Mean Well EPP-500-54 Power Supply

I've wanted to buy this power supply for a long time, but one thing or another kept getting in the way. Now, at last, it's in my hands. I'll say right away that I like it, although it does have a few minor quirks.

As usual, a reminder: to keep up with new posts on the blog, I recommend following me on Instagram. I'll post notifications about every new article there, and might also write about something interesting from time to time — link to my account

A while ago, I got an RD6006P hybrid converter module and decided to build a compact adjustable power supply. I could have made a custom enclosure to fit some power supply, but first, I already had an enclosure—and it was metal—and second, I wanted a compact solution.

I chose the EPP-500-54 for a reason. The EPP-400-xx series would have worked for me too, and possibly even the EPP-300-xx series, with some limitations. The catch is that neither series has a power supply with an output voltage above 48 V, and ideally I needed 65 V.
For some reason, brands don't make power supplies rated above 48 V. I wanted to buy an LLC-topology power supply for power amplifiers, but it turned out that a suitable one would cost about as much as a Mean Well.

In the end, I ordered two EPP-500-54 units... and recently received the first one :)
I ordered the power supplies from Taobao simply because in Ukraine they're not only available by special order, but also cost an absurd amount of money. Even the Russian ChipDip listed the EPP-500-54 at a much more reasonable price.

I ordered two power supplies; I'll tell you about the second one another time. I was pleased to find a mouse pad included as a bonus—nice :) The packaging, though, could have been a lot better than just a piece of bubble wrap. That's a bit odd, considering the price.


A brief overview of the series. Here and below, the screenshots are from the datasheet.


The power supply has an open-frame design, but that doesn't bother me. According to the label, this really is a Mean Well EPP-500-54. Since even the Chinese don't counterfeit these while keeping the name completely intact, this is an original unit.


The series includes eight models, which is quite a lot. What sets it apart from other series is the model with a 54 V output.


It looks very nice, in its own way, of course. You can see how the designers tried to fit everything into a fairly compact package, and I'd say they succeeded.


There is essentially no enclosure. Its place is taken by an L-shaped aluminum plate, which also serves as a heatsink for some of the power components.


The power supply measures just 127 × 76 × 41 mm, or 5 × 3 inches. The dimension drawing shows not only all the measurements, but also where the fan sits relative to the power supply.


The input has a proper mains filter, which makes sense; I'd be more surprised if it didn't. What's unusual, though, is the two-pin power connector and the lack of a terminal for the earth wire. It seems to connect directly to the chassis.
Speaking of the enclosure—or, more precisely, the aluminum chassis—it has no mounting holes. The only option might be to use the four threaded standoffs for mounting the board.


A fuse, a bunch of capacitors and inductors, and a thermistor. The only thing I couldn't find was a varistor.
There's a spot on the board for a second fuse, but it's been replaced with a jumper.


Overall view of the board. As you can see, in addition to transferring heat to the chassis, the manufacturer has also used separate heatsinks.


1. The components are packed very tightly. I also noticed a relay, which seems to be in the inrush current limiting circuit: it engages almost immediately after power-on and releases some time after the power is disconnected.
2. The power supply's inverter transistors are mounted on a separate, small heatsink.
3. To improve efficiency, the manufacturer used a synchronous rectifier, although at this high output voltage it doesn't provide much of a gain.
4. Thermal switches are mounted on two heatsinks to protect against overheating.


1. I plan to connect the output using screw terminals. I think this is simply a way to standardize the design with the lower-voltage models, which are rated for a much higher output current. On the left is the fan connector; on the right, the output voltage adjustment and a pair of auxiliary connectors.
2. The output has four 220 µF capacitors and one that is probably 150 µF. I didn't check the marking, but it's smaller than the others. The capacitors are rated for 80 V, and the manufacturer seems to have squeezed in the largest ones possible: 63 V capacitors would have been less reliable, while 100 V ones would have been larger.
3. There's an additional small board on the side that provides the third auxiliary voltage: 5 V.
4. Essentially, this is the familiar standby supply, which means the power supply has:
Three output voltages: the main 54 V, 12 V for the fan, and 5 V for the auxiliary circuits.
A PowerON signal to control the main power supply's startup
A PowerGood signal to indicate that the power supply has started up correctly
S+ and S- measurement signals, more on those later.


The input capacitors are installed in such a way that you can't read their markings without desoldering them. I couldn't be bothered to desolder them, so I just measured their capacitance: about 250 µF for both, which means they're probably 120 µF capacitors.
While I was at it, I measured the output capacitance: 950 µF, of which 880 µF is accounted for by the four 220 µF capacitors, leaving some for the fifth one.


The datasheet included a block diagram of the power supply. The only things you can tell from it are that it has two independent converters and protection circuits, and that the PFC and main inverter frequencies are 90 and 100 kHz, respectively.

Later, though, I realized the block diagram has errors. For example, one of the three optocouplers shown at the bottom is drawn backwards, and all three are transistor-type.
You can also see that the S+ and S- contacts are connected to the output circuits through resistors. The datasheet says nothing more about them, but something tells me these are feedback contacts for a four-wire load connection.
The very sparse documentation is disappointing. It basically consists of a specifications table and a dimensioned drawing.


Connect the input cable and apply power.
The output is close to the specified 54 V. I was pleasantly surprised by the no-load power consumption: just 1.4–1.5 W, which is very low for a 500 W power supply.
The output voltage adjustment range was about 49–56.7 V. If you try to turn it up any further, the power supply enters hiccup mode, indicating that the OVP circuit (output overvoltage protection) has tripped.


The datasheet also explains how to connect the connectors. Besides the input and output, there are:
1. A fan connector, a standard two-pin connector
2. A connector for the feedback lines and PG output
3. A connector for the 5 V standby output and power supply control.

The last two use a relatively uncommon DF11-4DS-2C connector (mating part). I even considered buying one, but it's expensive and has to be ordered, so I changed my mind. Besides, the power supply works without them.


As I mentioned, the S contacts (probably short for Sense) are connected to the outputs, with 13 Ω of resistance between them.


1. The fan power connector provides 12 V, which, as far as I can tell, is regulated.
2. The standby connector provides 5 V. This voltage is available only on this connector and doesn't go anywhere else.
3, 4. The PS_ON contact is pulled up to +5 V, so the power supply starts up normally without it being connected. To shut down the main converter, ground this contact; power consumption then drops to 0.2–0.4 W.


A load test from 0 to 8 A showed that the power supply performs reasonably well. One thing to note, though, is that the output voltage drifts by about ±5 mV all by itself.
Below are the results at 0, 1, 2, 4, 6, and 8 A.


I measured efficiency over a load range of 0.5 to 11 A, so there are quite a few data points. Although the specified efficiency is 94%, the power supply actually reached 95% at maximum power.
I also found that the overload protection trips at an output current above 11 A, around 11.2–11.3 A. The power supply also operates correctly at around 550 W for considerably longer than the specified 3 seconds.


I measured ripple with the probe connected directly to the terminals and got about 150 mV peak-to-peak, compared with the specified 200 mV.
Oscillograms at 0, 3, 6, and 9 A. The first clearly shows the “green mode,” when the power supply switched to low-power operation.


However, there was 100 Hz ripple at the output, both at 3 A and at 9 A; the only difference was the amplitude. There may have been interference picked up by the probe, but I’m not sure.


The thermal test was conducted with the power supply simply sitting on the table. The temperatures might have been a little better if it had been standing on its side, but I was more interested in assessing the overall situation.


The datasheet includes graphs showing how output power depends on air temperature, additional cooling, and input voltage. But to put it very simply, at a mains voltage of 230 V there are two operating modes:
1. Natural cooling - 320 W
2. Forced cooling - 500 W.


I tested it with natural cooling, a mains voltage of 230 V, and output currents of 3, 6, and 7 A, giving load powers of 160, 320, and 375 W. Each stage lasted 20 minutes.

The first two modes looked pretty good overall, but when I increased the load to 7 A, the transformer temperature exceeded 100 degrees.
Point D. PFC is the PFC diode. It sits next to the power transistor, but I couldn’t measure the transistor’s temperature.


The same modes shown in thermal images. During the test, I noticed that the transformer winding temperature at 6 A was 97 degrees. In the third image, I focused the thermal camera on that exact spot, and it was already around 115.


The power supply has a slight positive temperature coefficient for its output voltage:
1. Cold power supply, no load
2. Warm power supply, 7 A load
3. Warm power supply, no load.

I should also note that the power supply showed a bit of voltage “overshoot” after the load was removed. It was initially 54.050, rose to 54.054 over about 20–30 seconds, and then started to drop.
Overall, though, the voltage seems very stable to me. A drift of 40–50 mV at 54 V is pretty good.


At this point, I could have wrapped up the review, but I’d also like to cover a small modification to the power supply for use with the RD60xx series converter.


The RD6006P needs at least 61+3.6=64.6 V at its input to work properly, but as I mentioned above, the power supply can’t output more than 56.7 V.
So we need to solve two problems:
1. Increase the maximum voltage, i.e. extend the adjustment range upward.
2. Raise the lower threshold of the overvoltage protection. The specs say it can be in the 56.7–59.4 range, but mine was at the minimum :(

I decided to start by adjusting the overvoltage protection, but it turned out not to be so simple, especially without documentation. The OVP label on the board and the resistor next to it were a bit misleading. This resistor is the upper resistor in the output-voltage divider, but it’s in the voltage-regulation feedback loop.
So that was almost a dead end, though it did help a little later.


It became clear that I’d have to dig deeper, so I unscrewed the board from the chassis.


That’s when I found out that the power component closest to the edge is the PFC diode, with its power switch to the left.


Overall view of the underside of the board.


Individual components and sections:
1. ICE3PCS03, the PFC controller.
2. L6599, the main PWM controller. Three optocouplers are visible on the right. They’re all the same type, but two face one way and the third faces the opposite way, which contradicts the block diagram in the power supply’s datasheet.
3. MP6922, the synchronous rectifier controller. The 7812 regulator is visible above it and presumably supplies power to the fan.
4. LS957B. I looked for a datasheet but couldn’t find one right away, so I’ll assume it generates the Power_Good signal.


There are also two current-sense shunts on the underside of the board, and an op-amp on the top side that amplifies their signal. Ideally, I should have checked whether it has a CC mode, but I don’t have a load that can regulate the voltage in the range I need at around 550–600 W.


I figured out fairly quickly which optocoupler handles the protection. I could even have simply shorted it out and nothing would have shut down, but “that’s not how we do things,” especially since it turned out that the overload protection is connected to the same optocoupler as well.

It turned out there are two feedback loops.
1. At the top, the lower resistor in the OVP divider is marked; there’s even a spot next to it for a second one.
2. At the bottom, the lower resistor in the voltage feedback divider.

Since the simplest way to adjust it is to add a resistor in parallel, that’s what I chose to do. So, to raise the voltage, the resistance of the lower divider resistor needs to be reduced.

There are at least four TL431s on the board:
1. Main output feedback
2. Current feedback
3. OVP
4. 5 V feedback

This really did make things confusing, especially in the overload protection circuit, because the complete circuit there is: shunt - op-amp - transistor - divider - TL431 - optocoupler. OVP is connected to this same optocoupler as well.


After a few simple calculations, I estimated the resistor values to connect in parallel with the existing ones. For the voltage feedback circuit, I could use 20 kΩ, and for OVP, around 7 kΩ. I didn’t have a 7 kΩ resistor handy, though, so I used 9.1 kΩ.

This made OVP trip at voltages above 62.7 V. I then added a 33 kΩ resistor in parallel with the 9.1 kΩ resistor, bringing the trip voltage up to 64.32 V.


But that still didn’t seem like enough, so I replaced the 33 kΩ resistor, which was in parallel with the 9.1 kΩ resistor I’d installed earlier, with a 20 kΩ one (6.25 kΩ total). That let me set the power supply output to 65 V, although I wouldn’t recommend doing this.


I was careful throughout, but still accidentally shorted the power supply output to the fan connector with a probe. Luckily, only the connector itself was damaged.


I’ve already written about the converter and enclosure, but now it’s time to put everything together.
1, 2. There are only two ways to install the power supply: with the chassis facing toward or away from the converter. I chose the first option, since the chassis will act as a shield and improve cooling.
3, 4. The hardest part was cutting a hole for the AC inlet. I wanted to mount it vertically where the terminals are, but then it would get in the way of the power supply.


It’s very important to add extra insulation. The problem with this power supply is that components on the “hot” side are close to the edge of the board. If I install it as planned, they’ll be very close to the enclosure.
The same is more or less true of the top edge of the board.

For insulation, I used plastic that comes in the packaging of various products.


And that’s pretty much it. The only things missing are a fan and its control board, but the fan is still on its way, and the board is pointless without it. I’ll cover that another time.


Power it up, check it, and everything works :)
And here I need to make an important point:
Obviously, any modifications are entirely at your own risk. But I should also mention that I raised the voltage by 20% over the original value. That’s A LOT, and I plan to bring it down to around 60 V (+10% over the original).
The point is that power supplies can usually handle +10% without any problems, but +15% is already harder on them, and +20% is really pushing it. Better not to risk it.


So, what’s my verdict? I like this power supply, especially its high power in such a small package, along with its high efficiency, reasonable ripple, and overall performance.
I’m not too keen on the very limited output voltage adjustment range, and in my opinion the transformer runs a bit too hot, but those are minor quibbles.

If the price works for you, I recommend it—you won’t regret it. It’s hard to find a more compact off-the-shelf power solution.

That’s all from me. I hope you found this information useful.