The el-cheapo power supply round-up

Xsonic SZ-450W P4

My first impressions about this unit are actually favourable… kind of. It doesn’t have more connectors than any of the others or anything like that, but out of all the power supplies in this round up, it feels the heaviest. The front of the unit has a honeycomb style vent and through it, I can see some of the internals, and the transformer looks to be an ERL-35, which is the biggest I’ve seen so far. However, it does have thin T-shaped heat sinks.

_

Sorry, but I’m not going to show you what I saw through the front vents just yet. As usual, I’m saving that for later. What we see here is the usual sad story for cheaper power supplies. We have more current available on the 5V rail than the 12V rail, only 1 SATA and 4 peripheral connectors and thin 20AWG wires.

Load Testing

Test 1 (115.77W Load – Cold)

Rail Load Voltage Ripple
12V 4.84A 12.09V 15.2mV
5V 5.01A 5.01V 11.0mV
3.3V 9.88A 3.26V 12.4mV
−12V 0A −12.03V 14.2mV
5Vsb 0A 5.08V 11.6mV
AC Power 145.23W
Efficiency 79.72%
Power Factor 0.64

 

Test 2 (195.99W Load – Cold)

Rail Load Voltage Ripple
12V 9.66A 12.07V 22.8mV
5V 9.82A 4.91V 13.4mV
3.3V 9.73A 3.21V 14.2mV
−12V 0.1A −12.14V 24.8mV
5Vsb 1A 4.99V 16.2mV
AC Power 246.17W
Efficiency 79.62%
Power Factor 0.62

 

Test 3 (252.19W Load – Cold)

Rail Load Voltage Ripple
12V 14.48A 12.07V 27.6mV
5V 9.88A 4.94V 14.8mV
3.3V 9.3A 3.07V 15.6mV
−12V 0.1A −12.28V 30.4mV
5Vsb 1A 4.99V 16.8mV
AC Power 311.27W
Efficiency 81.02%
Power Factor 0.62

 

Test 4 (302.24W Load – Cold)

Rail Load Voltage Ripple
12V 16.77A 11.98V 32.6mV
5V 14.52A 4.84V 16.4mV
3.3V 9.7A 3.20V 17.2mV
−12V 0.1A −12.55V 37.2mV
5Vsb 0.99A 4.96V 16.5mV
AC Power 387.28W
Efficiency 78.04%
Power Factor 0.63

 

Test 5 (354.98W Load – Cold)

Rail Load Voltage Ripple
12V 21.38A 11.88V ?
5V 14.64A 4.88V ?
3.3V 9.45A 3.12V ?
−12V 0.11A −12.74V ?
5Vsb 1A 4.98V ?
AC Power 468.65W
Efficiency 75.75%
Power Factor ?

 

The unit may have felt the best, but it sure didn’t perform the best. The voltage regulation was the worst I’ve seen from any unit so far in this round up, with the 3.3V rail dropping below the minimum 3.135V allowed in ATX specifications in tests 3 and 5. The ripple suppression was pretty good for a cheaper unit. I’ve seen worse from the Hairong, but I’ve seen better from the PowerTek. The efficiency was mediocre, peaking at about 81% during Test 3. During Test 5, the efficiency started out at about 75.75%, and started dropping faster and faster as the test went on. What was probably happening is that the unit was heating up with the increased load and as it did so, the increased heat caused the efficiency to drop, which in turn created more heat, causing the efficiency to drop faster. It lasted less than a minute at 350W load before we got a new years eve fireworks display. Unfortunately, I didn’t get to read the ripple or power factor before it failed.

Rail Test 3 (258.22W) Test 4 (314.25W)
12V
5V
3.3V
−12V
5Vsb

A Look Inside

_

As I thought, it’s the best looking unit thus far. The input filtering starts at the AC receptacle. The PCB there contains a common-mode choke, an X capacitor and two Y capacitors. The main PCB contains the same components as the AC receptacle PCB as well as two MOVs and another Y capacitor after the rectifier (which is rated at 4A), and the primary capacitors are two 470μF parts branded LCZ. The two switching transistors are Sino Microelectronics D304Xs, the same parts used in the Sun Pro ATX-550W. The screws holding them (as well as the 5Vsb transistor) to the 1mm-thick T-shaped heat sink were done up so loose that I undid them with nothing but my bare fingers, and it wasn’t the plastic washers melting as only one washer was melted, but I undid all 3 screws with my fingers.

The screws on the secondary heat sink were done up a bit tighter, since I did actually need a screwdriver to undo them, but I could still wiggle the rectifiers around a bit. The problem there is that only the 12V rectifier is good enough for the labelled rating (and then only just); it’s a 16A rated STPR1620CT fast recovery rectifier. A Schottky rectifier would have been preferable, since they offer better efficiency than fast recovery rectifiers. The 5V rail has an SBL3045PT 30A Schottky rectifier and the 3.3V rail has an SBL2040 Schottky rectifier rated at 20A. Bear in mind that the label claims 36A and 25A respectively on those 2 rails, so these parts are insufficient. The capacitors are all from an obscure company called Nicon. The general rule with capacitors is that if it isn’t from one of the 6 known good manufacturers, then it’s safe to assume low quality.

_

It looks like ‘sleeve bearing’ is the brand of this fan. Whoever made this fan probably decided not to put their company name on it so that no one knows who to send it back to if it fails prematurely. Although it is temperature controlled, it was audible during test 1 and only got louder as the load and heat went up.

Actual specifications

Real Wattage 300W
OEM Unknown
PFC None
Price $30 (AUD)
ATX Connector type 20+4 pin
Worst-case voltage regulation (12v, 5v, 3.3v) 1.0%, 3.2%, 6.9%
Worst-case ripple (12v, 5v, 3.3v) 32.6mV, 16.4mV, 17.2mV
Worst-case efficiency 75.75%
Input filtering Adequate
CPU Connector ATX12v (4 pin)
PCIe Connectors None
Molex (Peripheral) Connectors 4
FDD Power connectors 1
SATA Power connectors 1

Conclusions

Pros: Good input filtering, good ripple suppression

Cons: Poor voltage regulation, Can’t deliver 450W, few connectors, noisy, small heat sinks, low quality fan and capacitors, loose screws

Bottom Line: I wouldn’t buy a power supply from someone who can’t even use a screwdriver.

Score: 3/10

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