load-testing-the-ax1200i-digital-atx-psu

By Jonny Gerow, on 8 de Agosto de 2012

Hello everyone. My name is Jonny Gerow, but a lot of people know me by my alias "jonnyGURU". I started up a little website humbly called jonnyGURU.com, where we primarily review computer power supply units. I'm very happy to say that I am now working at Corsair as a technical marketing specialist for power supplies and I'm excited to be given the opportunity to play with Corsair's newest unit: the AX1200i Digital ATX Power Supply.

Last week, I cracked one open and showed everyone what it looks like on the inside and how it works. Today I get to put the AX1200i through the paces on some of this test equipment I happen to have sitting here.

For testing power supplies, I have a SunMoon SM-8800 power supply load tester in combination with an SM-220 auxiliary load. This machine has all of the same connectors as a standard ATX PC. It has a 24-pin, 8-pin CPU power connector, three PCIe power connectors, three Molex, a SATA and a floppy power connector. I can program a different load for each type of connector. The machine than tells me the voltage on each connector and the total wattage of the load. The SunMoon allows me to program up to 10 different scenarios. For today's testing, I'm going to duplicate 80 PLUS's loads at 10%, 20%, 50% and 100%.

 


Buttons, buttons, everywhere!

 

I also measure how much wattage if coming from the wall's AC outlet. When you take the DC output wattage of the power supply and divide by how much you're using from the wall, you're calculating efficiency. Any wattage not converted from AC to DC is exhausted by the PSU as heat. Speaking of heat, creating loads creates heat within the load testers that is exhausted by a few fans. To make the power supply's operating temperature more realistic, like that of a power supply inside an actual PC, I pump the heat from the load tester into the intake of the power supply, waiting 15 minutes between tests to make sure everything gets nice and toasty warm. I then use two thermal probes to measure both the intake and exhaust temperatures.

 

 

So let's hook up the AX1200i and do this...

 

Corsair AX1200i 1200W load tests
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Intake/
Exhaust
Simulated system load tests
10% 1.8A 1.8A 0.1A 8.5A 0.3A 119.4W/
139W
85.9% .99 29°C/
32°C
3.30V 4.99V -12.00V 12.05V 5.07V
15mV 16mV 38mV 14mV 15mV
20% 3.7A 3.7A 0.1A 17.1A 0.6A 240.3W/
266W
90.3% .99 31°C/
34°C
3.29V 4.98V -12.01V 12.05V 5.06V
15mV 16mV 38mV 16mV 15mV
50% 9.2A 9.2A 0.3A 42.8A 1.5A 600.5W/
649W
92.5% .99 39°C/
44°C
3.27V 4.96V -12.04V 12.04V 5.02V
17mV 20mV 34mV 18mV 19mV
100% 18.3A 18.3A 0.7A 85.5A 3.0A 1196.2W/
1343W
89.1% .99 47°C/
53°C
3.24V 4.92V -12.09V 12.02V 4.95V
16mV 24mV 37mV 20mV 24mV

 

To get 80 Plus Platinum, a unit needs to hit 90%, 92% and 89% at 20%, 50% and 100% loads. We clearly hit those numbers out of the park. But to accomplish efficiency like that, sometimes secondary filtering has to be reduced because the addition of caps and coils adds resistance that can decrease efficiency. Doing this can increase the units ripple.

Ripple is small variations in the DC output. ATX specification allows a power supply to have up to 120mV peak to peak (.12V) on the +12V and -12V rails and 50mV peak to peak (.05V) on the +3.3V, +5V and +5VSB rails. Too much ripple on the output of a PSU means that components receiving that power may get stressed by having to compensate for those small variations in voltage and could shorten the life of the part.

While each rail was loaded, I measured the ripple using an oscilloscope hooked up to the load tester. I put the peak to peak value in the above table, while below there are thumbnails of screenshots from the oscilloscope. Clicking on the peak to peak value in the table will bring up a bigger picture of the waveform captured by the oscilloscope. The screenshots show ripple on a scale of 2ms on the X axis and 10mV (0.01V) on the Y axis. As we can see, the ripple on the +12V never gets higher than 20mV. That's 1/6th of the ATX specification! The +3.3V never gets higher than 17mV and the +5V and +5VSB never gets higher than 24mV. This is well within ATX specification.

 


Even under full load, the +12V only has 20mV of ripple.

 

One of the other things I've noticed with highly efficient power supplies is that the voltage regulation often suffers. This is where the AX1200i's DSP should help since the voltage can be adjusted much quicker and more accurately than with conventional voltage sensing.

If we look at the +3.3V going from 10% load to 100% load, we see that it only drops .06V, or 1.8%. The +5V drops .07V, or 1.4%. The +12V dropped from 12.05V to 12.02V, which is only .03V, or .25%. That's right. There's a decimal before that 25. We're talking one quarter of one percent. That's the best I've ever seen!

And operating at high temperatures? Well... 47°C can hardly be considered comfortable, but the AX1200i toughed it out with 100% load on it.

So all and all, the AX1200i looks like it could very well be the best 1200W power supply on the market. But don't just take my word for it. Let's have a look at the AX1200i that Ecova Plug Load Solutions (the 80 PLUS guys) tested and compare it against the other Platinum 1200W units they've tested.

First, the Supermicro PWS-1K25P-PQ:

 

Supermicro PWS-1K25P-PQ 80 Plus Test Results
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Simulated system load tests
20% 2.6A 2.6A 0.0A 18A 0.5A 240.3W/
242.44W
90.43% .99
3.3V 5.1V -12.1V 12.1V 5.1V
50% 6.5A 6.5A 0.1A 44.9A 1.4A 604.37W/
656.3W
92.09% 1.0
3.3V 5.1V -12.1V 12.1V 5.1V
100% 13A 13A 0.3A 89.7A 2.7A 1203.72W/
1347W
89.36% 1.0
3.2V 5V -12.1V 12.0V 5.1V

 

The High Power HP-1200-G14S(C)-PT:

 

High Power HP-1200-G14S(C)-PT 80 Plus Test Results
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Simulated system load tests
20% 2.9A 2.9A 0.0A 17.8A 0.5A 250.15W/
274.2W
91.23% .96
3.4V 5.2V -12.4V 12.5V 5.2V
50% 7.2A 7.2A 0.1A 44.5A 1.3A 624.46W/
674.6W
92.57% .99
3.4V 5.2V -12.4V 12.5V 5.1V
100% 14.4A 14.4A 0.3A 88.8A 2.7A 1240.61W/
1367.4W
90.73% 1.0
3.4V 5.1V -12.5V 12.4V 5.1V

 

The Enermax Platimax 1200W:

 

Enermax Platimax 1200W 80 Plus Test Results
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Simulated system load tests
20% 2.6A 2.6A 0.1A 17.9A 0.5A 246.7W/
273.8W
90.1% .95
3.4V 5.1V -12.0V 12.3V 5.1V
50% 6.4A 6.5A 0.2A 44.9A 1.3A 614.21W/
665.9W
92.24% .98
3.4V 5.1V -11.9V 12.2V 5.1V
100% 12.8A 12.9A 0.5A 89.6A 2.7A 1216.39W/
1358.9W
89.51% .98
3.4V 5.0V -11.9V 12.1V 5.0V

 

And finally, the Super Flower SF-1200P14PE, which is also sold by PC Power & Cooling as the Silencer Mk III 1200W:

 

Super Flower SF-1200P14PE 80 Plus Test Results
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Simulated system load tests
20% 2.2A 2.2A 0.1A 18.3A 0.5A 247.87W/
269.7W
91.91% .99
3.4V 5.2V -12.0V 12.4V 5.1V
50% 5.5A 5.5A 0.2A 45.6A 1.1A 618.17W/
667.1W
92.66% .99
3.4V 5.2V -11.9V 12.3V 5.1V
100% 11A 11A 0.5A 91.1A 2.3A 1231.19W/
1363.0W
90.33% 1.00
3.4V 5.1V -11.9V 12.3V 5.1V

 

We'll take a look at these results and compare them against the results Ecova got with the AX1200i:

 

Corsair AX1200i 80 Plus Test Results
Load % +3.3V +5V -12V +12V +5VSB DC Watts/
AC Watts
Eff. Power
Factor
Simulated system load tests
20% 3.7A 3.7A 0.1A 17.1A 0.6A 241.32W/
265.8W
90.79% .99
3.3V 5.0V -12.2V 12.0V 5.1V
50% 9.2A 9.2A 0.3A 42.8A 1.5A 602.73W/
653.0W
92.30% 1.00
3.3V 5.0V -12.2V 12.0V 5.0V
100% 18.3A 18.3A 0.7A 85.5A 3A 1200.67W/
1344.0W
89.34% 1.00
3.3V 5.0V -12.2V 12.0V 5.0V

 

Before we compare these results, I'd like to go ahead and point out there are a few inconsistencies with the Ecova's testing. First off, different power supplies have different specifications and therefore require different loads to accomplish 20%, 50% and 100%. For example, the AX1200i is capable of a combined 180W on the +3.3V and +5V, while the Platimax can only output 120W, so you couldn't load the +3.3V and +5V up on the Platimax as much as you can the Corsair. There's certainly no helping that. Secondly, sometimes the distribution of loads that Ecova uses vary from one unit to the other. For example, they use the same loads on both the +3.3V and +5V on the AX1200i, but the Platimax has a 0.1A higher load on the +5V than on the +3.3V during the 50% and 100% test. Also, sometimes the -12V rail is loaded during the 20% test, while other times it is not. Finally, they only measure voltages to 1/10th of a volt. That means voltage can drop as much as .04V and not be recorded as a drop in voltage. All that said, Ecova's goal is to measure efficiency and they do so with great accuracy. Wattage is measured to the 1/100th of a watt and efficiency is calculated to 1/100th of a percent.

 

Aggregated 80 Plus results +3.3V Drop +5V Drop +12V Drop +5VSB Drop 20% Efficiency 50% Efficiency 100% Efficiency
Corsair AX1200i 0V 0V 0V 0.1V 90.79% 92.30% 89.34%
Super Flower SF-1200P14PE 0V 0V 0.1V 0V 91.91% 92.66% 90.33%
Enermax Platimax 0V 0.1V 0.2V 0.1V 90.10% 92.24% 89.51%
High Power HP-1200-G14S(C)-PT 0V 0.1V 0.1V 0.1V 91.23% 92.57% 90.73%
Supermicro PWS-1K25P-PQ 0.1V 0.1V 0.1V 0V 90.43% 92.09% 89.36%

 

Using Ecova's results, the Corsair AX1200i is the only unit to not show a drop in voltage on the +12V, as well as there not being any drop on the +3.3V and +5V.

If we look at the efficiency numbers, we can see that the Corsair AX1200i is not always the most efficient of the five power supplies, but other than the Enermax Platimax 1200W, the Corsair is the only power supply on the list that is fully modular. Because each connection can create a little bit of resistance that can result in a drop in efficiency, it's not unusual for a fully modular power supply to have slightly lower efficiency results. The Super Flower is semi-modular, but when it is tested for efficiency, none of the modular cables are used in testing; only the fixed cables are used.

Up next, I'm going to install the AX1200i in a system and have a look at the Corsair Link software, which allows me to monitor loads, voltages, temperatures and fan speed and control the OCP (Over Current Protection) for the +12V rails.


Comentários