Compact power supply with power factor correction and a claimed output of 400 W

Another power supply has landed in my hands. I didn't buy it—my friend did—but that doesn't matter here. What caught my attention first was its claimed 400 W output, fairly compact size, and power factor correction.


This power supply was originally intended for a DPS5012 converter board, but we'll see how that goes. The tests didn't produce the results I expected.
But let's take things in order. I'll try to show both the pros and cons in this review, and perhaps this PSU will find its target audience.

The power supply has an open-frame design, meaning it's intended to be installed inside an enclosure rather than used on its own. Its overall dimensions are approximately 128 × 82 × 43 mm, which is quite compact for the claimed power rating.


A label on the side of the case tells us that the PSU has:
Input voltage: 100–240 V
Output voltage: 48 V
Output current: up to 8.3 A
Power factor: over 0.95

The power supply is mounted on an aluminum chassis that also serves as a heatsink. A protective film is applied to prevent damage; it should be removed before installing the PSU in a device.


The store page also has a more detailed description. Since it's in Chinese, here's a rough Google translation, which is quite enough to understand the specifications.
Besides the power and voltage, I was also interested in the claimed ripple level of 150 mV, which is very low for a 48 V, 400 W PSU.


The power supply is assembled fairly tightly, though I wouldn't say it's extremely cramped. For example, the Mean Well PSUs I've reviewed before are often packed much more densely.


The input has a pretty decent filter—I was even surprised. It includes two common-mode chokes, two X capacitors, and three Y capacitors. The only thing missing is a varistor. There's a spot for one on the board, but they "forgot" to fit it. Add a varistor, and you could call this a textbook filter.
The common-mode chokes also have plastic bobbins, which is the right way to do it.


1. The input also has a thermistor to limit the capacitor charging current, as well as a fuse. There's only one fuse, installed in the live line (as marked on the terminal block), and it's rather small, which isn't ideal.
2. The three Y capacitors for suppressing interference are connected to the PSU case.
3. The bridge rectifier is pressed against the case through an insulating pad, even though its own case is insulated. I think the pad is simply being used instead of thermal paste here.
4. The input capacitor has a capacitance of just 100 µF. Active-PFC PSUs have less stringent requirements for input capacitance than conventional PSUs, but 100 µF is still too little for 400 W—roughly 2–2.5 times less than needed. The active PFC choke is visible to the right of the capacitor.


All the power components are pressed against the aluminum chassis through insulating pads.


The transformer in this PSU doesn't have the gap typical of ordinary flyback converters. That certainly increases its power capacity, but it still looks rather modest for 400 W of output power, especially given the claim that it can operate with passive cooling.


1. Two small electrolytic capacitors are installed between the PWM controller board and the transformer. This area will probably get quite hot, which could shorten their service life.
2. The feedback circuit uses a single optocoupler, and the interference-suppression capacitor is the proper Y type.
3. The PSU output is also connected to the case, but through a regular high-voltage capacitor. Given that Y capacitors are used in the primary circuit, this arrangement is perfectly acceptable and safe.
4. The output has a large storage choke and three 470 µF, 63 V capacitors. The filter's total capacitance is only about 1,500 µF for an output current of over 8 A, which is too little. There's also no output choke to reduce ripple.


Looking at the output section of the board, you get the impression that it's a universal design for different voltage/current configurations, simplified somewhat for this particular version. The trimmer for adjusting the output voltage is visible on the left.


The unit is very easy to disassemble: three screws hold the power components in place, and another four secure the board. Here I’d give it a plus. Despite its relatively compact layout, the unit is quite repairable.


Some component leads are covered with PTFE insulating sleeves. A small detail, but a very useful one.


The input bridge rectifier is a KBL406, the PFC transistor is a 20N50, and the PFC output diode is an SFF10006A.
The inverter transistor is an FHA9N90, and the output diode assembly is a MUR3060PT. At a glance, all the components are rated with plenty of headroom for current and voltage.

But I also found signs of some minor, though obvious, cost-cutting. The insulating pads were made by cutting a standard insulator in half.


Insulators like these usually cover the entire power component, as with the component mounted on the left side of this board. I think it’s now clear “how it’s done.”


The whole power supply is controlled by a PWM controller, the CM6800G, mounted on a small daughterboard.


The CM6800 is a combined PWM controller that can also control the power factor correction stage in addition to its main function. It’s made by a company called Champion, which I hadn’t heard of before.


I didn’t redraw the power supply schematic, but it’s very similar to the one in the controller’s datasheet, so it should be useful for repairs.


The board is double-sided and very well made.


The high-voltage section: the input EMI capacitors have a discharge circuit, and some of the traces are reinforced with solder.


The output section is also quite neat, but the solder “blob” between the negative output terminals is a little unusual. Apparently, there’s a PSU version where these points are separate.


I initially thought I’d explain how the PFC works with some pictures, but then remembered the video. I think that’ll make it easier to understand.
In a nutshell:
A conventional power supply draws power from the mains only when the (peak) voltage is higher than the voltage across the input bulk capacitor. As a result, there are large current spikes when the capacitor is being recharged.
A power supply with PFC essentially has a step-up converter at the input, but without a bulk capacitor (that’s placed after the PFC stage). It can draw current from the mains over a wide range of input voltages, making its load profile much closer to that of a resistive load.
After the PFC stage, there’s still a conventional PSU, but it runs from a regulated voltage. This means you can reduce the regulation headroom and thereby improve efficiency. However, the PFC stage itself is less than 100% efficient, so overall, “it all comes out about the same.”


This power supply uses a single-switch forward topology, whereas the more common low-power single-switch PSUs use a flyback topology.
The block diagram highlights the forward converter stages on the right that aren’t present in the flyback circuit on the left. The forward converter adds a diode and an inductor, and one of the transformer windings is connected with reversed polarity (this is important).
There’s another difference, too: in a forward converter, the transformer core has no air gap, which is essential in a flyback converter.


This topology is very similar to a classic step-down converter.
In both circuits, the input switch “pumps” the output inductor, which then delivers energy to the load through a diode during the off period. The difference is that in a forward PSU, both the transistor and the transformer, along with one of the output diodes, act as the switch.
The corresponding stages are shown below, highlighted in the same colors for clarity. I think it’s now clear why I said earlier that this PSU has no output filter inductor: the one that’s installed is an energy-storage inductor. Do not short this inductor under any circumstances!




There’s not much else to inspect, so let’s move on to the tests.
The PSU was initially set to 48.3 V, though that doesn’t really matter. The adjustment range is 41.8–54 V, which is quite wide. But if you need to set the output voltage precisely, the coarse adjustment may make that difficult.


The first test checks how accurately the output voltage is maintained as the load changes, and measures efficiency and power factor. I used two multimeters, a wattmeter, and an electronic load for the test. An autotransformer was added to the 110 V test.
During testing, I found that at load currents of around 4–5 A and above, the output voltage starts to drop. It doesn’t just drop and stop, either—the process takes a noticeable amount of time. A similar effect can occur when the feedback resistors heat up rapidly, if ordinary resistors with a high TCR are used. But when I switched off the load while the power supply was hot, I saw that the voltage immediately returned to its original value, so in this case it was most likely just a straightforward power supply overload.
Below are two photos: before the test, with the power supply cold, and immediately after removing the load while it was hot. You can see that heating has practically no effect on the voltage.


Test results table. At low voltage, the efficiency measurement error is noticeably higher, so the figures are only approximate. The final test at low voltage and maximum current was not performed, since the power supply was clearly overloaded.


In practice, I tested the power supply at almost its maximum current. The load was actually drawing nearly 400 W, but the power supply struggled in this mode: it got very hot and the output voltage dropped—not by a huge amount, but enough to notice.
One interesting observation: when powered through the transformer, it (the transformer) started buzzing when the power supply's load current was in the 2.5–3.5 A range, peaking at around 3 A.


Measuring the output ripple is a fairly important test, since this parameter has a significant effect on both the proper operation of the load and the service life of the device's output stage.
To measure it, I have the recommended filter installed ahead of the probe. It consists of a pair of capacitors: a 1 µF, 63 V electrolytic and a 0.1 µF ceramic capacitor.


Ripple is measured at the converter's switching frequency (HF) and at twice the mains frequency (LF).


Overall picture of the power supply's operation
1. 230 V input, high frequency, 100% load
2. Same, but low frequency
3. 105 V input, low frequency, 75% load
4. Same, but low frequency.

As strange as it may sound, the power supply stayed within the specified 150 mV ripple. The only issue was that the first test showed large low-frequency ripple, which disappeared during the experiments.


Another important test is checking the thermal performance. The description said the power supply was designed for passive cooling, and even at the start of testing I had doubts that it could handle this at maximum power, since it was already getting very hot after a short test. I estimated that the maximum continuous current would be around 6 A, but as it turned out, I was way off. The most I could get for continuous operation with relative safety was just 4.2 A, or about 200 W.
The thing is, power losses in the supply in this mode were about 35 W, and dissipating that much heat in such a compact unit is only possible with active cooling.


In the end, the test was run on a shortened schedule: first at 2.1 A, or 25% load, then at 4.2 A, or 50% load. But since the power supply was clearly getting very hot even at 50%, the next run was only at 4.5 A, which is still less than 55% load.
Each test took the usual 20 minutes, for a total test time of 1 hour.
Since there were a lot of measurement points, I'll list them in the same order as in the table:
Input bridge rectifier
Power factor correction transistor
Power factor correction diode
Power factor correction inductor
Inverter input capacitor
Inverter transistor
Transformer
Output diode assembly
Output storage inductor
Output capacitors.


The thermal images show roughly the same picture.
1. Temperatures after the voltage stability test, with a total test time of about 15 minutes and the current gradually increased from 0 to 8.3 A.
2. After completing the test at a load current of 4.2 A
3. After another 20 minutes under a 4.5 A load.

The thermal camera also “saw” some component installed between the transformer and the output inductor, and showed a maximum temperature of up to 145 degrees. Changing the camera's angle changed this reading; the actual temperatures of the main components are listed in the table above.


So how does this make sense? The rated power is 400 W, but it can only deliver 200 W continuously. What's going on?
The answer is simple, and it's right there in the product description: this power supply is for audio amplifiers, where the average power is about 200 W even when the peak is 400 W.


And now for the verdict.
The power supply is reasonably well built. It has a proper input filter, all the power components are mounted on a heatsink, safety-rated Y capacitors are used, the inductors are wound with copper wire, and there's an active PFC stage. The only things to fault are the lack of an input varistor and the fact that the inductors are held in place only by the stiffness of their winding wire.
Most components have plenty of headroom, but that doesn't apply to the input and output capacitors. Their capacitance is fine for 200 W, but clearly too low for 400 W.
The ripple level is also fine overall, apart from the strange effect where the ripple amplitude suddenly increased above 8 A, then went away and never came back.

Based on the test results, I can say that this power supply has a maximum continuous power of just 200 W. With active cooling, you can easily raise that to 300 W, while 400 W is only available for short periods (a few minutes). You could increase the capacitor capacity, and the power supply would work a little better, but its continuous power would hardly change.

So, an interestingly mixed bag of a power supply. That's all from me; I hope you found the review useful.