Pass the abacus!
Input and Output Power? Eh?
There are two sets of power figures to be reported because not only does the power supply output a power to the PC components that are connected to it, it also draws power from the mains to do so. So the supplies were drawing from the 230V, 50Hz mains supply we fed them, which obviously draws a power, measured in watts. The 6842A provided the input power measurement and it was recorded verbatim with no correction factors.
Output power is then calculated as follows:
P = ∑(V1xA1 to VixAi) x PFC. That's simply the sum of voltage multiplied by current (VxA) for all the connected rails, multiplied by PFC which is power factor correction. So you total up the power draw from all connected rails and scale it by PFC. PFC is simply a measure of resitance to reactance of an AC supply, as a ratio between 0 and 1. If it's 1, you're converting all input power into output power from the AC mains. If it's not 1, the lost power is held in the reactance as the AC supply oscillates. Dan knows more about it than I can be arsed to convey, and he's funnier than me with it too. Dan will also tell you that the upshot of PFC is that regardless of whether or not it contains an active or passive PFC circuit, it'll always be a negligable factor to care about, overall.
Since Dan doesn't care and I measured PFC to be as near as damnit to 1 every time I plugged a supply in, I assume it to be 1 and therefore ignore it for the purposes of this article.
So we just summed power from all the rails, reporting the total. As an example, say I loaded up a PSU with the following currents and measured the following voltages, on the following rails:
| Rail | Rail voltage (volts) | Applied Load (amperes) |
| +5vsb | 5.002 | 1 |
| -5V | -5.039 | 0.25 |
| -12V | -12.122 | 0.4 |
| +3.3V | +3.118 | 18 |
| +12V1 | +11.922 | 16 |
| +12V2 | +11.804 | 16 |
| +12V3 | +12.039 | 15.5 |
| +5V | +4.921 | 18 |
The output power is therefore:
(5.002 * 1) + (5.039 * 0.25) + (12.122 * 0.4) + (3.118 * 18) + (11.922 * 16) + (11.804 * 16) + (12.039 * 15.5) + (4.921 * 18) = 722.033W
Which is otherwise known as a very beefy ATX2.0 supply due to the fat output power and multiple +12V rails. The difference between ATX2.0 and previous ATX power supply specifications is the provision for less current draw available on peripheral rails like +3.3V and +5V, and a lot more available current for +12V, spread across multiple rail supplies. With your CPU and graphics card(s) wanting to draw ever more power from your PSU, that's the reason for ATX2.0's creation.
Efficiency
When you see power consumption figures quoted in other reviews on the web (since we don't publish numbers like that for the time being), you'll see the figure quoted as being from the mains, since that's the easiest place for you to measure power from. Using a watt meter in between the mains and your PSU, you can see what the PSU is pulling. That's your input power as explained above. However, since power supplies aren't 100% efficient, which you can feel directly as the heat being produced by the wasted power, there's the output power to measure too, which I showed you how we calculated above.
Efficiency as a percentage is simply (input power / output power) * 100. So for that output power example above, if the supply was drawing an average of 860.8W of input power at the time, its efficiency would be (722.033 / 860.8) * 100 or 83.88%. For reference, that'd be an excellent efficiency for a power supply that capable, with calculated efficiencies somewhat lower as you'll see soon. More efficiency equals less wasted power and since you're billed normally by the kilowatt hour, less cash given to your energy supplier. Hoorah!
So more efficient supplies given the same output power will be better for your wallet, provided the price differential is less than the money saved by higher efficiency. The good news is that it nearly always will be, given the disgustingly tiny margins on many of the more popular supplies, just so they can compete with the vendors selling craptastic supplies with high supposed output power at ridiculously low prices.
And that's really all you need to care about
While I measured more data points for this article than any other I've ever written, it's great that I can really distill them down to two things that you'll be caring about: whether the supply can give you what it says on the box, and whether if or when it does so, it does it with good efficiency.
Those two things were the mantras we set out before we started any testing. You the consumer need to have a choice of supplies you can trust to do what they say they will, which ironically is often way more than you'll ever call upon them to do, and you need them to do that job while not costing you a shit load in wasted money and unwanted heat in your chassis due to poor efficiency.
I'll obviously tell you those things and a bunch more stuff relevant to the testing, like how hot they got and what cabling they have. I'll do so as follows, to keep things simple and to the point.
What you'll see on coming pages
The next lot of pages will sum up each power supply in real terms. Firstly I'll sum up the PSU in physical terms. That means its colour, whether it was gloriously shiny or boringly dull, whether it had vomit-inducing LED fans or UV glowing bits, how big it was, what cable runs it has and what ATX spec it supports. Then I'll tell you whether it passed or failed testing and what general type of PC it's suitable for.
After a billion or so pages of that, two or three to a page, I'll show you the graphs that sum up the testing data. They'll show you how well each PSU did in top-to-bottom ranking terms based on metrics like efficiency and how hot they got.
There'll also be a funny page on what happened when PD and I blew shit up. Electricity is fun!
Finally, I'll go on to sum everything up, present a list of what supplies I heartily recommend you can invest your money into (or will I!), giving you the HEXUS big picture of the units we tested. Definitive stuff we hope. Let's get cracking.
