EPP-100-12 power supply, the second of six

So, today I have the second review in a series of reviews of power supplies made by Mean Well. This time I chose an open-frame power supply, i.e. a board without an enclosure. As with the last one, it has its own "twist," but you can read all about it in the review, which, as usual, includes tests, oscilloscope traces, and conclusions. All in all, I think anyone interested in good power supplies won’t be disappointed.


Like the previous one, this power supply was bought on TaoBao and sent to me for testing by one of my regular readers, whom I’d like to thank separately. The next power supplies will be interesting in their own ways too, but people asked me about this one in the comments, and I was very interested in it myself.

This power supply came without any packaging other than a small piece of film, but it arrived in perfect condition, which is always nice.


First, a few words about the key features.
Power supplies in this series are designed to operate across the full input voltage range (90–264 V), have a built-in PFC, offer efficiency of up to 92.5%, and include a full set of protections against overload, overheating, overvoltage, and short circuits. Output power is up to 75 W with passive cooling and up to 100 W with active cooling, so the power supply can be connected to a fan.
At the same time, no-load power consumption is no more than 0.5 W.


Detailed specifications. There’s more information here, of course, but the short description is enough for the most part. Still, I’ll refer to this table during the tests.
Power supplies in this series are available with output voltages of 12, 15, 24, 27, and 48 V. The model reviewed here has a 12 V output; it’s highlighted in the table.


The power supply’s appearance.



The power supply uses the same form factor as the RPS-120-27 shown earlier, roughly one by two camels 2 by 4 inches.


As you’d expect from a proper branded power supply, the datasheet gives the full dimensions, including the location of the cooling fan.


For another way to get a better sense of its size, here it is next to a “people’s” power supply.
The one under review is a little smaller, but the “people’s” power supply board looks much emptier.


The input capacitor has a label with all the necessary specifications. Its capacitance is 82 µF, its voltage rating is 420 V, and it’s made by Nichicon.


As expected, there’s an input filter to keep interference from the power supply out of the mains. This time, though, there’s only one fuse (I’m comparing it with the RPS-120-27), but the input terminals are marked ACN and ACL, for neutral and live, respectively. For safer operation, it’s best to connect them according to the markings.
The photo shows two common-mode chokes, with a bridge rectifier tucked between the second (green) choke and the board.


The input filter also includes four Y1 capacitors. The first pair connects the power supply input to ground, and the second pair connects to the output side of the filter. Ferrite sleeves are fitted over the leads of the two middle capacitors. This is needed to reduce radiated interference: because the board is so densely packed, these capacitors are raised above it and have long leads that could act as antennas.
A varistor is installed at the input to protect the power supply from voltage surges.


A heatsink is also clearly visible in the center of the board, almost completely covered in what looks like silicone sealant. The power factor correction transistor is mounted on it.
Another noise-suppression choke and a temperature-sensing thermistor are also embedded in the sealant. That’s partly why I didn’t take the whole assembly apart.
If you take a close look at the previous power supply review of the SDR-120-24, you’ll see a lot of similarities; it’s just easier to inspect those components there. In fact, the input sections of these two power supplies are very similar.


Another view shows both the EMI suppression inductor (embedded in the white compound) and the PFC inductor (at the bottom of the photo), as well as the transformer (on the left).
The PFC inductor is connected using just its winding wires. At least, I couldn't make out any leads on the inductor, but it's firmly fixed in place—quite literally, "for good". It seems to be glued in with sealant as well.


You probably noticed how this power supply differs from the RPS-120-27: there's an additional board on the side. It carries essentially all the control circuitry: the PFC controller, PWM controller, feedback circuits, and protection circuits.


1. The board has a sort of "window" for the EMI suppression capacitor. A ferrite bead is fitted over the lead on the "hot" side. It serves the same purpose as the sleeves on the primary-side capacitors: reducing radiated interference.

The output section of the power supply is pretty conventional.
2. The output uses two 40L45CT diode assemblies. Interestingly, they're connected "backwards"—on the negative output rail.
3. There are two output capacitors, rated at 1500 µF and 16 V, made by Rubicon. There's no output inductor.
4. The additional board has a trimmer for setting the output voltage, with a fan connector above it. As far as I can tell, the fan is powered from a separate winding and, accordingly, a separate rectifier. The RPS-120-27 was built the same way. This may seem a little odd here, with the main output at 12 V, but it's mainly for standardization, since these power supplies are available with output voltages up to 48 V.

A thermal switch is mounted on the diode heatsink. It completely shuts down the power supply if it overheats.


While looking for more information, I found a second (or perhaps first) version of this power supply. It differs in:
The control board design
The heatsink
The location of the EMI suppression capacitor.

The thermal switch mounted on the heatsink is also easier to see here.


Since most of the components are on the additional board, the underside of the main board is relatively bare. The soldering is excellent, and the board is clean.


Now, a little about what controls this power supply.
1. First is the PFC controller. Here it's an NCP1605 made by ON Semiconductor.
2. Next is the L6599 from ST. This is a resonant controller, which is the main difference from the RPS-120-27 and SDR-120-24. The board was coated with conformal coating, and even after cleaning it off, the marking on the IC package is hard to make out in the photo.
3. Two optocouplers are used in the feedback circuit, as in the two previous power supplies. One handles feedback, and the other protects against output overvoltage.
4. This time, the "cold" side also has an LM258A op-amp. Its job is to amplify the voltage from the current-sense resistor. This is the second notable change compared with the previous power supplies: here, overcurrent protection is implemented on the secondary side.
5. Two inverter transistors are on the underside of the board. They're marked 13NM60N and rated at 650 V, 11 A, and 0.36 Ω. At first, it may seem odd that the transistors have no heatsink. That's thanks to the resonant controller, so the only components that need heatsinks here are the PFC transistor and the output diodes.
6. The "bypass" diode. It connects the PFC input and output and takes the initial surge when the power supply is switched on.


The circuit is fairly complex and the layout very dense, so I didn't redraw the power supply schematic. But I'll try to explain how it works without one.
The block diagram is in the datasheet. Essentially, it's much the same as the SDR-120-24 diagram, with the only difference being an additional output for connecting a fan. Interestingly, the manufacturer even specifies the operating frequencies of the PFC and PWM controllers.

After the mains filter comes the PFC (power factor correction) stage, followed by the power stage, which is controlled by the PWM controller (Control). The controller monitors power overload (OLP), temperature (OTP), overvoltage (OVP), and voltage feedback.
The block diagram doesn't show overcurrent protection (OCP) or protection against overheating of the output diodes. I do have some thoughts about the latter, though, which I'll discuss in the overheating tests section.


The PFC controller's application circuit is in its datasheet. The PFC stage isn't particularly complicated; technically, it's a step-up voltage converter, but there's no filter capacitor ahead of it. More precisely, there is a capacitor, but its capacitance is very small (Cin). The PFC lets the power supply draw energy from the mains almost continuously, unlike conventional circuits, which draw power only at the peaks of the sine wave.
Besides improving the power factor, it can extend the power supply’s input voltage range (though not always), provide the inverter with a stable voltage, and thus make its job easier and improve its efficiency.

Unfortunately, the input voltage range can only be extended in relatively low-power power supplies, so you may well see a power supply with PFC but an input voltage range of 190–260 V.
The question of efficiency is a bit more complicated. Yes, the inverter will be more efficient, but the PFC itself reduces efficiency, so in the end I’d say it more or less evens out.


The controller in a resonant power supply is a much more interesting thing. Or rather, the principle itself is interesting.
In a conventional power supply, the switching that controls the energy delivered to the transformer is fairly abrupt, so switching usually occurs while a relatively high current is flowing through the switching transistor. Since most of the heat can be generated not while the transistor is on, but during turn-on and especially turn-off, the transistors run hotter. Increasing the switching speed too much can cause more interference, so this parameter is even limited to find an optimal balance.
In a resonant circuit, the controller drives the transistors so that switching always takes place under the most optimal conditions—ideally, with no current flowing while the power switch turns on or off.

The controller’s internal circuitry is quite complex, perhaps partly because it is relatively "old": the datasheet is dated 2009. It can also control the PFC controller, switching it off when power consumption is low. When I tested the power supply, I could clearly hear the PFC start up at an output current of around 300–400 mA.


Simplified block diagram of the power stage.


Here’s how the datasheet shows an example power supply with active PFC and the PWM controller mentioned above. I think it’s now clear why I didn’t redraw the entire circuit—and the output section in the diagram shown here is heavily simplified anyway.


I mentioned above that this power supply’s output current is monitored on the secondary side. To do this, an op-amp is installed on the auxiliary board, and a 3 mΩ shunt is fitted on the main board. This solution makes it possible to set the shutdown threshold more precisely when the output power is exceeded, and we’ll check that toward the end of the review.


Overall view of the power supply from above.


That wraps up the theory; now let’s move on to the practical part—the tests.

First, a word about adjusting the output voltage. The power supply surprised me with its very narrow adjustment range: the output voltage can only be set from 11.4 to 12.9 V, or just −5% to +9%.
All right, let’s set it to the specified 12 V and start testing.


This time I won’t include many photos of the testing process. There were a lot of steps, and the entire review would end up being nothing but photos, so I’ll just show the test "setup" and put the results in a table.
The first test involved:
1. UT-181A multimeter in voltage measurement mode
2. UT-61E multimeter in current measurement mode
3. Wattmeter
4. Electronic load.
5. Pen and paper.

I also used a TN61 transformer in an autotransformer configuration to obtain 100–110 V.


For this review (and possibly future ones), I decided to tweak the power supply testing procedure a little to bring it closer to some kind of "standard." In the first stage of testing, I’ll check and measure:
1. Output voltage regulation across the entire power range
2. Efficiency
3. Power factor

All tests were performed at two supply voltages: 220–230 V and 105–110 V. The power factor test at 107 V is somewhat questionable, since I’m not at all sure I can trust my wattmeter here. Next time I’ll try to come up with a more accurate method.


Ripple measurements were taken with a different electronic load to improve the repeatability of the results.


The power supply has a "green mode," and these spikes periodically appear at the output while it’s idle.


There are a lot of oscilloscope traces, so most of them are hidden under the spoiler.


Above, I showed measurements to assess how the power supply performs in different operating modes. For a clearer comparison, here are the waveforms at maximum power and with different input voltages.
The top traces show high and low frequency at 230 V; the bottom traces show high and low frequency at 110 V, respectively.
In practice, there’s little difference, thanks to the power factor correction circuit. As I mentioned above, it is essentially a step-up voltage converter that also regulates the voltage, so differences in input voltage mainly affect heating and efficiency.
The specifications claim 120 mV peak-to-peak ripple. I measured about 100 mV in my test, but there were very short spikes with a peak-to-peak amplitude of 175 mV. I don’t know exactly what the manufacturer measured or what method they used, but in my opinion, even a small output inductor would have improved things considerably. At the very least, it would have eliminated the short spikes.


Thermal testing is one of the most important and, in its own way, challenging tests, since it’s very difficult to simulate real-world operating conditions. So I usually try to create conditions that are more demanding than the manufacturer intended. This also lets me test the overtemperature protection, if the unit has one.

This power supply is more “delicate” than the previous one. The SDR-120 is rated for full power at ambient temperatures up to 55 (60) degrees, whereas this one needs its power derated once the temperature reaches 50 degrees. Also, low-voltage power supplies dissipate more power because of the higher output current through the output diodes.
And that’s not even counting the fact that a fan is needed for sustained power above 75 W.



By the way, the fan specifications list 20 cfm, or 20 cubic feet per minute. In units we’re more familiar with, that’s about 12 cubic meters per hour.
For reference, that’s a small fan, like the EB40101S2-999—40x40x10 mm.

The power supply can’t control the fan speed. With no load, it runs at about 50% speed, and at currents above 0.6–0.8 A it switches to full speed.


But since I specifically wanted to put the power supply through its paces, I didn’t use the fan. I also covered the power supply with its lid to restrict ventilation. I ran the test under two conditions: with “our” 220–230 V mains supply and with “foreign” 100–110 V mains, which is the lower limit for operation without derating.


Table notes:
Bridge rectifier — input diode bridge.
Input cap — input filter capacitor.
PFC transistor — I tried to measure the highest temperature around the PFC transistor, since it’s covered in sealant.
Inverter transistor — temperature of the hotter of the two inverter transistors.
Inductor — PFC inductor.
Transformer — transformer.
Output diode — outermost output diode assembly.
Output cap — one of the output capacitors.

Each test ran for 20 minutes, and the table shows the results at the end of each 20-minute interval. The total test time was about 180 minutes, or 3 hours. Before starting the test at 110 V input, I let the power supply cool to room temperature.

So, it turns out the power supply can handle up to 7.5 A even without cooling, but it’s right on the edge: at 100 V, it’s literally on the verge of triggering the protection.
At 230 V, it can even handle 8.5 A, but it shut down 4 minutes before the end of the test interval. With 110 V input, the power supply shut down one minute after I raised the current from 7.5 A to 8.5 A.

The thermal protection latches, so the power supply can only be restarted after it cools down. Otherwise, it briefly applies voltage to the output and then shuts off again.

In my opinion, the power supply passed this test with flying colors. It’s rated for 6.5 A, but it handled 7.5–8.5 A not only without active cooling, but also in a relatively enclosed space. I also couldn’t damage it: when it overheated, it simply shut down.


Thermal images taken at the end of the test at 230 and 110 V, just before shutdown. You can see that in the second case, the area around the PFC circuit heated up.


And finally, the last test: response to an overload.
I ran the load in DC test mode. In this mode, you set the maximum current and minimum voltage, and then the load gradually increases the current on its own. At 12 V, the power increased by about 1 W per second.
I tested different settings, with shutdown triggered when the output voltage dropped to 10 and 5 V, but as it turned out, there’s no difference: the protection monitors the output current.
Also, as expected, the voltage doesn’t sag as the current increases; it stays stable right up until shutdown. In other words, the voltage is either normal or the output is off. Shutdown at 10.9–11 A.


So, to sum up, this is a pretty good power supply: it has active PFC, a full set of protections against just about everything, high build quality, a conformal-coated auxiliary board, quality components, and high efficiency.
The power supply can continuously deliver up to 7.5 A without overheating or active cooling, even though its rated current is 6.3 A. It can sustain 8.5 A for a while, even without forced cooling. A fan can also be connected.

Overall, I think it's quite good. There is perhaps just one "fly in the ointment": short output spikes reaching 175 mV peak-to-peak, due to the lack of an output inductor.

Video version of the review.



That's all for now. As usual, I look forward to your questions in the comments and suggestions for tests of the other four power supplies.