A visit to the Enermax Lab in Hamburg – 6 PSU tested
High-end equipment and high precision measurements are not going to do you any good in the absence of an equally good methodology, one which can identify and single out both the strengths and weaknesses of each tested unit. Personally, I find none of the methodologies used by the dedicated PSU review sites 100% relevant and representative for the end user (no offense). So we had to come up with the following 7 stage load pattern for our tests:
- 10% – low load scenario, would qualify for idle values on high output units (think over 1KW), entry level units tend to display very poor low load efficiency.
- 20% si 50% – standard ECOVA 80+ certification load.
- 40%, 60% and80% – monitoring the relevant parameters as the load increases.
- 100% – full rated output, relevant for obvious reasons, for ECOVA 80+ certification as well
- Overload – an ignored aspect by most PSU reviewers, but a very important one for us. The PSU is pushed to the point the protection is triggered (OCP, OPP, etc) or the electrical parameters go out of spec (voltage or ripple noise below ATX specifications).
At each of these loads, we’ll provide current and voltage measurements together with ripple noise on four rails (two 12v rails and both minor ones, 5v and 3.3v) on the DC side. On the AC side we’ll measure the PFC and the power drawn, which will allow us to precisely calculate efficiency. For reference, the ATX limits for ripple noise are as follows: 50mV for the minor rails (3.3v and 5v) and 120mV for the 12v rails.
If you are suspicious in any way, I can tell you that all the units are retail versions, picked from the warehouse and opened in front of me. This methodology is conceived by yours truly with no input or directions meant to favour the tested units in any way.
Another typically ignored aspect is the load applied for each specific rail, which shouldn’t be random or judging strictly by the PSU label. More specifically, a 20A rating for 12v1 means nothing if that specific rail only feeds the 20+4pin ATX connector, which in turn only has two 12v wires each supporting up to 7A each. So that’s 14A total, 15A if you take into account the tolerance margins, far from the 20A rating on the label. Applying 20A on this connector will only overload the wires and the resulting voltage drops shouldn’t surprise anyone, and besides that, there’s no way on this earth that the motherboard will ever draw 20A (240W) through the 24-pin ATX connector.
That being said, it’s important to know the connector-rail arrangement before connecting them to the test board, as well as the number of wires which will support the loads so we can roughly calculate the load on each one. Today we’ll test 6 power supplies from Enermax and before moving on we’ll present the connector-rail arrangements for each one:
Enermax NAXN 450W
Enermax Revolution87+ 850W
Triathlor ECO 650W
Enermax Revolution X’t 530W
Enermax Platimax 1350W