/       /       /    Mean Well NET-35 or a puzzle from the manufacturer
Support the project on Patreon


Mean Well NET-35 or a puzzle from the manufacturer

$28.35
Go to store
The other day, someone brought me a power supply to repair. Well, not exactly for repair: it had already been repaired and was working, but according to the person who brought it in, it was behaving strangely. The result was a little unexpected, and that’s what I wanted to talk about today, because I think anyone could run into this “quirk.”

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

The backstory...
One of my readers got in touch and asked for help with a power supply. It was installed in industrial equipment, failed, and was repaired by its owner. But according to the measurements, it was putting out not the specified -5 V, but around -15 V—three times as much. After the repair, they tried putting it back into service. It ran for about half an hour to an hour, then the equipment shut down with an error. The problem seemed strange, so to speed things up I suggested bringing it over to me.
To get ahead of the story, I’ll say that I almost immediately suggested that if this was for a production line and the power supply was needed urgently, they should simply buy a new one. It turned out that wasn’t so easy: finding this power supply for sale here is practically impossible. There isn’t even an equivalent from another series, ordering one from China would take too long, and production was at a standstill. And I’ll say it was actually a good thing that this exact power supply wasn’t available for sale. You’ll see why as we go along.


The next day, they brought me a little box with the power supply, its power cable, and a few components.


This power supply originally comes in a perforated metal cover, but this one was missing it. As it turned out, it had been left in the cabinet because it was not only rusty, but also stuck fast.


The board shows signs of repair, as well as just how hard a life this power supply has had. By the way, the board had been cleaned after the repair—many thanks for that. I think readers remember the example of repairing a power supply from a mini-PBX.


The operating conditions are very harsh, especially in terms of humidity. You can see that the solder mask has peeled off some of the traces.


According to the markings on the board and the model number I was given, this is a Mean Well NET-35A.


According to the datasheet, this model has +5 V, +12 V, and -5 V outputs. That’s another reason these power supplies are hard to buy: most have one output voltage, less often two, and three is very rare. Of course, the particular combination of voltages and load currents matters too.


I quickly checked the original capacitors that were in the box. They’re not like new, but they’re in perfectly decent, working condition.


The box also contained a schematic for a similar power supply, along with a section of the output circuit redrawn by the person who brought this one in.


The schematic wasn’t redrawn quite correctly, but that wasn’t really critical for me. However, while checking continuity, I spotted what had confused the owner during the repair.
On the left is the typical winding connection for a power supply with three output voltages; on the right is how it’s done here. I can’t say I studied exactly what was going on, but that’s what I found when checking continuity.


We put the components back, connect a light bulb and the power, and see what we get at the output.
And the output is +5 V, +15 V, and -15 V, which is nothing like what a NET-35A is supposed to produce. You could say that +15 V is really +12 V, just too high—this can happen when the capacitors on the main output (here, +5 V) have dried out. But -15 V instead of -5 V blows that theory out of the water, especially since I checked all the capacitors right away.


This suggests it’s either a NET-35B or a NET-35C. Knowing how a switching power supply works, I can say that this kind of voltage spread is technically impossible: the voltage ratios are tightly determined by the turns ratio of the secondary windings. Some variation is acceptable, but only a little—not several times over.

At this point, I suspected it was actually a 35C. So I called the owner and asked where the information that it was a 35A had come from—the board or the case? It turned out the model number had been read from the circuit board, since the label was no longer on the case.
Then I remembered that I’d recently come across a Mean Well power supply with a board marked as a different model, and in that case the power ratings differed by almost a factor of two! That’s why we’ll assume that the 35A was simply the first power supply in the series, and that later boards used the same markings but had different specifications.
Another indication that the third channel isn’t -5 V is that it had 35 V capacitors. There’s simply no reason to have that much headroom.
Also, applying a small load to the -15 V channel didn’t cause the output voltage to sag, so it wasn’t being raised by spikes or ripple.

So, at this point we’re assuming it’s a 35C, though it could also be a 35B. The tests should tell us more.


But before testing, I want to get the circuit board back in shape, so I desoldered some of the components. There was also some thermal paste residue around the power transistor, which could have led to trouble in dusty, damp conditions.
The snubber resistors were disappointing, and I didn’t have anything to replace them with, so I had to clean them thoroughly. I was also surprised to find two 47 kΩ resistors in parallel in the snubber, for a total of about 23.5 kΩ, which is quite low.


I also desoldered the board beneath the input X capacitor, checked it and cleaned it, since there were scorch marks nearby and beneath the input filter capacitor.


I connect the light bulb again and check. The filament is clearly glowing, which is normal for this power supply because it has fairly low-resistance load resistors on the output. Overall, I got the impression that this power supply was pushed pretty hard: the snubber and output load resistors have low values, which usually isn’t necessary.


Since the power supply started up correctly, we can move on to load testing—this time, of course, without the light bulb.
The loads were:
+5 V channel — electronic load (rated current up to 2.5 A)
+15 V channel — 10 Ω resistor, calculated load current 1.5 A, power 22.5 W at 15 V (rated current up to 1.5 A)
-15 V channel — resistors with a combined resistance of 30 Ω, load current 0.5 A, power 7.5 W at -15 V (rated current up to 0.5 A)

But even with no load on the 5 V channel, I got low output voltages.


All right, I disconnect the load from the -15 V channel, and the +5 V channel returns to normal.
With no load on the -15 V channel, I load the +5 V channel to 1.5 A and get a voltage drop again.

So in practice, we’re getting about 27–28 W at the output, since trying to load it to 30 W causes the voltage to sag.


Okay, that makes sense: the overload protection resistor has the wrong value. It had burned out, and someone replaced it with one that seemed close enough.

The original resistor was 0.75 Ω. I’m adding a 2.2 Ω resistor in parallel, bringing the total resistance down to about 0.56 Ω.
The power supply immediately comes to life, and with no load on the +5 V channel, the voltages return to normal—or at least to what we’d expect from a 35C model. We now have at least 30 W at the output.


Now let’s gradually increase the current on the +5 V channel. That’s exactly why I connected the electronic load to it.
The power supply holds the voltage steady up to 2.5 A, which means we’re already getting about 42–43 W of output power, despite the rated 35 W.


When I tried increasing the current on the +5 V channel to 2.6 A, the voltage immediately sagged. So the power supply can now deliver a maximum of 42 W, or 120% of its rated power.

Based on my experience with Mean Well power supplies, I figured the overload margin was too small, so I replaced the added 2.2 Ω resistor with a new 1.6 Ω one. That gives a total shunt resistance of about 0.5 Ω, close to the standard 0.51 Ω value.
So if standard E24-series resistors were used, the original value should have been either 0.51 or 0.56 Ω. The former is more likely, which is why I used a 1.6 Ω resistor in parallel.


According to the datasheet, this power supply’s overload capacity is 110–150%. The 120% we measured is just above the lower limit, so 0.56 Ω is unlikely; the next value down is 0.51 Ω, which is what I used.


Let’s check. The power supply now works normally with current up to 3.5 A on the +5 V channel and starts going into protection at 3.6 A, giving a maximum output power of 47 W. I calculated it as follows: 17.5 (the +5 V rail) + 22.5 (the +15 V rail) + 7.5 (the -15 V rail) = 47.5 W, or about 140% overload.


Next came the usual load-testing stage. I kept the 5 V rail loaded at 3 A for as long as possible. The other load currents can be estimated from the load resistor values, but in practice the total load was about 45 W, or roughly 130% of the rated power.


During the test, I took temperature measurements to monitor how things were going and see how the power supply behaved overall.
First, I tested the 5 V rail at 2 A and 2.5 A (the ±15 V rails were loaded with resistors, for a total of 30 W). The test duration is shown at the bottom right of each thermal image. After the 2.5 A test, I loaded it to the 3 A shown above.
1. Start of the test at 2 A
2. End of the test at 2 A
3. End of the test at 2.5 A


And here are the results at the end of the test at maximum load, with a total load of at least 45 W.
1. After 25 minutes at maximum power
2, 3. Another 15 minutes later.

In my view, it passed the test. Notably, the transistor only reached 70 degrees, while the transformer stayed noticeably cooler, so there was still some headroom.


The final step: I wipe the circuit board with alcohol once more and coat it with protective varnish—one layer on top and two on the bottom. Unfortunately, all I have at home is Plastik-70, though “Urethane” would be better for conditions like these.


And that's almost it.
I didn't replace any components in the power supply. I don't have any suitable new capacitors at home, there's no time to order them, and I didn't want to install used ones, even branded ones with better remaining specs. Besides, the measurements showed that the original capacitors are in perfectly decent condition and should last for several more years.

The person who repaired this power supply was thrown off by the confusing—or, more precisely, unintuitive—labeling. If it had simply been marked NET-35, it would have been clear that this was the series and that the full model name had to be looked up separately. But the label clearly says NET-35A, and there's a datasheet with the specifications.

Now, about the power supply itself. Subjectively, it seems pretty good: it holds its output well and proved highly resilient under harsh operating conditions. But I got the impression that something went wrong during development and it was deliberately pushed into a harsher operating mode. I suspect this is because it has three output voltages and had to meet a certain regulation coefficient for all of them, since there is only one feedback loop and no additional regulators on the board.

That's all for now. Keep an eye on the labeling—it can sometimes be as confusing as the one in this article. I hope the power supply serves for a long time and helps with the work.
$28.35
Go to store

Questions and answers

Why did the MeanWell NET-35A output +15 and -15 V instead of +12 and -5 V?

The NET-35A marking was read from the PCB, but different versions in the series could use boards with the same marking. Based on the measured voltages, the author suspects it is a NET-35C or NET-35B.

What voltages and currents were tested on the MeanWell NET-35?

The tests covered +5 V at up to 2.5 A, +15 V at 1.5 A, and -15 V at 0.5 A. The author considered the test successful; the transistor reached 70 °C.

Were any components replaced in the MeanWell NET-35 power supply?

No. The capacitors were checked and found to be in good condition, and the board was cleaned and coated with protective varnish.

The figures from this review are in the measurement database: Power supplies


Discounted products


You may be interested in


Discounted products


Support the project

All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!

Comments1

Comments are published in their original language.

  1. 11 April 2023 22:13#1
    Дома пару раз сталкивался с БП MeanWell в металлическом кожухе.
    Мне был непонятен момент с нагрузочными резисторами - контора MeanWell ставила их слишком небольшого сопротивления. Менял их сразу и увеличивал их сопротивление минимум в 1,5 раза. Проблем не было.
    0

Discuss on the forumHave a question or some experience to share? Discuss the article with the author and readers: the forum is a convenient place to share schematics and photos.
Go to the forum

Only registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.

Log inRegister