Just how to test?
Testing power supplies used to be a simple business of placing the PSU into a fully-built system and taking a few measurements, stating that it seemed to work to specification after running Prime95 and a game or two.
How to test?
However, with PSU firms now devoting increasing amounts of money to research and development, all the while differentiating on what appear to be minor differences, a scientific, empirical approach is called for. This is why we have invested in our own power-supply-testing equipment, housed at the offices, and used to test supplies with far more accuracy. As such, we have the following equipment at our disposal:
Chroma programmable DC electronic load model 6314A
Chroma 63103a and 63107a load modules
Associated test harness for PSUs
iDRC CP-240 power analyser
Tektronix TDS 2012B oscilloscope
PCE-T 800 8-channel thermometer/data logger
Voltcraft VC170 digital multimeter
By using the above equipment we can measure efficiency and regulation quality across any range of loads, the AC ripple across the various lines at various loads, and temperature of the unit when under the cosh.
So what to test?
Efficiency and regulation
The next question is what to test? The equipment gives a near-infinite number of testing scenarios that can be undertaken. We think it makes sense to look at how the unit performs across a wide range of loads. We're choosing, somewhat arbitrarily, loads of 10, 25, 50, 75, and 100 per cent, meaning that supplies of differing capacities need to have the 3.3, 5, 5VSB, 12V and -12V lines loaded with specific amperage ratings; we do the calculations so you don't have to. Better supplies have higher efficiencies across the board.
Regulation shows just how well the PSU keeps to the voltage standard with varying loads. You don't want the 12V line dropping to, say, 11.2V when taxed.
But you also want to know AC ripple, or just how much the voltage varies when changing from mains power (AC) to PSU power (DC). This transformation from AC to DC (no heavy-metal reference, note) is caused by non-perfect suppression of the AC wave. Put simply, the lower the ripple the better, and a calibrated oscilloscope is used to measure just how much ripple exists on the various voltages lines. We use what's known as the peak-to-peak millivolts value (mV p-p). Better supplies have lower ripple and better suppression.
The onus on PSUs to push out huge amounts of 12V - CPU, graphics - power and relatively little 3.3V and 5V - SATA devices - juice means that loading isn't always proportional, that is, a set ratio between the voltage lines. Hammering one line and barely touching others is commonplace, and it can be tested by cross-loading. This type of examination puts specific emphasis on either the 12V or 3.3V/5V lines, to see how the supply copes with lopsided loads.
A supply that ticks all the above boxes but then gets super-toasty when doing so isn't ideal. We use the thermometer to measure temperature of the air just inside the PSU chassis, as it enters the PSU, and the temperature exiting the PSU This is done at a 75 per cent load.
We can clearly learn more than a few facets of PSU performance by looking at these four basic metrics. But there's more, so much more, that can be done, and we will be adding hold-up time and hot-box testing in the near future. Anyway, let's now blitz you with tables. Our testing is done on a 230V, 50Hz source, with an open-air, ambient 25°C temperature.