Test Methodology

Our new 2024 PSU review test setup consists of various interconnected components that feed measurements into a central server that's dedicated to data logging and analysis. All tests are performed with the power supply operating at 230 V / 50 Hz mains. The ambient temperature is regulated to 40 °C, to simulate an actual operating environment inside a computer case (not an open bench). A Rigol oscilloscope is used for ripple noise measurements and the noise levels are recorded using a calibrated microphone at a one-meter distance from the tested PSU. The dBA noise values are just for the PSU—there are no other noise sources in the acoustic testing environment. Multiple remote-controlled electronic loads draw power from the tested power supply. In order to be able to properly stress high capacity PSUs, we have allocated 4x 400 W for 12 V, 1x 200 W for 5 V and 1x 200 W for 3.3 V, which brings the total testing capacity to 2000 W sustained.



Our goal with this setup is to perform all the relevant tests and also build a software-defined infrastructure layer on top of it, where all the metrics are collected into a single piece of custom in-house software that enables automatic testing and that is able to provide data to other TechPowerUp APIs such as the relative performance builder and charting interface.

Each power supply is tested over a range of 5% to 110% of its rated output, in steps of 10% (5% for the first step).

Of course, these changes in the equipment and methodology mean that results from previously published reviews cannot directly be compared with the data of new reviews.