Does a Bigger PSU Mean a Quieter PSU (AXi edition)

By Jonny Gerow, on May 28, 2014

I admit that Corsair over-simplifies our power supply product's fan graphs by showing how fast the PSU fan should spin at a particular load. Of course, controlling the fan in a power supply isn't as simple as "the greater the load, the faster the fan spins", but we do this because it makes for an easily digestible graph that works in the relatively limited confines of a retail box or e-commerce website. While the results are fairly accurate as long as ambient temperatures do not exceed 25°C, the actual formula used to determine how fast the PSU fan should spin is quite a bit more complicated and takes into consideration the temperature inside the unit as well as how long the unit has been running at that temperature. Since loads in computers are very dynamic, this making temperatures within the PSU equally dynamic.  So results can certainly vary from PC to PC.

No doubt, any armchair engineer can come along and say "45W of heat is 45W of heat and you're going to need the same amount of airflow to evacuate that heat, therefore the fan will always spin at the same RPM at a given load, regardless of the size of the power supply." Sure, the basic math makes sense, but it's just not that simple and it just isn't true.  At least when it comes to Corsair power supplies.

Logically, you can look at an AX760i with a 600W load as 93% efficient and extrapolate that there's 45W of heat to dissipate. If an AX1200i is also 93% efficient at 600W, then there's still 45W of heat to exhaust. But we're going to present you with some graphs that tell quite a different story.

Using a PC with a Core i7-4770 @ 4GHz, 8GB of Corsair Vengeance Pro DDR3-2133 memory, two Gigabyte Radeon R9 280X graphics cards in Crossfire, on a Gigabyte GA-Z87X-UD4H motherboard, we ran Prime95 (two threads) and Unigine Valley benchmark simultaneously to heat the system up. Once with a Corsair AX760i PSU, and then again with a Corsair AX1200i PSU. This testing produced a DC load of just over 600W, pulling 645W from the wall. Our ambient temperatures were between 26.5 and 28.4°C (tests were performed first, early in the morning and then again after lunch time).

We measured the temperature of the PSU, as well as the temperature of the CPU, so our readers could see the relationship between the two. We also measured the fan RPM. The data was collected in 2 second intervals over a period of 30 minutes. We'll start with the AX760i data:

As you can see, as the CPU temperature rises, the PSU's temperature rises. For a good 15 minutes, the PSU fan's RPM averages 840. After 15 minutes, the power supply decides it's time to start really evacuating heat and spins up to an average RPM of 1122.

Now let's take a look at what happens when we use the AX1200i to run the exact same tests, on the exact same system:

We can see that our CPU temperature rises at the same rate it did before, naturally. But the PSU fan never exceeded 720RPM during the course of testing. Of course, the AX1200i itself ran a tad bit hotter than the AX760i because of this (49.4°C versus 42.4°C), but for a PSU made to run at 100% continuously in an ambient temperature (temperature outside of the PSU) as high as 50°C, this is a minimal difference.

So how does this work? Corsair's fan controllers are typically custom programmed for each PSU. If the same programming was used for the fan's profile from one unit to the next, our results would be a lot closer to one another. So how do we determine how we want the fan to operate to maintain the lowest noise, yet provide adequate cooling?

First, the AX760i and AX1200i have fans with different diameters. While the fan blade shape, number of blades, etc. remains the same, the AX1200i has a slightly larger 140mm fan as opposed to the fan inside the AX760i, which is a 120mm fan.  Here is a graph showing the CFM of both fans, running at the saame RPM.   CFM is the cubic feet per minute of air moved and RPM is the revolutions per minute of the fan:

As you can see, the 140mm moves slightly more air than the 120mm. But the difference is really negligible, and it certainly doesn't explain why the fan in the AX1200i starts up much later. So let's take a look at some of the other differences between these two PSU's...

For one, there's the placement of the thermistor. On the AX760i, the thermistor is on the main transformer:

The thermistor for fan control in the AX1200i is located on the PCB where the +12V FET's are installed:

Another thing to take into consideration is the size differences between the two units. The AX1200i is a rather large 200mm deep. This is 20mm longer than the AX760i. This means there's a lot more area for hot air to go.

Finally, there's the components that are used so one PSU is capable of a higher maximum output than another.  The keyword here is "maximum".  The FETs, the caps, the coils, the transformers... almost everything is bulked up.  And while the components tend to be tolerant to heat up to a certain point regardless of their capacity, their tolerance for heat at lower loads is greater. The delta between where a component's de-rating curve starts and that component's maximum thermal tolerance is often quite large. So while the component's de-rating curve temperature may be exceeded, the lower loads means there is less stress on the components so they have a greater tolerance for those higher temperatures. This allows the higher wattage units to run at higher temperatures at lower loads.

If your goal is a quieter PC, and the budget allows, it may make sense to get a larger than necessary power supply if the data shows that the same fan controller is not used across the entire product line. Of course, the AXi series, with its DSP topology, is quite unique from other units on the market. And the AX760i and AX1200i are quite different from each other. So in a future installment of this experiment, we'll have a look at how two different RM Series power supplies vary under the same load.


Comments