Step-down voltage converter with an output current of up to 20 A

I needed a powerful step-down converter for one of my projects, so I decided to give it a quick test before putting it to use.
A brief inspection, some tests, and conclusions.


What I actually needed was up to 40 A at 4.8–5 V. The load can be split, so I could use two 20 A converters. But I wasn't keen on risking an order for two right away, so I decided to get one to try first.

Incidentally, this is already the second one I've ordered. I tried ordering one a while ago, but they sent a lower-power model. The most annoying part was that I only noticed after the buyer protection period had expired. I had to place the order again, this time from a different store.

The packaging is as basic as it gets: an envelope and an antistatic bag. The converter is compact, measuring 60x52x28 mm.


Specifications (from the product page)
Input voltage: 6–40 V DC (10–40 V recommended)
Output voltage: 1.2–36 V DC
Output current: 20 A (max.), 15 A (recommended)
Efficiency: 95% (24 V to 12 V, 20 A)
Output ripple: ≤ 50 mV
Connection type: terminal
Short-circuit protection: self-recovering (not suitable for a prolonged short circuit)
Size: 60x53x27 mm/2.36x2.08x1.06"

It looks reasonably tidy. Nothing is loose or dangling, and the heatsinks are secured with small screws rather than hanging off the component leads. There are four mounting holes.



1. On the input side, there's a screw terminal block, a switch, and a power indicator LED. The switch controls the PWM controller's enable signal. The terminal block is pretty flimsy.
2. The output side has an identical terminal, with two trimmers next to it for setting the output voltage and current limit.
3. Two 470 µF, 50 V input capacitors.
4. Three 270 µF, 35 V output capacitors with branding that looks a bit too good to be true, though they could well be genuine. Hard to say.
5, 6. This is a synchronous rectifier converter, so the heatsinks carry two MOSFETs rather than a MOSFET and a diode assembly. The MOSFETs are identical: NCE8290, N-channel, 82 V, 90 A, 8.5 mΩ, which is actually pretty decent.


The layout isn't especially cramped, but it's still not great: the capacitors are right up against the power inductor, which tends to get quite hot during operation.


The PWM controller, op-amp, shunt, and other small components are on the underside of the board.


The PWM controller is visible at the top right: LM25116. Below it are a 4 mΩ shunt and an op-amp to amplify its signal: LM321



Key features of the PWM controller include synchronous rectification, a built-in driver with up to 3.5 A of drive current, operation from supplies up to 42 V, adjustable current limiting, and an output voltage range of 1.21–36 V.
In short, it's a very interesting controller.


The datasheet includes a typical application circuit. There's nothing unusual here: you can see the controller, the power MOSFETs, and the current-sense shunt. Note that the datasheet shows two application circuits, and in both the controller and power stage are powered from separate sources. The converter under review uses a single source, which is also allowed, but that limits the input voltage range to the controller's maximum of 42 V.


In practice, the output voltage range is a little less impressive.
1, 2. The minimum is more or less as specified, although with no load it wanders around between about 1.24 and 1.45 V.
3. But the maximum I could get was only 30 V.
4. That's with the input set to the specified maximum of 40 V, so this isn't a limitation caused by the input voltage. The feedback divider just doesn't seem to have been calculated quite right.


Power consumption when switched off is practically zero. When switched on with no load, the current is about 20 mA at 12–24 V, but rises noticeably to 60 mA at an input of 36 V. This measurement is rough, but I don't think that's a problem.


The current limit works, but the lowest setting is only about 700 mA. The highest I tested was 12.2 A; I didn't go any higher because fuses for my multimeter are expensive. At some current settings, the converter gave off a faint whine.


Next, I checked how accurately it maintained the output voltage at load currents from 5 to 20 A. To start, I set the output to 5 V.


Then I measured the output voltage at currents of 5, 10, 15, and 20 A. The multimeter was connected to the circuit board traces beneath the terminal block.
Across the 0–20 A range, the voltage drop was 0.12 V. I wouldn't call that bad, but it's noticeable at low output voltages.


The same test, but with a 12 V output and a 24 V input.
First, with no load


Then at currents of 5, 10, 15, and 20 A.
We get the same 0.12 V difference. I suspect there may be an issue with the circuit board layout.


While running the converter in different modes and taking photos for the review, I noticed it was getting warm. I was surprised by how high the temperature was, especially since I wouldn't say the preliminary tests had taken very long.


I also noticed that efficiency depends noticeably on the input voltage—or, more precisely, on the difference between input and output voltage.
For example, with a 12 V input, 5 V output, and 20 A current, the converter draws 114.5 W.


At a 24 V input, it draws 117.3 W, and raising the input to 36 V increases that further, to 121.6 W.
So, at 5 V and 20 A output, changing the input voltage from 12 to 36 V gives us a power draw of 114.5 to 121.6 W.
In my case, the input will be 10–14 V, so it's all fine, but this could be a concern for some people.


I measured efficiency in several modes. Below are three graphs for a 5 V output and 12, 24, and 36 V inputs. The horizontal axis shows the load current, from 2.5 to 20 A in 2.5 A increments.

The results are fairly rough because I estimated input power from the power supply's readings, so voltage drop in the wires between it and the converter affected the measurements. I think the actual efficiency is about 1% higher.


Here are three more graphs for different modes: a 12 V output with 24 and 36 V inputs, and a 24 V output with a 36 V input (top graph).
Note that in the 36-to-24 V test, the load current was 15 A, giving an output power of almost 360 W—well above the rated maximum of 300 W.


As I mentioned earlier, the converter gets noticeably hot. To check this, I ran a test at 5 V output, 12 V input, and load currents of 10 and 15 A. Note that this is one of the more favorable operating modes; it may get even hotter in other modes.
1. The converter was already slightly warm when the test began.
2. After 20 minutes at 10 A, the temperature was within the normal range.
3. After another 20 minutes at 15 A, the heating became more noticeable. The input transistor reached the highest temperature: 106 degrees.

Based on the test results, I recommend either limiting the output current or considering active cooling.


Ripple.
Overall, it's actually pretty good. I was expecting worse.
Output voltage: 5 V; input voltage: 12 V.
1. No load
2, 3, 4. At currents of 5, 10, and 20 A


There were actually some spikes in the ripple spectrum, but they aren't visible because I used an adapter on the probe (1 µF + 0.1 µF) for the test.
Below is an oscilloscope trace with the probe connected directly, at 20 A and a 12-to-5 V input/output ratio.


The same load currents—5, 10, and 20 A—but with different input/output ratios: 30-to-5 V on the left, 24-to-12 V on the right.


If you look closely at the oscilloscope traces above, I think you'll notice that the “horizon is tilted”—each successive pulse is higher or lower than the previous one.
This caught my attention, so I increased the time base and got this rather unpleasant-looking trace. The overall ripple amplitude is about 80 mV. This occurs at output voltages of 12 V and above, and at currents around 15 A or higher. The lower trace was taken at 12 V output, 24 V input, and 15 A.


To wrap up the review, here's a comparison photo of other converters in the same form factor: the step-up converter is in the middle, and the step-down converter on the right is rated for 10 A. I think I'll write short reviews of them too, if anyone's interested.



To sum up, the converter basically works, but there are quite a few caveats.
1. Heat: you can't draw more than 15 A from it continuously without additional cooling, though this is mentioned in the description. Even 15 A is already close to the limit, especially with a large difference between input and output voltage.
2. Current adjustment starts at 0.7 A
3. Output voltage reaches only 30 V, despite the claimed 36 V.
4. The input capacitors are low quality.
5. The terminal blocks are flimsy, especially for the claimed 20 A.

In short, the manufacturer started with a generally decent set of components, but ended up with a mediocre converter. I think some of the problems come down to PCB layout errors. It’s perfectly usable, and I even liked it to some extent, but the cooling needs some attention.

That’s all for now. I hope you found this useful.