Step-Down Voltage Converter, Up to 10 (12) A

Not long ago, I published a review of a step-down converter module rated for up to 20 A. As I mentioned then, though, it was already the second one I’d ordered from AliExpress. Today I’m reviewing the one I received about a year ago instead, with my first order.


I didn’t order this converter at all—the seller mixed something up and sent it to me instead of the 20 A version. I noticed too late, so I had to make do with it. But I’ve now decided to test it, and in the process found a link to this exact model from the seller who sent it to me.

The packaging is the same classic setup: an antistatic bag, with a compact converter inside. The board has a heatsink on each side.


Specifications from the product page (original translation)



At a glance, it looks neat, although as it turned out later, that’s far from the reality.
On the input side, there’s a terminal block, a 15 A fuse, and a trimmer for setting the output voltage. There’s no provision for current adjustment.
On the output side, there’s the same terminal block as on the input, an LED indicating the output voltage, and a spot for connecting a fan.
I should note that the terminal blocks are a bit flimsy for the claimed 10 A, and their overall quality leaves a lot to be desired, though they should work.


There are two capacitors on the input and three on the output. They’re all rated at 330 µF and 35 V. They do have some name printed on them, but I just call these no-name capacitors.


The layout is cramped. The inductor isn’t secured and is held in place only by its winding wire, and the capacitors are packed tightly together (presumably so they don’t freeze). When I took a closer look, I noticed that the screw holding down the diode assembly was rusty...


The converter is built around an XL4016E1 PWM controller, with an MBR2045CT diode assembly mounted on the second heatsink.
Both power components are insulated from the heatsinks. The heatsinks themselves are screwed to the circuit board, by the way.


And this is where things get interesting. It turns out that there are at least two versions of the datasheet for this chip: an older version 1.2 and a newer version 1.3. The older one specifies a maximum current of 8 A and input overvoltage protection, while the newer one doesn’t mention this protection but lists the load current as 12 A—that’s one and a half times higher.


I found screenshots of the first datasheet in the materials for my own review of a converter rated for 10 A; I downloaded the second one from the chip manufacturer’s website.
I think many of you remember the review of this converter model, even though more than 4 years have passed since it was published.


The block diagrams for both chips are identical, but the typical application circuits differ. The older version showed an example with a 5 V output and 8 A current over an input voltage range of 8–40 V. The newer one is more complicated. It specifies:
1. With an input voltage of 8–20 V and a 5 V output, up to 9 A is allowed.
2. In the 20–36 V range, with the same output voltage, the load can be up to 12 A.
3. At the same time, the typical operating point is listed as 5 V, 12 A with an input of 8–36 V, which doesn’t match the first point.

By the way, the circuit on the right also specifies an MBR2045 diode assembly, so the module manufacturer apparently just followed the chip manufacturer’s recommendations.



The soldering is average, but the flux residue on the circuit board looks pretty bad.


But there were some oddities here, too.
1. For some reason, the fan connection terminals are labeled V_in and V_out underneath.
2. There’s an unpopulated connector and missing components for PWM control of the converter. Judging by the circuit, they connect to the chip’s second pin (FB).


Here’s a comparison of the two converters: the 20 A model is on the left, and the one under review is on the right. The 20 A version costs only one and a half times as much and, in addition to its higher power, has a current-limiting function.


To test the converter, I used a power supply based on the RD6012 as the source, with an output current of up to 12 A.


The description lists the maximum input voltage as 30 V, but the converter also worked without any problems at an input of 36 V and 150 W.

1, 2. With a 30 V input and no load, the output voltage can be adjusted across its full range, with a little travel left on the trimmer.
3, 4. The output was initially set to 22 V. With no load, the current draw was 4 and 5 mA at input voltages of 12 and 30 V, respectively.
5, 6. Since the indicator LED is connected to the output, I set the output to its minimum voltage and measured the current again. This time it was 3 and 2.8 mA.




Load testing and efficiency measurements were carried out with an EBC-A10H electronic load, so the maximum load current was 10 A and the maximum power was 150 W. To reduce the effect of the contacts and wires, the measurement leads were connected directly to the converter board, bypassing the terminal block.


Load test with an input voltage of 12 V and an output of 5 V. A slight rise in output voltage is noticeable at currents above 7 A; below that, the voltage remained stable. Although, of course, calling a 30 mV increase in voltage a rise is probably overstating it.


With the same output voltage but an input of 30 V, things were somewhat worse: the voltage had risen by 0.18 V. Not exactly terrible, but the difference is clearly there.


With 24 V input and 12 V output, performance was quite good. There was still a rise, but only 0.12 V, or 1%.
I also tested it with 30 V input, 15 V output, and a current of 10 A, and found no difference from the results above.


But with 30 V input and 24 V output, at currents of 5 A and above, the output voltage started to drop. At 6.35 A and 150 W, the difference came to 0.4 V.


Efficiency was measured in three modes: 12-to-5 V, 30-to-5 V, and 24-to-12 V. The highest efficiency was with 24 V input and 12 V output, and the lowest with 30 V input and 5 V output. On the horizontal axis, each unit represents 1 A of load current.


During the preliminary tests, I noticed that the converter got noticeably hot, but I went ahead and ran a thermal test anyway.
For the first test, the input was 12 V and the output 5 V. The test had three stages: 20 minutes at 5 A, 20 minutes at 7.5 A, and 5 minutes at 10 A. I stopped the last stage because the temperature exceeded 110 degrees. The inductor was the hottest component, but the PWM controller wasn't far behind. I think the inductor's proximity to the controller and its heatsink also played a part.


With 24 V input and 12 V output, I ran two tests: 20 minutes at 5 A and 20 minutes at 7.5 A. I didn't increase the current further, since even at this level the temperature reached around 106–107 degrees.


Finally, I measured the output ripple. The measurement was taken with 1 µF and 0.1 µF capacitors connected in parallel across the probe.


Here, the results were actually quite good. The waveforms were captured with 12 V input, 5 V output, and loads of 0, 2.5, 5, and 10 A. Even at the heaviest load, ripple was 25 mV peak-to-peak, and at 5 A it was no more than 15 mV.


I also tested other modes. On the left: 30 V input, 5 V output, and currents of 2.5, 5, and 10 A. On the right: the same currents, but with 24 V input and 12 V output.
Even at 12 V output and a load current of 10 A, ripple was around 30 mV. That's also a good result for a board without output filters. I think adding an LC filter could easily reduce it several times over.


Conclusions.
Well, somehow I ended up not only reviewing this board despite never planning to, but also buying it in the first place. As you can see, that wasn't for nothing: the converter doesn't really stand out, and in my opinion it falls short of the 20 A version.
I wasn't impressed by the poor build quality, the grime on the board, the rusty screw, or the flimsy terminals. Still, the board delivers up to 10 A without any problems and has relatively low ripple, so it could be useful for some applications.

In short, it's usable, but it's better to pay a couple of dollars more and buy the 20 A board.