Our testing methodology for testing Power Supply Units is limited to the multi-meter to show each rails performance due to us not having access to an ATE load tester. We are making plans to invest in equipment as we were directed by Cooler Master when Simon visited their test labs in June 2010.

Although this does constrain us slightly we can still monitor fluctuations on the rails and give the unit a good assessment when we feature it in one of our builds. The way the testing will run is simple, voltages will be observed at idle wattage and then a single GPU will be installed within the system and a combination of OCCT and MSI Kombustor will be executed to provide simulation with just 1 GPU. This will be run for 10 minutes. After this another graphics card will be added to the system giving the setup a dual GPU configuration and again the stress tests will be applied and after 10 minutes voltages monitored. In both instances the CPU will also be overclocked to maximise power consumption and strain. This should give us a good indication of how the Corsair TX850 V2 will perform under gaming stresses and beyond. This is a severe load test indeed and a good representation of real-world analysis.



System Components
CPU Intel Core i7 920 @ 4GHz
Motherboard ASUS Rampage III Formula X58
Graphics 2 x NVIDIA GTX 480 (SLI)
Memory 12GB Kingston HyperX 1800MHz CL9
HDD 2 x Samsung Spinpoint 500GB SATA-II
PSU Corsair TX850 V2

I deliberately configured SLI within this test system to give the PSU as much of a draw as it could take. We all know how Fermi is notorious for its big power draw! As I mentioned there is no ATE load tester available so generating as much power as possible is key to observing the fluctuations on the rails.

For voltage tolerance levels, ATX specify that a 5% variable fluctuation is acceptable. (Form Factors - ATX Specification PDF - Page 10). So what this means is:

3.3V Rail - 3.135V - 3.465V acceptable
5V Rail - 4.75V – 5.25V acceptable
12V Rail - 11.4V – 12.6V acceptable



Idle test - The system settles in windows without any applications or stressing. 1 GPU (217W)
Single GPU - OCCT/MSI Kombustor run for 10 minutes (421W)
Dual GPU - SLI is setup for 2 x GTX 480. OCCT/MSI Kombustor run for 10 minutes (752W)

Firstly, its good to see that all the voltages are within the ATX standards for tolerance levels - being within 5%. The 3.3V rail typically used for low powered peripherals did not sustain any significant voltage fluctuation at all. Just 0.02v difference with 2 GPUs loaded plus the system. The 5V rail similarly had very little oscillation. This stayed within 0.04v. The one we really want to pay attention to though is the 12V rail. This is the rail that provides the juice for the graphics card and needs to be able to sustain plenty of voltage load not just for one GPU but for two. Here the performance is once again solid. Bear in mind that MSI Kombustor is a comprehensive and thorough stress testing tool and puts enormous pressure on the GPU's. The results are pleasing - just 0.05v difference between idle and using just the 1 GPU and then with 2 GPUs being fully loaded we have 0.07v difference between idle. This is within the specifications and under 50 and 88% load can be deemed as excellent. The TX850 V2 holds out very well.

These results clearly show that a sustained stress on the PSU causes very little fluctuations and the consumer can rest assured there will be no sporadic performance hits. We have a reliable PSU in the Corsair TX850 V2. During the tests the Yate Loon 140mm fan delivered a quiet cooling solution. Even under 50 and 88% load the noise levels from this unit were not disruptive by any means, clearly Corsair have done well in opting for the Yate Loon fan.