/       /       /    Sanmim GPA50A-12 power supply: a very interesting model
Support the project on Patreon


Sanmim GPA50A-12 power supply: a very interesting model

351₴
Go to store
Another power supply, and Sanmim again—but this time it’s a 50 W model, somewhat similar to the Meanwell LRS-50-12, which I reviewed just recently.
In short, it’s a very interesting power supply.

To stay up to date with new posts on the blog, I recommend following me on Instagram. I’ll post notifications about new articles there, and might also write about something interesting from time to time: link to my account

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


So, how is this test subject similar to the previous Meanwell LRS-50-12? Well, at the very least, they have similar specs: both are 50 W, both output 12 V, and both are made by well-known manufacturers. But they differ in size, design, and price—the price difference is almost twofold, which can sometimes be the deciding factor.
Speaking of price, this power supply is also available on AliExpress, although the price seems rather high for some reason. There may be cheaper listings—probably, in fact—but judging by what I’ve seen while browsing and searching, Sanmim is often cheaper here.

So, there’s a datasheet for this power supply, and as usual, I’ll include screenshots from it throughout the review.


I ordered one unit. It came in a thin cardboard box.


Same situation as with the previous Meanwell: I needed a 15 V model, but the only one readily available was 12 V. The product line does include six models covering all the popular output voltages, but as usual, the least common ones are 9 V and 15 V.


It actually looks quite nice, and you can clearly see Sanmim’s signature style.


Its dimensions are slightly different from the more common 2-by-1 matchbox size (4 by 2 inches): it’s a little wider, at 58 mm instead of 51 mm.
Yes, there are a few more or less standard sizes, and 2 by 1 inches is one of them.


Every time I open a datasheet for this company’s power supplies, I’m impressed by their approach to technical drawings. They look great :)


The power supply itself looks just as good. The component density is moderate, and, for example, you can replace the transistor without removing the heatsink. In general, almost everything is laid out really conveniently.


Almost everything, but not quite. Take the input filter, for example: it’s better than the one in the Meanwell. This one has a proper filter with two X capacitors, as well as a 14D417 varistor. The only thing I can fault is the varistor’s placement: if it fails short, it could damage both the inductor and the neighboring X capacitor.


All the electrolytic capacitors are made by Aishi, which is a perfectly respectable brand.
1. The input capacitor is 100 µF, 400 V. Next to it are a pair of capacitors and an inductor; these are part of the PWM controller’s power supply circuit.
2. The 12N60NZ transistor is mounted on a tiny heatsink.
3. There are two optocouplers, implementing two independent circuits: the main feedback loop and the output overvoltage protection. If the first one fails, the second will kick in.
4. The output capacitors are 1000 µF, 16 V. Here I have the same complaint as with the Meanwell: I’d like to see 25 V capacitors.
5. There’s also an inductor on the output to reduce ripple, followed by a 470 µF capacitor. But there are some other capacitors nearby too; more on those later.
6. EBR30L100F output diode assembly



The board is clean and neatly assembled. In my opinion, it’s even better than the Meanwell’s.


1. The PWM controller is an OB5269. By the way, the power supply starts up very quickly, with practically no delay.
2. As in the previous Meanwell, there’s a TVS diode in the snubber, but it’s noticeably smaller.

I’ll also point out that they included an overtemperature protection circuit, and used not one thermistor but two connected in series. That’s an unusual touch. As a guess, this gives better sensitivity, but I could be wrong.


The output side is also neatly laid out. Below you can see a spot for a synchronous rectifier controller. Well, there it is again, just like on the Mean Well. I wonder what chip is supposed to go there. Does anyone know?
The feedback circuit is a typical TL431-based design. Overvoltage protection is implemented with a zener diode and transistor; I think that's better than just using a zener diode and an optocoupler.


Now let's get back to those strange capacitors.
The photo shows a puzzling assembly on the left, made up of four capacitors and an inductor. The larger 470 µF 25 V capacitors are before the inductor, and the smaller 220 µF 16 V ones are after it.
Below, there's clearly a spot for something to be installed, and the output connector has an unsoldered contact. In other words, it should have a 5-pin connector, but it has a 4-pin one.


It's all quite simple: this series includes models with two output voltages. What's more, the manufacturer went to the trouble of providing not just two outputs, but two fully regulated outputs. That's genuinely cool.
Here's what I mean. In a typical design, regulation is based on the highest-power output, or the output with the highest load current. The other outputs are only nominally regulated, because although their voltage depends on the main output, it also varies with the load—or, more precisely, with the combination of loads on the different outputs.


Here, both outputs are fully regulated: the main one by the power supply itself, and the auxiliary one by a DC-DC converter board. Nicely done.
Even better, in the single-output version they left in the capacitors and inductor for the second output. So, in principle, you can buy the converter board separately and get two outputs. Or just have a second, filtered 12 V output.


There's no output voltage adjustment, and the no-load power consumption tends toward zero.


All tests were performed with an EBC-A10H electronic load, using a four-wire connection to improve measurement accuracy.


The power supply really surprised me in the load-capacity test: although its maximum rated current is 4.2 A, it delivered 8 A!
And its output voltage regulation was simply excellent.


The efficiency claimed for this model is around 86–87%.


My result was a little lower. It's hard to say whether the manufacturer or my wattmeter is to blame; I'd need to measure it using other methods.
Efficiency was measured at load currents from 0.5 to 8 A, in 0.5 A increments.


Yes, in the screenshot above, the power supply falls short of 8 A in the load-capacity test. In fact, it even went beyond that; it just happened to shut down a little earlier in this particular test. When I measured efficiency, the power supply ran happily at an output current of 8 A, delivering 96 W of output power.


But what about heat? The manufacturer's graphs show nothing unusual: the operating range is -30 to +60, and the power needs to be derated at higher temperatures. They also claim that the unit can deliver 100% of its rated power with an input voltage from 85 to 240 V.


I checked the temperature at load currents of 2, 4, 6, 7, and 7.5 A, first in 20-minute intervals, then for about 15 minutes at 7 A, and finally for 10 minutes at 7.5 A. The power supply never shut down; I stopped the test myself.
As it warmed up, the output voltage dropped slightly, from the initial 12.01 V to 11.95 V. This was due to the heat: after removing the load, the output remained at 11.95 V, then returned to the original 12.01 V once the unit cooled down.


The temperatures aren't very easy to make out, so I'll list them in text as well. I tried to keep the transformer in the center of the frame.
1. 2 A: transformer 45 degrees, diode bridge 50
2. 4 A: transformer 61 degrees, diode bridge 70
3. 6 A: transformer 77, diode bridge 92.
4. 7 A: transformer 87, diode bridge 104.

I'm including the diode bridge temperature because it was the hottest component on the board, and the thermal camera kept focusing on it.


At 7.5 A, things get more interesting.
1. Transformer: 93 degrees
2. Transformer winding: 107 degrees.
3. Transistor: 80 degrees. Something around the transformer was heating up to 99; I suspected the TVS diode, but it's underneath the board, farther to the left.
4. Output diode bridge: 128 degrees.

Yes, that's hot—very hot—but the temperatures still aren't extreme. So, naturally, the question is: why doesn't the protection kick in? The thing is, the thermistors are located under the transistor’s heatsink, and it only reaches 80 degrees. By the way, I was genuinely surprised that with an output power of around 90 W, the transistor on that tiny heatsink only heated up to 80 degrees.

Personally, though, I think the overtemperature protection is implemented too crudely, since overheating is possible. I’d tweak this circuit a little so the protection kicks in sooner. Still, the power supply worked fine in this mode. It’s also possible the protection was designed for a lower current but a higher ambient temperature.


Ripple, as usual, starting with the high-frequency component. No C+C filter was used; the probe was connected directly to the output terminal.
1. With no load, the frequency is reduced; ripple is 20 mV.
2. At a 2.5 A load, the main ripple is 20 mV, or about 35 mV including the “spikes.”
3. At 5 A, the picture barely changed—if anything, it went up to 25 and 45 mV, respectively.
4. At 7.5 A, it increased a little: the main component was 35 mV, or 55 mV including the “spikes.”

As we know, spikes are usually easy to suppress; they “settle down” even on the wires to the device. But even with them, and even with an almost twofold overload, the results are simply excellent.


The low-frequency ripple was also quite normal. Anything noticeable started to show up when I loaded it to 7.5 A, but that’s clearly well above the rated 4.2 A, so the result is to be expected.


Conclusions.
The only thing I can really fault is the “sluggish” overtemperature protection, because otherwise the power supply worked perfectly.
So here we have a case where Sanmim turned out to be better than Mean Well, both in circuit design and build quality and performance. It’s also smaller and about half the price.

What can I say? I’m very pleasantly surprised. And if you add the fact that a second output is already fitted, just without the converter, then I think it’s simply great.
I didn’t test the second output, by the way. Since it comes from the same diode assembly but has its own filter circuit, I wouldn’t have learned anything new.

In short, it’s definitely worth buying—it’s worth every penny.

That’s all for today. I hope you found the review useful, and as always, thanks for your comments!
351₴
Go to store

Questions and answers

How much power did the Sanmim GPA50A-12 handle in testing?

Although rated at 4.2 A, it ran at 8 A output and delivered 96 W.

What ripple did the Sanmim GPA50A-12 produce?

At 2.5 A, the main ripple was about 20 mV, or about 35 mV including spikes. At 7.5 A, those readings rose to 35 and 55 mV, respectively.

Does the Sanmim GPA50A-12 have a second output?

The board retains components and space for a separately filtered second output. A separate DC-DC board is needed for a fully regulated second voltage; the author did not test it.


This model is included in the summary table: Comparison of power supplies: ripple, efficiency, and temperature rise

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

Calculate manuallyAll calculators

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!

Comments2

Comments are published in their original language.

  1. 22 December 2023 15:24#1
    Спасибо за обзор.
    Вариант на 2 напряжения как раз то что мне нужно для питания самодельного электронного микроскопа на базе модуля Sony EFIO. Вчера случайно выяснил что питание от сетевого GaN PD блока существенно портит качество изображения.
    Странно что вариант 5В+24В стоит на 120 грн дешевле чем нужный мне 5В+12В. 
    0
  2. 22 December 2023 20:23#2
    Давно смотрю на БП от Sanmim. Есть один минус (Россия) - это высокая цена и она больше, чем на аналогичные БП от Meanwell.
    Есть непонятный момент с БП Sanmim. Эти БП очень похожи на БП от Meanwell. Хорошо заметно, что Sanmim много взяло у Meanwell. Но почему же Sanmim не стала использовать копеечную вещь, а именно применять ферритовые бусины ? Для меня это загадка. В платах открытого типа и БП Meanwell, эти бусин очень много.
    И этот момент сразу бросается в глаза.
    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