The 2014 El-Cheapo Power Supply Roundup

Numan AT-580H

Numan is a brand I’ve never come across before, so I really have no idea what they are like. Given that this unit was purchased for only $23 on eBay, though, my expectations aren’t high.

The label claims that there are two 12V rails, but this is not true; as we will see later, it is a single rail unit. The label does say that this unit is only rated for 580W peak power, but does not specify what the continuous rating is. I generally assume it to be about 50W less if that is the case, so we will assume the continuous rating to be 530W. The fan grille is a little unusual. Most of the cheap power supplies with 120mm fans use wire grilles, but this one uses a punched out honeycomb grille. It is not much more restrictive, but this does bring back memories of a somewhat similar looking power supply reviewed by Oklahoma Wolf over at JonnyGuru, which was an absolute disaster of a unit.

Test Results

Test 1 (74.34W Load)

Rail	Load	Voltage	Ripple
12V	2.4A	12.11V	12.2mV
5V	5A	5V	10.0mV
3.3V	4.96A	3.37V	25.8mV
−12V	0.09A	−11.25V	14.4mV
5Vsb	0.5A	5.02V	12.6mV
AC Power	97.7W
Efficiency	76.09%
Power Factor	0.63
Intake Temp	35°C
Exhaust Temp	41°C

.
Test 2 (102.45W Load)

Rail	Load	Voltage	Ripple
12V	4.74A	12.04V	14.0mV
5V	5.01A	5.01V	9.6mV
3.3V	4.96A	3.37V	25.0mV
−12V	0.09A	−11.33V	15.8mV
5Vsb	0.5A	5.01V	13.8mV
AC Power	130.3W
Efficiency	78.63%
Power Factor	0.65
Intake Temp	35°C
Exhaust Temp	43°C

.
Test 3 (156.02W Load)

Rail	Load	Voltage	Ripple
12V	9.28A	11.92V	19.4mV
5V	5.03A	5.03V	10.0mV
3.3V	4.93A	3.35V	25.6mV
−12V	0.1A	−11.48V	20.2mV
5Vsb	0.5A	5.0V	16.0mV
AC Power	195.1W
Efficiency	79.97%
Power Factor	0.66
Intake Temp	36°C
Exhaust Temp	44°C

.
Test 4 (199.48W Load)

Rail	Load	Voltage	Ripple
12V	9.28A	12.07V	22.8mV
5V	9.8A	4.9V	17.6mV
3.3V	10.06A	3.32V	40.2mV
−12V	0.1A	−11.76V	24.4mV
5Vsb	0.99A	4.95V	23.4mV
AC Power	249.2W
Efficiency	80.05%
Power Factor	0.66
Intake Temp	37°C
Exhaust Temp	48°C

.
Test 5 (252.96W Load)

Rail	Load	Voltage	Ripple
12V	13.83A	11.98V	32.8mV
5V	9.8A	4.9V	20.8mV
3.3V	10.03A	3.31V	37.6mV
−12V	0.1A	−11.97V	34.0mV
5Vsb	0.99A	4.93V	27.6mV
AC Power	319.7W
Efficiency	79.12%
Power Factor	0.63
Intake Temp	37°C
Exhaust Temp	51°C

.
Test 6 (301.21W Load)

Rail	Load	Voltage	Ripple
12V	18.02A	11.87V	33.8mV
5V	9.84A	4.92V	21.4mV
3.3V	9.97A	3.29V	38.2mV
−12V	0.1A	−12.17V	34.4mV
5Vsb	0.99A	4.93V	29.4mV
AC Power	390.7W
Efficiency	77.09%
Power Factor	0.64
Intake Temp	38°C
Exhaust Temp	54°C

.
Test 7 (354.89W Load)

Rail	Load	Voltage	Ripple
12V	22.66A	11.78V	47.2mV
5V	9.9A	4.95V	26.0mV
3.3V	9.79A	3.29V	?
−12V	0.1A	−12.45V	43.8mV
5Vsb	0.99A	4.93V	33.6mV
AC Power	487.1W
Efficiency	72.86%
Power Factor	0.63
Intake Temp	39°C
Exhaust Temp	59°C

.

The 12V rail started at 12.11V and dropped to 11.78V in Test 7, giving us 0.22V (1.83%) worst case regulation, and a 0.33V (2.75%) drop. The 5V rail had maximum and minimum values of 5.03V and 4.92V respectively, giving us 0.08V (1.6%) regulation and 0.11V (2.2%) variation. The 3.3V rail started at 3.37V and dropped to 3.29V, giving us 0.07V (2.12%) regulation and a 0.08V (2.42%) drop. These results are well in spec and perfectly acceptable, although not amazing.

The efficiency just barely exceeded the 80% mark in Test 4, but was below 80% for the rest of the testing. Admittedly, the weather was extremely hot on the day I tested this unit, and my workshop was at 39°C during Test 7 (yes, you read that right – I was working in 39°C/102°F heat). This may have made the efficiency numbers a bit worse. Speaking of temperatures, the exhaust was 6°C warmer than the intake to start with and 20°C warmer in Test 7, which is quite high. The power supply was also not able to make it up to 580W. It didn’t even make it to 530W. After a couple of minutes at 350W load it blew up, and the explosion was quite spectacular – one of the biggest I’ve seen for a while now.

Rail	Test 5 (252.96W)	Test 6 (301.21W)
12V
5V
3.3V
−12V
5Vsb

.

The ripple suppression was acceptable, but it was just above half the maximum on the 5V, 3.3V and 5V_SB rails. Unfortunately, the power supply failed before I could get screenshots or read the 3.3V rail’s ripple value in Test 7.

Disassembly

The input filtering consists of two X capacitors, two common-mode chokes and four Y capacitors, which is enough components. The primary capacitors are branded Cheng, and are rated for 680µF, but I tested their value after de-soldering them. One read 493.7µF and the other read 469.1µF – far below the allowable 20% tolerance. This suggests that they are actually re-sleeved 470uF parts. The primary switching transistors are a pair of 13009s, presumably rated for 12A. The 5V_SB uses a two-transistor circuit, with a Silan Microelectronics SVF2N65F MOSFET rated at 2A as the main switching transistor and a 2SC1384 as the smaller transistor. Unfortunately, the critical capacitor is branded H.Q. – not one of the high quality manufacturers.

The 12V rail uses a U30D20C rectifier. I can’t identify its manufacturer, but it probably has similar specifications to the MOSPEC U30D20, which would mean that it is a 30A fast recovery rectifier. The other rails both use an MBR2045CT rectifier. Again, I couldn’t identify their real manufacturer, but they are probably similar to the ON Semiconductor MBR2045CT, meaning that they are 20A Schottky rectifiers. The capacitors are mostly from ChengX, with a few from H.Q. and Asia’X. None of these are high quality parts. The secondary side controller IC is a Silan SD6109. It does support Over Power Protection, but obviously it was either not implemented or was set too high. It does not support Over Current Protection, and there is only one group of 12V wires, meaning that this is definitely a single rail design.

Another problem is that one of the four screws holding the PCB to the case was missing from the factory. The damage to the PCB around the screw hole suggests that a screw was previously there, and was over-tightened. Sure enough, I tried putting one of the other screws in, and the thread on the hole was stripped, it seems that the manufacturer then left the screw out so that it wouldn’t fall out damage or short anything. The other three screws were also slightly over-tightened.

Both of the switching transistors were severely damaged in the explosion, with one almost split in half. Notice the melted washers and the soot on the heatsink, 5V_SB transistor and PCB.

After removing the switchers from the heatsink, it was evident that they had also blown out through the back. Both of them had holes burned through the backs and the electrical insulation pads, and one of them had even burned a hole in the heatsink.

 

The fan may be branded Legend, but there is nothing legendary about it. It was only quiet for the first few tests. It was quite noticeable by the time we hit 250W load, and it was making some strange clanking noises, as if the blades were hitting something (which does not appear to have been the case – suggesting that it was bearing noise). The heatsinks aren’t too bad – they are reasonably thick and have a reasonable amount of surface to air contact.

 

Specifications and Conclusions

Real Wattage	300W
OEM	Unknown
PFC	None
Price	$23 AUD
ATX Connector type	20+4 pin
Worst-case voltage regulation (12v, 5v, 3.3v)	1.8%, 1.6%, 2.1%
Worst-case ripple (12v, 5v, 3.3v)	47.2mV, 26.0mV, 40.2mV
Worst-case efficiency	72.86%
Input filtering	Adequate
CPU Connector	ATX/EPS12V (4+4 pin)
PCIe Connectors	1x 6 pin
Molex (Peripheral) Connectors	2
FDD Power connectors	None
SATA Power connectors	3

.

Pros: Fireworks display was entertaining

Cons: Can’t deliver labelled rating (−2), Low quality capacitors (−2), Low quality fan (−1), Loud (−1), Missing and over-tightened screws (−1), Mediocre ripple suppression (−0.5), Mediocre voltage regulation (−0.5)

Score: 2/10

< Previous Page | Next Page >

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