Boost converter with power up to 400 W

For some reason, I rarely review boost converters, and I don't think I've ever reviewed one with this much power. But people often ask me about converters like this, so I bought this one specifically for a review.


The heading includes the price and shipping cost. In the end, shipping cost me a little less because I also bought a buck converter and various small items for review.

The converter is compact and, like the previous ones, came in an antistatic bag.

Specifications from the product page, in the original translation
Input voltage: DC 8.5 V–50 V
Input current: 15 A (max.); above 8 A, please increase heat dissipation
Quiescent current: 10 mA (12 V to 20 V, output voltage; the higher the voltage, the lower the current)
Output voltage: 10–60 V, continuously adjustable
Constant current range: 0.2–12 A
Temperature: -40 to +85 degrees (if the ambient temperature is too high, please increase heat dissipation)
Operating frequency: 150 kHz
Conversion efficiency: up to 96%
Overcurrent protection: Yes
Input reverse-polarity protection: No
Mounting: 4 threaded holes, 2.55 mm
Module size: 67 mm x 48 mm x 28 mm (L x W x H)
Module weight: 60 g

Apparently, “quiescent current” means the no-load current, and “heat dissipation” means cooling. The rest is clear enough: input 8.5–50 V, output 10–60 V, input current up to 15 A, output current up to 12 A.
It mentions overcurrent protection, but I'll discuss that separately, since there are some caveats.



1, 2. The input and output use ordinary, inexpensive terminal blocks. They look a bit flimsy for currents up to 12–15 A, so it would be better to connect the wires directly.
3. A 15 A fuse is fitted as protection against a short circuit in the load or converter. The fuse is simply soldered onto the circuit board.
4. There are two 220 µF, 63 V capacitors on both the input and output sides.


1. There are two trimmer potentiometers for adjustment: voltage on the left and current on the right. The current adjustment is implemented correctly, but the voltage adjustment works in reverse: turning it clockwise lowers the voltage instead of raising it.
2. It uses one of the most common PWM controllers, the TL494—a classic, you could say.
3. The power transistor is 160N75F03, rated at 75 V, 4 mΩ, 120 A.
4. The diode assembly is an MBR20100CT. Both power components are mounted on separate heatsinks with insulators.


The underside is completely bare. You can also see that both adjustment potentiometers are in the lower arms of their dividers, though with a slight difference: the voltage feedback divider is connected to the output voltage, while the current feedback divider is connected to the reference-voltage input of the second error amplifier. In other words, the signal from the shunt goes straight to the PWM controller input.
Perhaps that's also why the direction of rotation is reversed: with voltage feedback, increasing the resistor value increases feedback sensitivity, while with current feedback it decreases it.


Now for the tests and a few oddities in operation.
1. As specified, the converter starts up with an input voltage of 8.5 V.
2. But at 8.4 V or less, we get the first oddity: with no load, the current draw jumps and the output voltage rises to 85 V instead of the set 20 V. Raise the input voltage slightly, hear a faint click, and it's back to 20 V.
3. The lowest output voltage you can set is 11.77 V.
4. If you raise the input voltage above the set output voltage, the output will also start rising regardless of the setting. This is a characteristic of step-up converters, at least those with a conventional output diode. Because of this, it cannot limit the current during an output short circuit.
5, 6. The maximum output voltage I got was 67 V, and it remained stable at both 12 V and 24 V input. Keep in mind that the capacitors are rated for 63 V.


I also had a natural question about the PWM controller's power supply and the input voltage. As far as I remember, the TL494 has a maximum supply voltage of 40 V, while the input is rated up to 50 V. But under the heatsink I found a component that looks like a voltage regulator.


Sure enough, the PWM controller is powered at 17.5 V. I think this voltage was chosen to provide 15–16 V at the power transistor’s gate. By the way, you can make out its driver on the board, built around two transistors.
I applied 50 V and nothing burned out :)


Because of this topology’s characteristics, I used an LED matrix as the load to test current adjustment.
The current adjusts fairly smoothly, and you could say it goes all the way down to zero. At least, I can set it to around 30 mA, but if I try to go lower, it drops to zero.
The matrix was rated at 100 W at 33–35 V, so I limited the test to 3 A, though you can also set any value in between. A reminder: this isn’t quite the right way to adjust LED brightness, since the color temperature can shift.


To check how the current depends on the input voltage, I set the current limit to 1.5 A and the input voltage to 20 V, then lowered the voltage to 10 V. The current dropped slightly. Then I raised it to 30 V, and again the current was a little below the set value. But interestingly, when I set the input back to 20 V, the current returned to its original value. I think the reference voltage is just drifting a little, but it’s not a big deal as far as I’m concerned.


How the converter behaves in different operating modes.
1, 2. Input: 10 V, output: 40 V. No problem getting 2 A, then 2.5 A of output current, with the input current at around 11 A.
3, 4. But I noticed an unpleasant quirk: when I tried to raise the load current to 2.7 A, the power supply predictably went into current limiting, but the converter kept trying to run. The input was at 6 V and the output at around 5.2–5.4 V, but the input current was 12 A while the output current was 2.7 A. Apparently, the transistor had gone into linear mode and was dissipating quite a lot of power. The input voltage dropped even lower very shortly afterward.
I only noticed this problem later, while selecting the photos, since I usually just photograph the testing process and don’t always notice what’s happening.


The converter had gotten pretty hot during the previous test, so I let it cool down a bit and carried on experimenting.
1. Input: 12 V, output: 19 V, current: 6 A
2. Input: 12 V, output: 24 V, current: 5 A
3. Input: 24 V, output: 36 V, current: 7 A
4. Input: 30 V, output: 48 V, current: 6.5 A

The converter behaved normally in these tests, and it felt like there was still some headroom. I also noticed that the diode assembly usually gets hotter than the transistor.


Next, I was going to test the efficiency. I turned off the load and went to get a notepad and pen to write down the readings. When I came back, I noticed out of the corner of my eye that the power supply’s display was flickering (it had been left on). The current was jumping from zero to 12 A, and the voltage was changing too.
I switched it off and tried starting it again, but the power supply always went into CC mode. The output voltage barely changed and stayed around 3–4 V.
I took a closer look at the converter and saw that the plastic insulator for the mounting screw had melted. So I suspect this is what happened: I was experimenting with different loads, then switched the load off while the converter had gone into linear mode without noticing. I stepped away for literally a minute, and by the time I got back, the transistor had suffered thermal failure, so the power supply was limiting the current. The transistor hadn’t failed as a hard short; it still had some resistance and even tried to work, but unfortunately, it was done for.


I wanted to continue experimenting, so I first tried installing a brand-new IRF3205. The converter worked without a problem, but the IRF3205 is rated for a maximum of 55 V, compared with 75 V for the original. Then I remembered I had some 110N8F6s left over from an electronic load. They’re rated for up to 80 V, though their resistance is one and a half times higher.
There was another dilemma here, too: the IRF3205 has higher resistance but noticeably lower gate capacitance, while the 110N8F6 is the opposite—slightly lower resistance, but higher gate capacitance (9.1 nF). Ideally, I’d have used the original transistors. I actually liked their specs, both for resistance (4 mΩ) and capacitance (6.7 nF), but I don’t have any :(
I also added thermal paste; there wasn’t any to begin with. I could have left it as it was, but the insulating pads have a waffle-like texture, with little square indentations, so I figured the paste couldn’t hurt. I applied it under the diode assembly as well as the transistor.

I can already hear the question: wouldn’t it be better to insulate the heatsink from the board rather than the transistor from the heatsink? From a heat-dissipation standpoint, yes, this is better, but you’ll also end up with an antenna radiating at the switching frequency—at the very least, from the transistor’s heatsink.


Efficiency was measured in several operating modes. First, with an input of 12 V and outputs of 19 and 24 V; maximum output power was 131 W.
Here and in the next test, each horizontal division represents 0.5 A.


Here are three tests: 24 V input and 36 V output, plus 30/36 V input and 48 V output.
As you can see, the converter reached its claimed 96% efficiency in this mode. The maximum load power during the test was 333 W (48 V, 8 A).


I noticed that the output voltage depends on the load current. Here’s an example from the test at 48 V output, with current ranging from 0.5 to 8 A.


For the thermal test, the board was simply left on the table with no active cooling.


The test was run in two modes. First, with 12 V input and 24 V output, at load currents of 2, 3.7, and 4.5 A. The first two runs lasted 20 minutes each, and the third lasted 10 minutes.
The converter performed quite well, which is why I went ahead with a third test at 4.5 A.
The output diode got hottest, reaching 85°C. The transistor and inductor were about 7–10°C cooler.


The second test used 30 V input and 48 V output, with two 20-minute runs at 3 and 4.5 A.
The temperature was considerably higher here. And since the input-to-output voltage difference was also greater, the transistor heated up more, exceeding 100°C.


For a clearer picture, I plotted converter losses in three modes—12–19 V, 24–36 V, and 30–48 V. The scale along the bottom is in 0.5 A increments of load current.
Based on this graph and the previous measurements, we can estimate the maximum operating conditions and temperatures.


Output ripple was measured as with the previous converters, by connecting 1 and 0.1 µF capacitors in parallel with the probe.
I expected the ripple to be high—this is typical of step-up converters—but I didn’t think it would be this bad.
First, with 12 V input and 24 V output, at load currents of 0, 1.7, 3.4, and 5.1 A, the ripple under load ranged from 0.4 to 1 V!


Next, a shortened test in other modes:
1, 2. 12 V input, 19 V output, currents of 3.5 and 7 A
3, 4. 24 V input, 36 V output, currents of 3.5 and 7 A
5, 6. 30 V input, 48 V output, currents of 3.5 and 7 A.

At the specified voltages and load currents, output power was roughly 40–50% and 80–100%, respectively.
In the last mode, ripple reached 1.2 V. Sure, you could say that the main ripple isn’t all that high and that the total is made up of short pulses, but they’re fairly wide. The blame lies with the converter topology itself, the lousy capacitors, and the suboptimal PCB layout.


And finally, a comparison photo of the four converters: three step-down and one step-up.
1. 10 (8) A
2. 20 (15) A
3. 12 (10) A
4. the one reviewed here



Now for the conclusions, and I’m afraid they won’t be encouraging.
No, the converter works, and in its own way it does a decent job, but it has a lot of shortcomings that could cause problems, so keep them in mind when using it.
1. Below an input voltage of 8.4 V, it operates unstably and produces an elevated output voltage.
2. If the input voltage drops under load, it may switch to linear operation; the only way to save it is to disconnect the input. This happens with power supplies that have current limiting. It’s unlikely to happen with batteries, but make sure the input voltage doesn’t fall below 9–10 V.
3. Heating is moderate, you could say, but it depends on the operating mode.
4. Ripple: for proper operation, replace the output capacitors with actual capacitors, not dummy components that merely match their size and weight. It would also be a good idea to add an LC output filter.
5. Short-circuit protection is provided only by a fuse. Remember that the output voltage cannot be more than 0.5–0.6 V below the input voltage.

What needs to be improved right away:
1. Replace the output capacitors.
2. Apply thermal paste and check that the transistor and diode assembly are clamped properly.
3. To improve efficiency, you can install a more efficient diode assembly.
4. It would be best to replace the original terminal blocks, or remove them altogether.

In short, it will work—possibly even quite well. But if the voltage drops significantly under load and the power supply goes into CC mode, you’re in trouble. It should work well when powered by batteries, but it’s still best to filter the output ripple.

One possible use is charging batteries—for example, charging an 18–20 V battery from a 12 V car battery.
With no load connected, set the required voltage. Then turn the current adjustment all the way to the left (until the trimmer starts clicking, or just about 20 turns). Connect the discharged battery (load) through an ammeter and set the charging current.

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