The el-cheapo power supply round-up

Okia 420ATX

Well folks, I’ve saved the best (or perhaps worst) power supply until last. This one makes all of the others in this round-up feel heavy. The ventilation on the case also looks very poor, with only a few slots on the front of the case and a poorly designed fan grille at the back, and I really don’t like that grainy and unfinished look to the case. Although Okia power supplies aren’t available here in Australia and this one came from the US, they are available for only $10USD in the USA. My prediction is that it will fail miserably.

_

Another 5V heavy design. I’ve said what that means enough times now. Unfortunately, the UL number on the label (E162369) returns no results at the UL website, which usually means that the logo is forged or that the manufacturer has gone out of business. However, Okia/Broadway Com Corp only use a couple of OEMs – CWT and an obscure Chinese company called Yue-Lin. Maybe a look at the internals will tell us which one made this.

Load Testing

Test 1 (112.33W Load – Cold)

Test 2 (196.43W Load – Cold)

Oh dear. There goes my bottom jaw. That is officially the most appalling performance I’ve seen from any unit on here on hardwareinsights.com. 200W was as much as I could pull from it. I asked it for 250W and one of the secondary rectifiers burned out. Unfortunately, this means no explosions and that it would have killed the motherboard. At least the resistors in my load tester can handle a bit of rough treatment from cheap units like this. However, not being able to deliver more than 200W is the least of its problems. The voltage regulation was just so-so for all of the rails except for the 3.3V, where it was absolutely awful. The voltage there was below the minimum allowed, even during Test 2 when the load on all rails was almost equal. The ripple suppression was equally appalling. It was out of spec on all of the major rails during test 1 and on all but the −12V rail during test 2, although even the −12V rail wasn’t in spec by a large margin. It’s not just 1 or 2mV over where your hardware wouldn’t even notice the difference, we’re talking over three times the maximum allowable limit here. I even had to adjust the scale here on the oscilloscope. Normally, each line going horizontally across the graph represents 20mV, but it was off the scale at that setting, so on the shots below, each line represents 50mV. Such poor performance can not only make a PC unstable, but significantly shorten its life. In fact, I can’t even consider this a 200W unit, since we consider the maximum wattage to be what it can deliver in spec. OK, let me put my bottom jaw back on and we’ll pop the cover.

Rail	Test 1 (112.33W)	Test 2 (196.43W)
12V
5V
3.3V
−12V
5Vsb

A Look Inside

_

There goes my bottom jaw again. The “input filtering” consists of just a single 2kV ceramic capacitor (not a Y2 type as required to meet safety standards) after the primary rectifier, which probably explains why I never see them in Australia. Without proper input filtering, they are illegal here. The PCB is screened for more input filtering components, but they weren’t included to cut costs (and corners). There are four 2A diodes instead of a single-piece bridge rectifier. Tied with the PowerTek PTI-400AR-T, this unit has the smallest primary capacitors – only 330µF. The two switching transistors are Sanyo 2SC5763s rated for just 7A, lower than the switchers on any other power supply in this roundup, although in this case, one of the rectifiers on the secondary side failed first which saved them, but if it had been in a PC, it would have damaged the motherboard.

Speaking of the secondary side, the 12V rail doesn’t have a proper rectifier. Instead, it uses two FR302 Fast Recovery diodes soldered on to a bracket. They are rated for 3A each, so that means that the 12V rail is only good for a pathetic 6A, not enough to power a modern PC and only a third of the 18A claimed by the label. The 5V rail uses an STPS1545CT rectifier rated at just 15A, which is nowhere near enough for the 40A claimed by the label. The lack of a separate toroid coil for the 3.3V rail usually means that the power supply uses DC-DC conversion from the 5V or 12V rail, but not this time around. Tracing the circuit out reveals that it uses the main toroid coil, which is unusual and not recommended. The rectifier is an STPS10L45CT, rated at just 10A. All up, this is enough to allow the unit to deliver about 200W (including the 5Vsb and −12V rails), so it’s little wonder I burned the 5V rectifier out by loading it to 250W. The secondary capacitors are all from the dreaded Fuhjyyu. They are only 470µF, which is the smallest used of all units in this roundup. The resulting ripple problem was only made worse by the fact that the PI filter coils are all missing and have jumper wires in their place.

_

The fan is from Ruilian Science, the Fuhjyyu of the fan world. You may have noticed that I decided to take the PCB of this unit out for shots, mainly just for a better view of the internals, or rather, the sheer lack thereof. The OEM isn’t immediately obvious, but looking at a similar Okia with green tape everywhere suggests that Channel Well are behind this unit. Now I can see why they forged the UL number – because they don’t want people to know that they are capable of making power supplies this bad, although this doesn’t necessarily reflect badly on them. CWT build to order, meaning that if they built a unit this bad, it was because they were asked to by Okia. Let me get my bottom jaw back on (again) and we’ll score it.

Actual specifications

Conclusions

Pros: None whatsoever. Not even a cheap price justifies such poor performance

Cons: Couldn’t even deliver 250W, Almost no input filtering, No SATA connectors, Only has a 20 pin ATX Connector, Low quality fan and capacitors, No PFC, Only 6A available on the 12V rail, Poor voltage regulation, Poor efficiency, Appalling ripple suppression, Noisy, Looks ugly, Makes my bottom jaw fall off.

Bottom Line: This is not a power supply. It’s a motherboard killer.

Score: 0/10

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