If you cast your mind back to last year, you should remember that HEXUS published the mother of all PSU reviews. In the world's most-definitive first PSU group test, 34 of the blighters were run through the mill by HEXUS' Group Technical Editor, Ryszard Sommefeldt, with a little help from our PD. Having established a testing method during our time at FSP Group's Taoyuan facility, we used an almost identical process this time around, albeit on slightly different and more modern equipment, and a different (Ed. though not so modern?...) man in the driving seat.
HEXUS' Manager of Performance Analysis, James Smith had the pleasure of trying to blow up the PSUs this time, though in this instance we fortunately didn't have to travel outside of the UK to do so, because FSP Group's (recently departed) Product Marketing Manager, Grace Liang, had arranged for FSP Group to be the first professional PSU manufacturer we know of to acquire the requisite (expensive!) test equipment and facilities here in the UK. We streamlined testing somewhat, but the overall goal was the same.
We've said before that some of the power output ratings on certain PSUs, especially 'too-good-to-be-true', el-cheapo models are a little ambitious. It's important that a PSU be able to provide what it says it can, lest a user attempt to put it under that much load and cause a catastrophic failure. So, of course, we load-tested the PSUs to see if they could deliver as advertised.
For our load testing, input power was provided by a Chroma 61604 Programmable AC Source, which we programmed to give 230V/50Hz - the ideal UK mains supply. We used Chroma 6314 Programmable DC Load mainframes with a combination of modules that gave us testing capabilities for up to 4 x +12v rails along with +3.3, +5, -12, and +5vSB rails. The connector board we used had, along with an array of 4-pin Molex, ATX and PEG power, two SATA power connectors this time around.
Easing the PSUs in gently, we started by demanding only half of their rated load. Once satisfied that the PSUs were coping, we gradually increased the load power towards the rated output power, checking that the PSU was working properly under the load as we increased it. We stopped increasing the load when the rated output power was reached, or the PSU was unable to deal with a load level. Once at full output power, the PSUs were given a 30-minute load test.
Also key to our PSU testing was temperature readings. Using a CHY 110 infrared thermometer, we took ambient temperature readings from room walls unaffected by any heat sources. We aimed the thermometer near the mains inlet, near to where the PSU exhaust vent was located, to get PSU temperatures. Further, we took the PSU temperature at idle (before testing) and after our thirty-minute full-power load test.
Finally, ideal & actual ouput power, along with input power was recorded to calculate PSU efficiency. Ideal output power is the output power from the current draw on each rail assuming the rails are at their ideal voltage. Actual output power takes into account the fact that the voltage rails may have dropped under load (or indeed been higher).
So, without getting into any grittier a level of detail, that's how we conducted testing. On to the results.