Test Equipment and Methodology

For the testing of PSUs, we are using high precision electronic loads with a maximum power draw of 2700W, a Rigol DS5042M 40MHz oscilloscope, an Extech 380803 power analyzer, two high precision UNI-T UT-325 digital thermometers, an Extech HD600 SPL meter, a self-designed hotbox, and various other bits and parts. For a thorough explanation of our testing methodology and more details on our equipment, please refer to our How We Test PSUs - 2014 Pipeline post.

Cold Test Results

At room temperature, the Seasonic S12G 650W unit easily honors its 80 Plus Gold certification, reaching up to 94.4% efficiency at 50% load. Inside the normal operating range of the power supply, which is between 20% and 100% load, the efficiency remains above 90% at all times. In fact, if the S12G was just a little more efficient while heavily stressed, it could hit 80 Plus Platinum status.

As expected, the efficiency drops sharply when the load is below 20% (130W in this case), dropping down to 76% with a 32W load, but it goes up to 84% when the load is increased to 65W. These are considered "out of range" conditions and do not affect the 80 Plus certification of the unit, but considering many modern PCs will idle at 40-60W, it's still worthwhile to look at performance at these loads. Of course, at 32W load the difference between 75% and 85% efficiency only amounts to around 5W at the outlet.

Both the air and the heatsink temperatures might appear a little high for a unit of this capacity and efficiency operating at room temperature, but that is because the fan of the Seasonic S12G 650W PSU is being "lazy" -- or "quiet" if you prefer. The fan is thermally controlled and it looks like Seasonic decided it didn't need to spin faster under such testing conditions. During our SPL testing, the fan reaches audible levels only after the unit hits ~80% of its rated output.

The Seasonic S12G 650W PSU Hot Test Results
POST A COMMENT

77 Comments

View All Comments

  • E.Fyll - Saturday, March 1, 2014 - link

    That will happen too. Companies (and consumers) generally favor large PSUs at the moment and there are very few high performance, low wattage models. I certainly do not plan on testing high output units alone though, neither focusing on just a single "class" of products. Patience, I can only do one at the time. In time, there will be a good variety of reviews, of every output/range/class. Reply
  • pandemonium - Friday, February 28, 2014 - link

    Nice read. I went for the big brother instead about a month ago when my Corsair TX650 decided enough was enough: http://www.newegg.com/Product/Product.aspx?Item=N8... Reply
  • Talcite - Saturday, March 1, 2014 - link

    Glad to see PSU reviews have returned! Reply
  • CknSalad - Saturday, March 1, 2014 - link

    There's a 360 watt version of this Seasonic brand line. Honestly anything 660/7870 and less just get a quality 360-400 watt psu so your efficiency is also decent under non-gaming/non-intensive loads. 550 watt for a high end single-gpu system (Titan/780 TI/290/290x). GPUs in between those mentioned above a quality 450w psu is good enough. Reply
  • FriendlyUser - Saturday, March 1, 2014 - link

    I applaud the newer testing methodology for cases and PSUs. Great job.
    I am a fan of Seasonic PSUs and have been buying them for a few years. Since I started using Seasonic PSUs I have observed a very low component failure rate for my HD, GPUs etc. Maybe because of very low ripple or because of excellent voltage regulation? Anyway, worth investing in a high-end PSU.
    Reply
  • E.Fyll - Saturday, March 1, 2014 - link

    Indeed. Most people do not realize that but power quality (ripple/noise suppression mostly) is, in my opinion, by far the most important aspect of a PSU. It hardly affects the performance of the PSU itself but it has a dramatic impact on the longevity and reliability of your system's components.

    Do not be fooled however, low ripple does not mean low noise; you can easily have a low ripple signal with horrible levels of noise in it. We can only present ripple readings at the moment. We will get a super-fast oscilloscope that can filter noise out of a ripple signal soon; until then try not to confuse these two figures please.
    Reply
  • FriendlyUser - Saturday, March 1, 2014 - link

    Thanks for the clarification. I really want to see noise figures for PSUs. I would also be curious to see if PSU quality affects analog sound output quality in a measurable way. That would be a cool comparison to make in the future, if you have access to the measuring instruments. Reply
  • tynopik - Saturday, March 1, 2014 - link

    Performance when the current is normal and all the components are behaving nicely is boring.

    We want real torture tests!

    - What happens to the output in either severe brownout or overvoltage conditions? What happens when the input is unstable and bounces all over the place? Hook that sucker up to a variac and play with the knob like you're a DJ.

    - Transient analysis- aka does the output remain in spec when 10 hard drives spin up at once

    - Over-current protection- What happens if a component shorts? What if it just draws a lot more power than it should? Can you melt a cable? Can you blow a fuse? Can you hurt other components on other cables?
    Reply
  • FriendlyUser - Saturday, March 1, 2014 - link

    Torture tests are only meaningful, in my opinion, if they correspond to possible usage scenarios. Although interesting, the performance of a PSU in situations that will almost never arise should not influence a buying decision. Reply
  • E.Fyll - Saturday, March 1, 2014 - link

    And I want a high output, high frequency, programmable AC power source. Do you happen to have 20.000 USD to spare? ;)

    The VARIAC (oh, I am using one) is entirely useless for the tests that you mentioned. Simply "rotating the knob" is entirely wrong. I will be adding such tests in the future, when I have better equipment available.

    As for most of the last tests that you mentioned, these cannot be tested. They are assumptions and performing any one of such "tests" will damage the components of a PSU indefinitely, rendering it useless. For example, if I short the rectifier bridge, it will permanently damage it and probably other components of the PSU as well. Unless you can supply me with 100 units of each model, I cannot test what would happen if every given component would fail.

    Oh, and if the distibution grid around your area shifts like "a DJ is rotating the knob", which is not really possible (infinite bus theory), you need to complain to the local authorities, not to the manufacturer of an otherwise proper product. The designer of a PSU is not really to blame if you decide to power it with a voltage signal generated by a pathetic, home-made contraption. Your grid is the problem, start by fixing that.
    Reply

Log in

Don't have an account? Sign up now