There are ordinary 120mm fans from Protechnic Electric in each PSU. The junior PSU has a MGA12012HS-O25 fan with a rated speed of 2500rpm.
The medium model uses a MGA12012HB-O25 fan with the same performance and speed but running on a ball bearing instead of the HS model’s slide bearing. Both fans have translucent blue impellers, but lack any highlighting.
The senior PSU model employs a MGA12012YB-O25 fan. Unfortunately, the product catalogue didn’t work on the Protechnic Electric website so I couldn’t find its exact characteristics. But judging by the consumption current indicated on the label (half as much again as that of the MGA12012HB-O25), this is a high-performance model with a rated speed of about 3000rpm.
The fan speed diagrams are very similar between the three PSUs. This is normal because if the PSU electronics is guided by the temperature of the heatsinks, it is going to maintain the same fan speed under identical conditions even with different fans.
The speed is changing linearly, from 1150rpm at min load (50W) to 2300-2500rpm in the Epsilon 700 and 800 to 2800rpm in the Epsilon 900. The initial speed isn’t low (as opposed to the early Epsilon PSUs whose fan would start out at a speed of 825rpm) and the linear regulation is no good in terms of noise. Ideally, the fan speed should not change at all until a certain temperature. After that, it should be increasing to protect the PSU from overheat.
So, today’s FSP Epsilon PSUs are noisier than average. They are not silent at min load and become louder at higher loads.
The PSUs all have an efficiency of over 80%, up to 87% in some point of the diagram, which is an excellent result. But as I wrote above, the Epsilon architecture is conceived to achieve a high efficiency by means of reducing the loss on the rectifiers’ diodes. The power factor is over 97% through most of the load range as it typical of PSUs with active Power Factor Correction.
Alas, the new FSP Epsilon PSUs leave an ambiguous impression. On one hand, these PSUs are neatly assembled and offer all the necessary connectors. But on the other hand, the stability of the output voltages and the voltage ripple are poor and their noise can be characterized as “worse than average” whereas I know that Epsilons can be very quiet. I hope FSP Group will be working to correct the mentioned problems and reestablish its reputation.