Raxfly charger (power supply), 5 V with a claimed current of 7.2 A

Not long ago, I was asked to review a USB charger. Since I like tinkering with all kinds of power supplies, I agreed. So I’ll take it apart, study it, and test it.


The original plan was to review a different model, with a higher output current and a USB Type-C port, but for some reason they sent a simpler model instead. Also, the model I ordered plugged into the wall with a cable, while the one I’m reviewing is a wall-plug style adapter. In my case, the cabled model would have been more convenient, but we’ll make do with what they sent. :)

Key specifications-
Output voltage - 5 V
Load current - up to 7.2 A (for some reason, it says volts).


Part one: inspection.

It comes in soft packaging, with the power supply itself inside an additional soft pouch.


It looks quite neat: a black case with a finish that resembles Soft Touch. The construction is fairly sturdy, so I don’t think I’ll have anything to complain about here.


It’s fairly large, but when plugged into a standard outlet it doesn’t block the neighboring sockets, since it stays within the outlet’s footprint. Still, I prefer power supplies with a cable; I don’t like large adapters plugged directly into the wall. And for the use I had in mind, there’s no way to connect this adapter at all :(


Dimensions listed on the store page.


The same specifications as in the image above are printed on the side of the case: 5 V, 7.2 A, input voltage 100–240 V.
The device has four ports: two for Android devices and two for Apple devices. This didn’t matter much to me, since my devices are usually charged with cables that don’t have data wires.


The case consists of two glued-together halves. A thin screwdriver and a minute later, one half comes away from the other. Since I wasn’t concerned about damaging the appearance in this case, I didn’t take much care, but the device still came apart quite neatly.


But the board inside was also secured with something like glue, so I had to literally tear it out of the case. You’ll see marks from the side cutters below.


What can I say? At first glance, it looks neat, with nothing sticking out.


By the way, you can see the wires to the mains plug at the bottom of the photo. They have silicone insulation and can easily withstand the temperature of a soldering iron tip. A bit of extra protection, in its own way.
The layout is fairly dense, but there is still some free space. In my opinion, though, the heatsinks could have been larger.


I also have no major complaints about the soldering quality, though there are a few spots with small solder splashes (for example, at the top right). Otherwise, it’s neat. The board has safety slots, but the mounting point for the output diode’s heatsink is only about 4 mm from the input circuitry. Another questionable spot is near the safety slot: ideally, the slot should have continued to the hole above it. Dust is unlikely to get inside this power supply, so you could say it’ll do, but in my opinion, they still didn’t think it through.


I mentioned above that the heatsinks could be larger for a reason. For example, this 5 V 6 A power supply has its output diode assemblies (two of them!) mounted on a large heatsink whose plane is parallel to the plane of the case. The heatsink in the unit under review is much smaller, and there’s only one diode assembly.


I’ll go through the main sections of the power supply and try to analyze them.
1. There’s no input filter—not even a thermistor to limit the inrush current when the filter capacitors charge. But this is common practice in devices like this; compact size and low cost come at a price.
2. The input filter consists of three 10 µF capacitors. The power supply’s rated output power is 7.2 × 5 = 36 W. The minimum input filter capacitance should be 35–40 µF (for a voltage of 198–242 V). This one has 30 µF, which is clearly too little—and even more so for the 100 V input voltage listed in the specifications.
3. High-voltage transistor: MDF6Т60.
4. The transformer is low-profile, and its apparent power is roughly in line with the rated output.
5. The output diode assembly is an SBR20100CTP, and there’s a Y-type interwinding noise-suppression capacitor. No complaints about either one. Though using two diode assemblies would have been much better; I’m not even dreaming of a synchronous rectifier.
6. They also cut costs on the output capacitors: 3 × 1000 µF, and they’re clearly cheap ones. Both the capacity and the quality leave a lot to be desired.



The high-voltage section. I was a little surprised by the "zoo" of resistors in the high-voltage transistor circuit; apparently, they were selected to set the maximum output current more precisely. The total resistance is 0.755 Ω.
What’s strange, though, is that there’s no diode-and-resistor network in series with the MOSFET gate—or anything at all. There’s a direct trace between the PWM controller and the transistor.


The PWM controller is marked 5673A. I couldn’t find a datasheet for it, but this controller often turns up in Chinese power supplies. For example, in this power supply. I also wrote there that it could be a 63D39, or the better-known FAN6862.


There are four USB ports on the output. Their power rails are connected in parallel, which makes sense since the device doesn’t support QC. In principle, you can draw the full power from a single port, though I think it would probably just melt.
The data pins on the left pair of ports are connected in parallel and wired to voltage dividers, while the right pair has separate dividers for each pin.


Here’s a brief explanation of how the whole system works, quoted from my own review.
The thing is, USB charging is designed so that the power supply has a specific resistor combination on its terminals. The device detects that combination and knows how much current it can draw. Since the device might be connected to a computer, whose port can supply no more than 0.5 A (0.9 A for USB 3.0), this safety measure is useful.
More details here
For Apple devices, the second option shown below is used.


When the data lines are connected together, there are many possible configurations. This one uses a 33/10k divider, which is closest either to option 4 in the example above or to the last option in the description below. In any case, it’s marketed as "For samsung galaxy".


Of course, I redrew the schematic; maybe it’ll help someone with a repair or modification. Interestingly, the schematic is quite similar to that of this power supply. The difference isn’t very big, and both power supplies have the same specifications.
The output voltage is fixed and cannot be adjusted, but the divider resistor values are chosen so that the output voltage is higher than 5 V (for 5 V, the resistor values would be equal).


Part two: testing.

It wasn’t very convenient to connect everything, so I had to improvise. I also wanted to take the positive lead directly from the power supply’s heatsink.


While connecting everything, I noticed that the heatsink was very close to one of the USB ports. I was about to criticize the designers, but then decided to check. It turned out they’d used a diode array whose center pin is insulated from the heatsink tab. That’s the right and safe way to do it, but on the other hand... If the manufacturer uses a different type of diode, there’s a risk of a short circuit.


Anyway, I managed to get it connected and moved on to the tests.
First up was testing output voltage regulation under different loads, efficiency, and ripple.
I planned to start at zero and gradually increase the output current in 1.5 A steps.
1. No load; power consumption is below the power meter’s measurement threshold. Output voltage: 5.14 V.
2. Load: 1.5 A; output power: 7.725 W. Efficiency: 75%.


1. Load: 3 A; output power: 15.5 W; efficiency: 79%. Ripple is about 100 mV.
2. Load: 4.5 A; output power: 23.26 W; efficiency: 80%. Ripple is about 120 mV.


1. Load: 6 A; output power: 31.1 W; efficiency: 79%. Ripple is about 140 mV.
2. Load: 6.7 A; output power: 34.7 W; efficiency: 79%. Ripple is about 150 mV.


You’ve probably noticed that the last test wasn’t run at the rated current of 7.2 A, let alone the 7.5 A I’d planned.
The thing is, during the tests, the output voltage dropped below my set threshold of 4 V.


To find the maximum output current, I ran an automatic test, which returned a value of 6.8 A. For the final test, I simply lowered that value by 100 mA, giving 6.7 A. This is the current at which the power supply operates stably.
So we can say that it failed the test for “output current.”


But output current is one thing, and maximum output power is quite another. To find the maximum output power, I ran a long-duration load test. Because of the power supply’s design and the way it had to be connected, I had to test it as shown in the photo. The front panel wasn’t fully flush, which improved cooling, but the power supply was sitting on the table, which made cooling worse. I figured the two effects would cancel each other out.


The test was the same one I always run: gradually increase the load current at 20-minute intervals. In other words, apply a load, wait 20 minutes, measure the temperature, increase the current, wait another 20 minutes, and so on until the end.

I won’t include all the test photos; I’ll just put everything into a table later. But there’s one detail worth mentioning. After increasing the current to 4.5 A, I noticed something strange in the output voltage graph. It wasn’t there when the unit was cold; the problem appeared after it warmed up.


I had to get the oscilloscope out again and check what was happening at the output.


The output had ripple at around 1 kHz, with a 0.4 V peak-to-peak amplitude and spikes reaching 0.8 V. The power supply was also quietly buzzing, but the electronic load’s fan was noisy, so I didn’t notice it at first.
Below are the oscilloscope traces at 4.5 A and 5.5 A.


The whole test took about 1 hour 50 minutes: five stages at 1.5, 3.0, 4.5, 5.0, and 5.5 A. The extra 10 minutes accumulated while I was taking measurements.
The current spike roughly halfway through wasn’t a load glitch. After 4.5 A, I initially set it to 5.5 A, then decided it would be better to check 5 A first.


Since I prefer to run non-destructive tests, I stopped when the transformer temperature exceeded 100°C. The diode assembly was at around 120°C. The power supply was giving off a strong smell of overheated varnish.
I could keep experimenting, and I think the power supply might even have managed 6–6.5 A, but I wouldn’t have felt like repairing it afterward.


Final output power test results—or, more precisely, temperature results.
I’d say the power supply can handle currents up to 6–6.5 A for short periods, but no more than 4.5 A continuously. I’d put the maximum continuous current at around 5 A, but I wouldn’t recommend running it that way.
It failed the test for “output power.”


Conclusions.
Despite its fairly neat and sturdy construction, the power supply has a number of problems.
1. Maximum output current is 6.8 A instead of the claimed 7.2 A.
2. Continuous output power is 22.5–25 W instead of the claimed 36 W.
3. Ripple at around 1 kHz appears at output currents above 4.5 A, once the unit has warmed up.
4. The input capacitors have less capacitance than required for this power level.
5. The output capacitors are low quality and have less capacitance than required for the claimed output current.
6. There is no input or output filter.

While putting together my conclusions, I remembered a review of a similar power supply by malimana. They included a photo of the same model, but rated for 5.2 A, which is much closer to reality.
Below is an example of how to increase a power supply’s output current by changing a single digit. Apparently it’s that simple—I had no idea. I was changing components and upgrading heatsinks, when all I had to do was change a number and it became almost one and a half times more powerful.
In general, fitting a 35 W power supply into a case this small is possible either with a high output voltage or by using a synchronous rectifier on the output.


That’s about it. In my opinion, the power supply itself is usable, but $11 is clearly too much for these specs, especially given the shortcomings listed above. Also, keep in mind that in reality this is a 4–4.5 A continuous, 5–6 A short-term power supply—not 7.2 A as stated in the description. The seller seems to mention a one-dollar-off-nine coupon at the bottom of the page, but I couldn’t figure out how to get it.

I can already hear the objection that without a QC port, it’s completely pointless. Sure, a port wouldn’t hurt, but I personally don’t need one, and I doubt I’m alone. There are plenty of applications that don’t need QC, and the manufacturer could simply have released two versions, with and without QC.