Testing Methodology

So now that I’ve shown you the Stingray… now it’s time to show you what I do with it…

I set the time division to 2ms and the Voltage divider to .05V.

JG Testing Methodology

That means each square going across the graph horizontally is 2ms. Each square going vertically is .05V.

The center line of the graph is whatever the output voltage of whatever rail I’m testing is. So if the above is a 12V rail putting out 11.9V, then each line going North from the center line is 11.95V, 12V, 12.05V, etc. Going south from the center line is 11.85V, 11.80V, etc.

In the above photo, we can see that the voltage fluctuates less than .05V.

JG Testing Methodology

As testing progresses, and the load becomes greater, the 12V rail shows more variation. Here we see the 12V rail fluctuate as much as .1V.

Unfortunately, because of the type of load I put on the load tester, a static resistive load, the oscilloscope can’t reflect the effects drive motors may have on the rail. So consider the waveform the best case scenario in most cases. Hey… it’s better than nothing. 😉

Crossloading:

Switchmode power supplies typically have rails that interact with each other. An extremely high load on one rail can cause another rail to go out of spec.

Recently, Legit Reviews reviewed an Ultra X2 550W power supply in a rig with a pair of ATI X1900 video cards and saw a drop in voltage below the acceptable 11.4V threshold, yet the PC did not lock up or crash or anything of the sort. Something I found quite interesting because this same power supply on the load tester dropped only .5V over the entire range of the load tester.

Using the specs that were listed on Legit’s website for the test rig, I then calculated what the load on each rail of the power supply would be. As I suspected, I found that the test rig had a very high 12V load (about 30A total) and a very, very low 3.3V and 5V load (with only two sticks of RAM, one hard drive, no PCI cards, etc. I figured on about 30W between the two rails.) So essentially, the power supply really wasn’t overloaded, it was just crossloaded. Which is another word for a “lop sided” or “uneven” load on a power supply.

Well, the results of these findings are certainly valid information and since they were accomplished using a real machine, it is certainly something that could happen to someone with a build similar to the one in the review. So I decided to turn the 3.3V rail load on my tester all of the way down to 3A, and the 5V load all of the way down to 4A. Once the PSU was crossloaded, if the voltages were below 11.4V, I then increase the load on the 3.3V and 5V rails until the 12V rail is back in spec.

Sure enough, this experiment resulted in 12V rail voltages of 11.38V and 11.32V. Once I increased the load to 4A on the 3.3V and 6A on the 5V, the PSU was at 11.45V and 11.40V on the two 12V rails. So I think it’s safe to say that if this machine had a second hard drive, or a PCI card or two or maybe more USB devices, it may have saved the 12V rail. On the other hand, the build Legit Reviews was a monster and, quite frankly, I would have never considered throwing an X2 550W in there.