The 2014 El-Cheapo Power Supply Roundup

A-Power P4-A680

The name A-Power may very well remind you of a unit we tested in the 2012 Roundup. That was one of the worst power supplies we have ever seen. Let’s see if the 120mm fanned version can do any better.

This unit is rated identically to it’s 80mm-fanned cousin, which is a little worrying, as it suggests that it may be similar internally. This unit certainly looks better than than the other A-Power, however. It has a 120mm fan and is black in colour.

Test Results

Test 1 (73.12W Load)

Rail	Load	Voltage	Ripple
12V	2.39A	12.07V	17.6mV
5V	4.95A	4.95V	91.8mV
3.3V	4.88A	3.32V	19.6mV
−12V	0.1A	−11.71V	36.2mV
5Vsb	0.49A	4.92V	27.4mV
AC Power	104.0W
Efficiency	70.31%
Power Factor	0.59
Intake Temp	27°C
Exhaust Temp	32°C

.
Test 2 (100.25W Load)

Rail	Load	Voltage	Ripple
12V	4.69A	12.24V	31.6mV
5V	4.95A	4.95V	93.6mV
3.3V	4.87A	3.31V	20.0mV
−12V	0.1A	−11.88V	43.4mV
5Vsb	0.49A	4.91V	29.4mV
AC Power	137.5W
Efficiency	72.91%
Power Factor	0.58
Intake Temp	27°C
Exhaust Temp	33°C

.

Test 3 (151.1W Load)

Rail	Load	Voltage	Ripple
12V	9.11A	11.75V	43.6mV
5V	4.96A	4.96V	97.2mV
3.3V	4.82A	3.28V	35.2mV
−12V	0.1A	−12.2V	62.2mV
5Vsb	0.49A	4.89V	33.2mV
AC Power	203.4W
Efficiency	74.29%
Power Factor	0.55
Intake Temp	27°C
Exhaust Temp	37°C

.

Test 4 (194.22W Load)

Rail	Load	Voltage	Ripple
12V	9.28A	11.97V	55.2mV
5V	9.54A	4.77V	86.0mV
3.3V	9.79A	3.23V	28.6mV
−12V	0.11A	−12.76V	78.2mV
5Vsb	0.97A	4.83V	35.2mV
AC Power	265.5W
Efficiency	73.15%
Power Factor	0.55
Intake Temp	28°C
Exhaust Temp	41°C

.
Test 5 (242.41W Load)

Rail	Load	Voltage	Ripple
12V	13.54A	11.8V	66.8mV
5V	9.54A	4.77V	88.8mV
3.3V	9.7A	3.2V	34.0mV
−12V	0.11A	−13.22V	107.6mV
5Vsb	0.96A	4.82V	38.4mV
AC Power	342.1W
Efficiency	70.86%
Power Factor	0.55
Intake Temp	28°C
Exhaust Temp	47°C

.

The 12V rail read 12.07V in Test 1 and dropped as low as 11.75V in Test 3. This gives us 0.25V (2.08%)  worst-case regulation and a 0.32V (2.67%) variation. The 5V rail read 4.95V in the first two tests, but in the last tests, it read 4.77V – just barely above the minimum in ATX specifications. We have worst-case regulation numbers of 0.23V, or 4.6%, and a drop of 0.18V, or 3.6%. The 3.3V rail started at 3.32V and dropped to 3.2V in Test 5. This gives us 0.1V (3.03%) regulation and a 0.12V (3.64%) drop. Averaged out, we have 3.24% regulation and 3.03% variation. This is what we would consider to be a mediocre result. Yes, it’s in spec, but not by a large margin, especially on the 5V rail.

The efficiency on this unit was woeful. Not only did it never manage 80% (the minimum I require not to remove any points), but it didn’t even achieve 75%. At 300W, the exhaust air was a full 19°C hotter than the intake – making it hottest-running power supply I have tested to date – even hotter than the TFX form factor In Win IP-S300FF1-0 at the same 250W load. Given the efficiency numbers, though, I can’t say I am surprised. This unit was unable to deliver more than 250W. I did something a little different with the load pattern this time and increased the load on the 5V rail to 20A, as I figured it probably had more headroom and would give us a better chance of the primary side failing first and a fireworks display. Unfortunately, though, the 5V rectifier burned out, which is interesting. If the rectifier is anything like the one in it’s 80mm-fanned cousin, it should have handled 20A load with room to spare. We’ll find out for sure what it is soon.

Rail	Test 4 (194.22W)	Test 5 (242.41W)
12V
5V
3.3V
−12V
5Vsb

.
The ripple was above the maximum allowed limit on the 5V rail throughout the testing. The other rails all had acceptable ripple levels, even if they were a little on the high side. This is a similar result to what we saw on this unit’s 80mm-fanned cousin.

Disassembly

Internally, this power supply looks almost identical to the A-Power Elite 680W, with the exception of a few of the components. The input filtering is exactly the same, consisting of only a single non-safety-rated ceramic capacitor, which is both completely inadequate and a safety hazard, as it can short to the case if it is overstressed by a power surge. It has the same 2A bridge diodes, but the capacitors are rated 470µF, as opposed to 330µF (although we suspect that these may be 330µF parts in reality, and that they may be over-rated). The switching transistors are simply marked 13007, and are presumably rated for 8A. As I said last time, these parts are ridiculously undersized for a 680W product; 250W is about all that they are usually capable of.

The 12V rail uses an F16C20C rectifier, which is a Fast Recovery Rectifier rated at 16A. It’s better than the 12A in the other unit, but still very undersized, considering what the label claimed about the 12V output. The 5V rail uses the STPS3045CW, and the 3.3V rail uses linear regulation from the 5V output, with a PHP45N03LT MOSFET, which is rated at 45A. Linear regulation, as is used here, works by using the MOSFET as a dynamic resistor in series with the load to lower the voltage. It is very simple and gives a very clean output, but it is extremely inefficient. The capacitors are mostly from ChengX, with the exception of one, which is branded EU Cap. Both of these are obscure brands, and not high quality Japanese parts.

I don’t normally comment much on the soldering in these roundups, but I felt compelled to after seeing how bad it was on this unit. It looks as though someone burned off a pad when soldering in the controller IC and used a piece of wire to connect the pin to the next joint.  Not only that, but there is no insulation on the wire, and it is too close for comfort to neighboring joints. This is not the kind of workmanship we would expect to see from the factory.

The fan is another cause for concern. As soon as I took the cover off, I noticed that there was oil all over the label. Peeling it off revealed that the rubber plug was not inserted properly, and had allowed oil to leak from the bearing. The fan is wired directly to the 12V output and was noticeable throughout the testing. The heatsinks have a reasonable amount of surface to air contact, but they are extremely thin. As was the case with the 80mm fan version, I can bend and snap them with my bare hands. I believe that this is why the 5V rectifier failed at a much lighter load than normal; the heatsinks are too thin to conduct the enough heat from the rectifiers to the fins in order to keep the parts from overheating.

Specifications and Conclusions

Real Wattage	0W
OEM	Unknown
PFC	None
Price	$30
ATX Connector type	20+4 pin
Worst-case voltage regulation (12v, 5v, 3.3v)	2.1%, 4.6%, 3.0%
Worst-case ripple (12v, 5v, 3.3v)	66.8mV, 97.2mV, 35.0mV
Worst-case efficiency	70.86%
Input filtering	Inadequate
CPU Connector	ATX12V (4 pin)
PCIe Connectors	1x 6 pin
Molex (Peripheral) Connectors	2
FDD Power connectors	1
SATA Power connectors	3

.

Pros: None

Cons: Can’t deliver anything with ripple in spec (−10), Inadequate input filtering with unapproved parts used, Low quality capacitors, Mediocre voltage regulation, Appalling soldering quality and PCB work, Low quality fan, Very inefficient, Loud, 5V-heavy, Runs very hot.

Score: 0/10

< Previous Page | Next Page >

Pages: 1 2 3 4 5 6 7 8 9 10 11