Compact, high-efficiency step-down converter based on the MP2225

I've already reviewed various converters, both large and powerful and really tiny ones. Today I'm going to take a closer look at a very compact version of a fairly powerful converter, with an output current of up to 4 A.
Tests, oscilloscope traces, graphs—as usual :)


Today's review will be very compact, much like the converter itself. There's not much to say about it, and probably no particular need to.

Specifications from the product page
Input voltage: 6.5 V–16 V
Output voltage: 5.3 V
Efficiency: up to 98%
Output current: 0–4 A
Voltage accuracy: <1%
Load regulation: <0.5%
Output ripple noise: <20 mV
Quiescent current: <500 µA
Operating temperature: -40 ℃ to 85 ℃


It's packed in a small bag, probably partly to make sure it doesn't get lost—it's that compact. I think you can see that in the photo next to the matchbox. The board measures 18 × 13.5 mm.


It's built around a PWM controller with an integrated power switch and synchronous rectifier: MP2225


The board is tiny. Besides the PWM controller, it has feedback resistors, input and output capacitors, and a pull-up resistor on the Enable input. The controller is enabled by default; to turn it off, connect the EN input to ground.


The controller's application circuit is ridiculously simple. In addition to the components mentioned above, there's a bootstrap capacitor for the power transistor driver and a capacitor for the internal voltage regulator.
The graph here also shows efficiency versus input voltage and load current. Incidentally, the datasheet specifies a maximum input voltage of 18 V, while the product description says 16 V.



The block diagram shows that it uses a synchronous rectifier, which is what gives it its high efficiency. Also worth noting: the converter operates at 500 kHz, allowing the inductor to be made significantly smaller, which also helps efficiency.


For testing, I soldered both wires and thick leads to the board, which I'll use to connect the load. The goal is to measure the voltage at the board's output as accurately as possible.


In operation, the board outputs 5.35 V, which is a little unusual since standard voltage levels like 3.3 V, 5.0 V, and so on are more common. I didn't measure its own current consumption separately, but the power supply doesn't register any load current at all.
With a 12 V input, current consumption is 405 µA in normal operation and 120 µA when the EN input is connected to ground.


First, load tests, along with measurements of how the output voltage varies with load current.

With a 12 V input, the protection kicked in at 4.5 A. The voltage remained very stable across almost the entire range—in fact, the difference was only in the third decimal place. But above 4.2 A of load current, it started to rise slightly.


The board is specified for a minimum input voltage of 6.5 V. The results were much the same, except that the board shut down at just 3.8 A.


But of course I was curious to see how the converter would behave at even lower voltages, for example 6 V, which is the minimum voltage for a 2S lithium battery pack. Of course, the voltage in a setup like this can be lower, but the effective range for lithium batteries is still closer to 3–4.2 V.

The results were almost the same as at 6.5 V, but once the current exceeded 2.4 A, the output voltage began to drop gradually. The board shut down when the current exceeded 3.4 A.


Efficiency—how could I not test this parameter, when it is the deciding factor both for this chip and for its intended applications? These are often battery-powered devices, where every bit of energy counts.

The manufacturer’s datasheet has some very appealing graphs, and the announcements claim efficiency of up to 98%. But look closely and you’ll see that this kind of efficiency is achieved over a very, very narrow load range.


I checked efficiency at two input voltages, 6.5 and 12 V, over a load-current range of 0.25–4 A. Accordingly, each division on the scale at the bottom represents 0.25 A.
At a 12 V input, efficiency peaked at 96%. At 6.5 V, it was higher at low and medium load currents, but as the current increased, it became the same as at 12 V.


Since the efficiency test takes quite a while, the board got noticeably hot and shut down at different load currents than before. For example, with a 12 V input, the board shut down at 4.25 A, even though it had no trouble delivering 4.5 A when "cold." The cutoff threshold was also lower at a lower input voltage.


The long-duration load test produced some interesting results:
1. At 12 V and a 2 A load, the board barely warmed up, reaching about 50 °C.
2. At the same voltage and a 4 A load, the board quickly went into "shutdown mode," so I had to run the test at 3.5 A instead. The PWM controller temperature was about 105–108 °C.
3. Next, I let the board cool down and tried running it with a 7 V input and a 3.5 A load. It shut down relatively quickly as well, so I tested at 3 A. The controller temperature was 83–87 °C.

Each test ran for about 15–20 minutes, which is more than enough to heat up such a compact board.


Ripple is another very important parameter, although in my opinion it matters less here than efficiency.
I initially tested it using a switching power supply, but the ripple seemed rather high, so I ended up testing with a linear power supply. I also disconnected the load from the computer and switched it to direct control to further reduce any possible interference.


With no load on the output, there’s a small sawtooth waveform with an amplitude of about 20 mV and a frequency of 300 Hz.


And here it is under load, at 2 and 4 A, or 50% and 100% load, respectively.
1, 2. High-frequency ripple. Overall, it looks fine: the board stays within the specified 20 mV.
3, 4. But at a slower timebase, you can immediately see that the full amplitude, including the spikes, is noticeably higher. The total "noise" here is 40 and 80 mV, respectively. This isn’t critical, and even the resistance of the wires will attenuate it almost completely. But in some situations, such as mounting the converter directly on a circuit board, it may matter.


These little boards are a bit cheaper on AliExpress. Banggood may also lower the price, at least for lots of 3–5 units. They also sell boards in a slightly different form factor, which may be more convenient for some people. The price and specifications are the same, but this version lets you choose the output voltage.



I’ll keep the conclusions very brief.
Technically, the board works and even delivers the specified performance, but with some limitations. For example, it can supply 4 A only briefly; for continuous operation, the limit is no more than 3.5 A with a 12 V input and 3–3.2 A with a 7 V input. You can increase the continuous current: there’s an exposed pad on the underside of the board, and you can solder a small copper tab to it. But I wouldn’t expect a big improvement—the IC package itself can’t dissipate much heat.

It regulates the voltage reasonably well, but at maximum load current it may rise slightly. Ripple stays within the specified 20 mV, but again, with a few "nuances."

I can already hear the question: can you use it in a car? My answer is no, because the input-voltage range isn’t all that wide, and voltage spikes in a car’s electrical system can easily exceed the specified 18 V.


That’s all from me. I hope this was useful, and as always, I’ll be happy to answer any questions.