460 watts in a tiny box? No problem—a compact 12 V power supply from HP

Not long ago, I started posting reviews of power supplies from the Banggood store. At the time, I mentioned that I’d be doing several reviews, with the power supplies presented in order of increasing power. Today I have the fourth and final review in the series, where I’ll take a look at a powerful branded power supply.


This power supply is marketed for use with ISDT Q6 and SKYRC B6 NANO chargers. At least, that’s what the product page says, but to me it’s just a power supply, nothing more.

The packaging is as simple as it gets: a piece of bubble wrap and a sticker with the part number. The product page shows this power supply with a power cable, but one isn’t included. Though if it was supposed to come with a plug like the one in the photo, that’s probably for the best—one less bit of clutter at home.


As you’ve probably gathered, this isn’t some universal power supply. It was originally designed for a specific purpose, then repurposed as a power supply for chargers. In fact, it’s a server power supply, as its construction and form factor make clear.


There’s a small label on top with the specifications and model name: DPS-460EB A.
Rated specifications:
Input voltage: 100–240 V
Output voltage: 12 V main, plus 12 V auxiliary
Output current: 38.3 A for the main output and 2.5 A for the auxiliary output.
Maximum output power: 460 W.

Bare metal on the bottom.


The power supply weighs just under a kilogram.


Its dimensions, excluding protruding parts, are 195 × 85 × 38 mm. Including those parts, the length increases to 240 mm.


One end has a power-cable socket, as well as a fan, handle, latch, and power indicator LED.
The latch is designed to secure the unit in a rack. The design is such that you can’t remove the power supply while the power cable is plugged in.
The other end has output terminals and an XT60U-F connector. That connector and a jumper are essentially all it takes to convert the unit into a power supply for a SKYRC charger. The wires are heavy-gauge, but the insulation on one of them was damaged. I’ll overlook that only because it’s the negative wire, so shorting it to the case is safe.


And of course, how could I not take apart such an interesting power supply?
I remove the four screws (one is under a seal) and lift off the cover, which is attached by makeshift hinges formed by slots in the case.


Here’s a closer look at the overall layout. It’s reasonably compact. Laptop power supplies are even more compact, for example, but this one has active cooling, so there needs to be enough room for the air to flow properly.


The input has a decent mains filter consisting of three inductors, a pair of X capacitors, and several Y capacitors. There’s also a fuse and a varistor, of course.


Judging by the relay and high-power resistor, and the absence of the usual thermistor, this unit uses a stepped capacitor-charging circuit: the input is connected through a resistor, which is then shorted out by a relay once the capacitor is charged.


One of the filter inductors, along with the bridge rectifier heatsink.


The input capacitor is rated at 330 µF, 450 V. To its right is what appears to be the inverter transistor heatsink (20N65), and farther right is the power-factor-correction inductor and its power stage.


A whole module packed with ICs controls the entire setup.


Alongside the analog circuitry, there’s a digital section too; you can even see an EEPROM chip on the left. Unfortunately, I won’t be able to use all the power supply’s features, since it has digital control and can probably report information about its operating modes.
You can also see five trimmer potentiometers here, with special access holes in the protective cover. I’ve come across information online about modifying a similar power supply to output 13.5 V, for example, to power transceivers.
There’s another trimmer near the output terminals, but as far as I can tell, the output-voltage adjustment trimmer is actually on the control module.


The output transformer has a somewhat unusual design. It looks like the output winding is made from a busbar.


The output rectifier. I’ll assume it uses synchronous rectification, since I couldn’t get a closer look. The transistors are mounted on copper plates.


This appears to be the auxiliary power transformer, at least judging by how it is connected in the circuit. That seems especially likely since there’s a K3532 transistor on the high-voltage side right next to it.


Two things here surprised me:
1. The output capacitors have a total capacitance of just 5900 µF: 2200+2200+1500, rated at 16 V.
2. The overheat protection thermistor isn’t near the components that get hot, but where the relatively cool air enters. I assume this isn’t the only thermistor, but I couldn’t find another one.


Logically, the next step would be to show the redrawn schematic, but I didn’t draw one, for several reasons.
1. It’s difficult.
2. It’s expensive: the time spent reverse-engineering is worth money, and it would take a lot of time.
3. It doesn’t make much sense, since at this price it’s easier to replace the unit than repair it.

As an extra, I removed the protective fabric tape from the output pads and scraped off the thermal glue. Nothing unusual, except that the output wires are soldered to only one side of the board. There are identical pads on the other side, and for a more reliable connection it’s better to draw current from both sides.


The solder marks on the other side are mine; I connected a multimeter.


First power-up: nothing blew up or sparked. The no-load output voltage was 12.3 V, and power consumption was about 7 W. The fan sped up a little at first, but quickly switched to “quiet” mode, where it was barely audible.


First, let’s check the following parameters:
1. Load capacity
2. Output voltage versus load
3. Efficiency
4. Power factor

For all of this, I used two electronic loads, a clamp meter, a multimeter, and a wattmeter.


I checked all the parameters at load currents in 3.8 A increments up to 38 A, then tested at 38.3 A (the rated current) and 40 A. The power supply performed excellently in every test.
Below are four tests at 3.8, 19, 38.3, and 40 A.



The power supply handled the full load and even a little more without any trouble. The problems started on my end, because the load wasn’t split evenly between the two electronic loads, so one of them started overheating.


After a while, the fan kicked in at full speed, and no-load power consumption immediately jumped from 7 W to 10 W. I could hear it too.... Silent, Super Silent, and similar features are all the rage these days. Well, forget about that here: this thing is seriously loud, easily competing with the noise from the four fans in my two electronic loads.

The maximum input power was almost 540 W. I accidentally set the current a little above 40 A; the input voltage during the tests was 227–228 V. It would have been good to test at a low voltage too, of course, but I temporarily don’t have that option at the moment.
Input current and power factor at no load, 50% load, and 100% load.


The same data in numbers and as a graph. 11 measurement points are shown, from 10% to 110% load.
Maximum efficiency is reached at 60% load and is 92%. However, keep in mind that the current was measured with a clamp meter, which isn’t designed for high measurement accuracy in the first place.


The power factor over the same range: it’s already 0.88 at 10% load, then rises to 0.96 and stays around that level, reaching a maximum of 0.98 at 110% load.


The hardest part of measuring ripple was... fitting everything on the table. In the end, I had to move the oscilloscope to the right of the power supply.


I once wrote that I can find problems even in a branded power supply. Well, in my opinion, the ripple here is a bit too high.
Oscillograms at no load, 33%, 66%, and 105%. You can see that the ripple amplitude doesn't depend much on the load and stays in the 500–600 mV p-p range.


The 100 Hz ripple, on the other hand, isn't all that high, partly thanks to the active power factor correction.


I suspect the output capacitors may have degraded, but they're difficult to access without desoldering them, and I don't have anything suitable at home to replace them with right now. So I simply connected another capacitor in parallel across the output terminals—a Panasonic 820 µF, 35 V—and the results changed dramatically.
Measuring ESR by simply connecting the meter to the power supply output gave an absurd reading of 5 Ω, which seems very unlikely in this case, but I'd still recommend replacing them just to be safe.

Load current: 10 A (25%); oscilloscope set to 50 mV per division. The left image is without the extra capacitor.


Naturally, I also tested it under load, using the same instruments plus a thermal camera.


The test was carried out in three stages at load currents of 13, 26, and 39 A—slightly above the rated 38.3 A. Each stage lasted 20 minutes, for a total test time of 1 hour.


I could only monitor the temperature with a thermal camera. The main measurement was the temperature of the exhaust air; I also checked the area around the power transformer and output rectifier, so I stuck a small piece of electrical tape there.
1. Test start: exhaust air at 35.8 °C.
2. After 20 minutes at a load of about 160 W, the temperature was 38.5 °C. I don't think the power supply even noticed we were trying to heat it up.
3. After 20 minutes at a load of 315 W, the temperature was 33.4 °C. That was unexpected...


The drop in temperature is easy to explain. Below are two thermal images taken 10 minutes into the maximum-load test. The fan is running slowly, and the air temperature is around 41–42 °C.
A couple of minutes later, the fan kicked in. By the end of the test (8 minutes later), the air temperature was 35.3 °C, even with a load of 470 W!

At this point I realized there was no point in running tests like this on this power supply. Its cooling system can handle more heat than the unit produces, so the fan may turn on at 50% load and stay off at 100%. The control logic goes like this: once it gets hot, the fan kicks in at full speed; after about 10 minutes, it switches back to quiet mode.


Overall view of the test setup after an hour of operation, and the temperature of the power connector. The connector temperature also depended on the fan, fluctuating between 55 and 60 °C at 39 A. That's a bit high in my opinion, since this connector is rated for up to 65 A, if I'm not mistaken.


The output voltage is practically unaffected by temperature: the cold power supply is on the left, and the one after the 39 A test is on the right.


A side-by-side photo of two power supplies with a similar design. They have almost the same output current—40.6 A for one and 38.3 A for the other—but different output voltages and, accordingly, power ratings: 48 V/2000 W and 12 V/460 W.


Update.
A regular reader of mine, DENS17, bought a similar power supply on AliExpress - link and kindly provided photos of its internals. Many thanks to him.
A visual inspection showed that this power supply is also well made. Although it differs from the one featured in the review, its build quality is just as high, so I can recommend it as well. There's more information in the comment below the review.



More photos under the spoiler


Well, what can I say...
This is definitely an interesting power supply: powerful, compact, with a popular output voltage, and a genuine, well-made unit. But keep in mind that it’s used. To be fair, I expected to find a pile of dust inside, or even a cheap Chinese circuit board. But you saw it for yourselves: everything is clean, neat, and very well made. Still, this power supply is 9 years old (the date on the label inside is 1031, most likely week 31 of 2010).

Of course, it has its drawbacks too. The first is ripple. It doesn’t matter for relatively undemanding loads (or for servers, which have separate converters on their circuit boards), but in some situations it can cause problems, so bear that in mind. An experiment with adding an output capacitor showed that it sharply reduces the ripple level; I’ve added the details to the relevant section of the review. Some of the interference is also filtered out by the RC filter formed by the power supply cable and the load’s input capacitance.

And then there’s the noise. Yes, friends, I’m going to spell it with a capital letter here: Noise. No, even better: NOISE :) It really is loud, but mainly at power levels above 200 W, and only intermittently. It probably helps that it goes something like this: 10 minutes quiet, 10 minutes noisy, rather than constantly ramping the fan up and down like a laptop. The culprit is the small size and the intended use, where active cooling is practically standard.


That’s all from me. I hope you found this useful. Best regards, your kirich.