Besides non-redundant Power Supplies (PSUs), the ones used by typical desktop PCs that most users have, we also evaluate redundant PSUs that are usually installed in servers that operate all around the clock (24/7). A redundant PSU consists of a mainframe where two or more power modules are installed. If one of them fails, another module handles the load, so it continues its operation without any issues. In most cases, the minor rails into a redundant PSU are generated by DC-DC converters installed onto the mainframe, so the modules only need to deliver the main rail (12V) and standby voltage (5V).

Because all power modules of a redundant PSU are online continuously, there is no point in measuring Vampire Power because it will be increased. Moreover, the systems that need redundant PSUs are destined for constant operation, so they won’t operate for long periods in standby mode, contrary to standard desktop systems.

In server PSUs, the 5VSB rail isn’t of high importance, too, since its load and its impact on efficiency are minimal compared to the overall load. So, for the moment, at least, we won’t take into account the average 5VSB efficiency, as we do in ATX12V PSUs.

We do take into account the power consumption of cooling fans in the power modules. On the other hand, we don't take into account the fan power consumption of any fan on the mainframe. If there is one, we will power it through an external power supply and we will provide its power consumption in our reports, only for reference purposes.

Our redundant PSU efficiency requirements will get constantly updated as we evaluate more PSUs and have more data to create a complete standard. 

We apply at least 1450 different load combinations in the DUT (Device Under Test). The overall efficiency and PF readings will be the average of all measurements, which cover from 2% to 100% of the PSU’s operational range. The overall efficiency will be the average of all measurements, which cover the PSU’s entire operational range.                                                                                                                                                                 

(1) In case the PSU is close enough to the limits (0.1%), we will ask its manufacturer if it is willing to send a second sample for re-evaluation. If the second sample registers a significant performance difference compared to the first one (>0.5%), we will ask for a third one, to verify the test results. Moreover, in case the difference with the upper limit is <= 0.05% (e.g., 91.95%), we will round the number to the second digit and the PSU will be taxed to the higher category. 

(2) If the PSU doesn’t meet the PF requirement, it will drop to the class that meets the corresponding requirement, despite its efficiency performance.