Since we don’t currently have access to an ATE load tester, a multi-meter is used to show each of the power supplies performance on the 3.3v, 5v and 12v rails. Although we can’t do full load tests, we can provide relative information regarding variance and fluctuation of current and ripple on the rails which is integral and indicative of quality in a power supply.
To perform the above tests, the methodology will be as follows:
I7 4770k @ 4.5GHz
1 x ASUS GTX 770
8GB (2x4GB) Kingston HyperX Beast 2400MHz
1 x 250GB Samsung 840 EVO
Voltages will be monitored via a multi-meter and the AC power draw will be monitored via a power monitor when drew from the wall. For the idle test, the system will be simply booted up and let to run into windows and after 5 minutes when the power draw has leveled out, the readings will be taken.
To load and apply power to the power supply, a combination of IBT and Furmark will be run to put as much strain on the power supply as possible. After 10 minutes the readings will be taken and to ensure maximum strain, the CPU will be overclocked to 4.5GHz. Depending on the power of the power supply, multiple graphics cards could be used in line with how comfortable I feel the power supply will cope with such load.
ATX themselves specify that a fluctuation and variable of 5% is acceptable so to number crunch it means that:
3.3V = 3.135V-3.465V is acceptable
5V = 4.75V-5.25V is acceptable
12V = 11.4V – 12.6V is acceptable.
Any readings outside of these figures will be an automatic fail.
|3.3V = 3.4v|
|5V = 5.1v|
|12V = 12.0v|
|3.3V = 3.41v|
|5V = 5.12v|
|12V = 12.2v|