And now we come to the fun part - the load testing. As always, the SunMoon SM-268 will be the Ed McMahon to my Johnny Carson, and will be doing most of the work for me. It's a fancy electronic load thinger that has the ability to simulate working loads on these power supply thingers in precise steps. It will be assisted today by a Brand 4-1850 power meter, a USB instruments oscilloscope, a multimeter, and a dual probe thermometer. Since this is a triple 12V unit, and my auxiliary 12V electronic loads are refusing to play ball today, I'll sum them into one big 12V rail for testing.

Those of you who have been to the site before know how my tests work already. For those of you who haven't, here's a quick rundown. First, some room temperature tests. Going off the load table from the unit's label, I will choose a set of five balanced progressive and two lopsided unbalanced loads, and then run the unit through its paces to see how good a unit it is in general.

Following that, I'll run a load test at a really low load to see what happens when the unit is operating at a load the 80 Plus program doesn't test for.

Then, I'll do some overshoot transient tests. These involve loading the unit to 100% and looking for turn on spikes with the scope that go out of the ATX spec. All units have these spikes, but some control them better than others.

Finally, the real fun starts - the hot box. I'll lock the poor Enermax in a hot environment and see if it can really handle full power at high temps. When that's done we'll look at the DC output scope shots, take the unit apart, and then I'll try to figure out how to score this unit. Let's get going.

First things first - the unit did meet its Gold certification. But only just. See that number on test five? That's just barely sticking to Gold. I have to wonder how this unit will handle heat, if we're just barely holding on here in the cold tests. I might be in for a surprise though - you see, units that are this efficient don't heat up a whole lot. While most units I test drop their certification at the hot box door, this unit is so efficient it actually has a chance to keep it.

Before we do the hot boxing though, we need to discuss this unit a little more as we look at the voltage readings. I must admit, I was expecting slightly better voltage regulation, particularly on the 3.3V rail. 3.41 volts is a rather high starting off point. While this is not outside ATX spec, it is slightly over the 3% mark. All this while the rail itself barely holds on to 3% regulation. I'd like to see that tightened up a bit, Enermax, especially since this unit uses the VRM approach to the minor rails.

Over on the other two rails, the 5V is also not quite as stable as it could be, though it is not in any danger of going out of spec either. We have just over 3% regulation on that rail, making it a slightly worse performer than the 3.3V rail is. But those two rails are the minor ones - what about the 12V side of it? Well, that's the highlight of the voltage regulation on this unit. While I did not get the magical 1% number I've seen on some of the better units I've tested, it's still within 2%, which is very nice to see. Wait, see? See? Gasp... I CAN SEE AGAIN! Oh what a glorious day! Now, all I have to do is not look at that fan or the box again, and I'll be ok.

Moving on up to the east side, let's take a look at the low load efficiency test. This is what I like to see here. This unit managed almost 85% at a load level that drops most designs below the 80% mark. Why do most designs act like that, you ask? Well, it's because when you optimize a design for efficiency at the loads 80 Plus tests at, it starts to affect efficiency at lower load levels. Most SMPS designs don't do so well with efficiency at low loads by nature. This unit almost pulled off a Silver result in this test. Excellent.

Overshoot Transient Testing - Enermax EMG700AWT
VSB On	VSB to Full, 12V	Off to Full, 12V

But, if the low load test was excellent, this test is actually below average. Most units I test keep these waveforms at slightly lower levels than this, on all three scope shots. Now, let me make this clear - this is not out of spec. The spec is for these spikes to stay below 10% over mean value, and none of these are even close to going that high. But all the same, I was hoping for a bit better spike suppression. Not a big deal, but just a little bit of a downer in the face of such an efficient design.