To test the power supply unit, we will not connect it to an everyday system as we would be unable to know the exact load that we would provide and wouldn’t be able to vary the load. Instead we will use power resistors which are nothing more than normal resistors but capable of consuming a good amount of power. How much power depends on the resistance and capability of the resistor, and can be easily calculated by Ohm’s law (Amperes = Voltage / Resistance, Power = Voltage * Amperage).

Hence for example a 1Ohm resistor attached to the 3.3V line will allow 3.3A through it and consume about 10W of power. When adding two of these resistors in parallel, you double the power consumption and allow 20W to be consumed as you halve the resistance this way. Adding more will increase the power consumption. Of course that is not entirely accurate on paper as the fault tolerance of these resistors is high and almost none are exact to their resistance specifications, but after using them we could calculate how much power each bunch of them would consume accurately. This way we can add fake load to any power supply unit, and on any power line we wish with good accuracy.

For testing the NorthQ 4800-400, taking into account its design and power specifications, we took the following steps:

Total Load	3.3V Load	5V Load	12V-1 Load	12V-2 Load
33% (~135W)	9A (30W)	8A (40W)	3A (36W)	3A (36W)
50% (~200W)	10A (33W)	12A (60W)	4A (48W)	5A (60W)
75% (~300W)	18A (60W)	14A (70W)	7A (84W)	8A (96W)
100% (~390W)	25A (83W)	20A (100W)	9A (108W)	8A (96W)

The room temperature when the test took place was 30 Celsius degrees. The performance of the 4800-400 does not differ from the 4775-400 much, but curiously the unit was considerably more silent, as it became loud only at 100% load. The voltage lines are stable enough, but the 3.3V line weakens considerably when the unit is heavily loaded. Not nearly below the specified standards, but enthusiasts will not like such fluctuations if they decide to appear.

dont have flash installed? then click here for a jpeg