And now my favorite part of the review - the load testing. Up until now, I've been doing this in three phases. Today, I'm adding a fourth just to see what happens.
As usual, the SunMoon SM-268 automated electronic load is the Teller to my Penn, featuring some assistance from other gadgetry I found around the house including a USB Instruments DS1M12 oscilloscope, three supplementary 12V loads, a dual probe thermometer, and a Brand 4-1850 power meter with a Kill-A-Watt thrown in for good measure.
The first thing I'll do is run a series of balanced progressive load tests ranging from just below 20% to 100% of full power. These will be followed by two crossload tests designed to throw the unit a curve ball and see how good the voltage stability really is.
After that, I have a new test I'm trying out: an unusually low load. Let me be honest... the cold testing has gotten a bit dull for me when it comes to efficiency. Most 80 Plus rated units on the market are almost boring in their predictability. They usually stay close to the 80 Plus test results, which anybody can go to their site to look at. This test will be somewhere around the 40-50 watt load level, where 80 Plus doesn't test and where your computer is more likely to be running at when you're just sitting there at the desktop trying to figure out which message boards you haven't posted at yet today. Most power supplies start to decline in efficiency the closer they get to zero load, and this test will see just how the efficiency is way down there in the pit. Now, this unit isn't 80 Plus certified yet, but I'll start this test with this here unit anyway and you can all rant at me in the forums if you don't want me to do this anymore.
Once that's done, I'll do my overshoot transient tests. That's where I turn the unit on and have the scope capture the turn on spikes inherent to all computer power supplies. I'll let you know if I find anything out of spec, which is defined as being a spike higher than 10% above mean value, with no dips into the negative range.
Finally, I'll repeat my progressive and crossload testing in the hot box, and we'll see how the unit does with the heat turned up.
All loads are chosen with respect to the maximum limits given on the label. Ready? Let's get going.
Results from Corsair HX850 COLD load tests
Simulated system load tests
Bang, zoom, straight to the moon! The efficiency is the highlight of the above table, where it starts at an already impressive 88.4% and immediately heads over 90%. It already looks like Corsair's out to turn some heads with this one. Sadly, from a maximum of 90.1%, the unit starts to drop off from there; eventually bottoming out at 84.9%. If we round up, that barely qualifies it for 80 Plus Silver, which is quite nice. Down in crossload test one, the result was an even 83%. This is an exceptional result when one considers that modern power supplies are not intended to power such a high 3.3V/5V crossload anymore, which used to be the norm back in the Pentium III days. To be honest, I don't think I've seen a unit of recent release that could hold the efficiency so far over 80% in this test.
Moving over to the voltage readings, the results aren't quite the best I've ever seen (Antec Signature, anyone?), but still pretty darned good regardless. On each of the main rails, regulation holds to 2% or ever so slightly better, with only the 3.3V rail really pushing past the 2% barrier. This is nothing to be ashamed of... it's already pretty tough to design a unit that holds to 2% let alone 1%. The connector quality alone can make the difference between the two.
Over in the temperature column, there's nothing really significant going on, as is usual for the cold tests. The intake vs. exhaust temperature gets up to a twenty-five degree delta by test five, but one can't really tell much from that until we put the heat on. About the only thing one can take away from that number is the assurance that the fan controller is still pretty laid back at that power level. And indeed, I couldn't hear the fan past the screaming ones in the load tester, so it seems that Corsair has some focus on the lack of fan noise here.
Let's whack her with an absurdly low load and see what happens, shall we?
Results from Corsair HX850 low load test
Well, then. You can see what I meant when I said that units start to drop off in efficiency once the loads get closer to zero, for this one is now well below 80%. Not to worry, as mentioned this is common for most designs now on the market. Let's move on now.
Overshoot Transient Testing - Corsair HX850
VSB to Full, 12V
Off to Full, 12V
Just as a quick refresher, this test is done in three stages. First, I set the SM-268 for test five's load levels, or 100%. Then, I go looking for those turn on spikes. First, I flip the main power switch and watch the 5VSB rail. Then I bring the unit out of standby to 100% while checking out the 12V rail. Last, I watch the 12V rail again, using the power switch to bring the unit from full off to 100%.
What do I think about the above scope shots? Well, I trimmed the scope cursors to show you the awesomeness. 4.4V, 7.7V, and 7.4V. The 5VSB results are only a bit better than average, but the 12V results are exceptional. I think I've only seen a select few units go lower than an 8V spike, and when you consider that the ATX spec only calls for the spike to stay below 13.2V it really hits home how awesome this kind of performance is.
Add our RSS feeds to your favorite RSS Reader or homepage.