitor; both were fine.
Next, I checked the PWM controller's supply capacitor. It measured 2 µF instead of the specified 22 µF, so it went in the trash.
The output capacitors were in decent condition, if not excellent.


I replaced the fuse and capacitor. The bridge rectifier had originally been mounted with thermal paste alone, so I glued it in place with thermally conductive adhesive Kafuter and only then installed it on the circuit board.



Power the unit through a light bulb, switch it on, and everything looks good.
In fact, the power supply also started up with the dead capacitor. At first, I just temporarily wired in the bridge rectifier, but those were intermediate diagnostic steps.


Once I confirmed that the power supply was behaving normally, I connected it directly to the mains and checked it again. No problems there either, as indicated by the output LED. Then I tested it under load and started putting it back in its case.


And this is where I remembered every bad word I know for Dell's engineers.
The power connector runs right up against the end of the input capacitor, with its contact touching the metal cover. If the unit were just 0.5 mm longer, or the capacitor were shifted by the same 0.5 mm, everything would fit perfectly. As it is, I have to literally force this board into the case.
To be on the safe side, I stuck three layers of special tape over the end of the capacitor. I don't want to risk a connection forming between, for example, the live wire and the capacitor's case.


Next, we close up the box, check everything once more, and that's basically it. To be honest, I still don't understand why the bridge rectifier was overheating, and that's a serious problem, because a repair technician's job isn't just to fix a device, but also to figure out why it failed. It's like a doctor giving a patient a pill and making them feel better, without knowing what was wrong with them in the first place or whether it has actually gone away.


Of course, if you want to change the output voltage of your power supply, there's a 99% chance the circuit will be different. But the general principles shown in this article will still apply, and I think you'll be able to do it if you pay close attention.
Key points:
1. The divider resistors are usually precision resistors.
2. The divider resistors are usually high-value: around 10–100 kΩ for the upper resistor and 1–10 kΩ for the lower one.
3. The feedback divider is connected after the filter inductor, but before the two-winding inductor.
4. It doesn't matter where the divider is connected; don't worry about it at all.
5. Don't make a drastic change. When adding a new resistor in parallel, choose one with a resistance about 10 times higher than the one already there.

That's all for today. I hope this was useful, and as usual, I look forward to your comments—especially if you have any tips of your own for this situation.