One of the biggest faults with my testing methodology in the past has been the temperature of the testing environment. Typically, my power supplies were placed on a bench in a room that’s 25°C. The problem with this is that it’s never 25°C inside an actual computer.
Previously, I would run the power supplies for an hour on the bench before getting my readings, but the power supply’s temperatures would tend to plateau within only a couple minutes because the only heat the PSU would have to exhaust was the heat it produced. Unfortunately, power supplies are typically subject to very warm ambient temperatures and are usually the primary exhaust fan within a PC, because a power supply is designed to suck hot air in from the inside of the case and exhaust it out the rear.
This has motivated me to improve my testing methodology so it better reflects actual operating temperatures. What I’m now doing is installing whatever PSU I’m testing inside an actual case.
Inside this mild-mannered looking Ultra Wizard case is a PCB with a number of female PSU connections on it. I think the PCB is spare parts from either a SunMoon or TechRed ATE. Can’t remember which. The connectors are soldered to severed power leads that then plug into my SunMoon SM-268 ATE (SMPS load tester.)
Above is a picture of the Wizard case with a PSU installed (looks like a BFG 650W from here) and the necessary power leads plugged in.
When I put the cover on, you can see the duct that pumps the hot air into the case. I simply reversed the TAC vent so it was on the outside of the panel and clamped an airbox intake hose (PepBoys: $24.99/3′) onto it.
The 3″ diameter hose is “reduced” to a smaller sized hose from a Sunbeamtech “overclockers cooling kit” which is then attached to an 80MM to 120MM fan adapter that’s bolted to the back of the SunMoon SM-268 load tester. This one 120MM outlet is where the SunMoon exhausts all of the heat generated from loading up power supplies.
I then program my loads and fire up the PSU. The greater the load, the more heat that’s generated by the SM-268. That heat is then pumped back into the case, which is then sucked up by the PSU. The PSU’s de-rating curve kicks into effect at an exponential rate.
Since these photos were taken, I had added a “helper fan” to the case-end of the duct because the fan in the SunMoon wasn’t strong enough to push all of the heat down the four feet of duct work, but in the above photo you can see the lead for the thermistor that I use to measure the temperature of the air coming into the case.
The Compu-Nurse seen on the previous page has long been replaced. The thermistor in the photos is attached to a Type-K electronic thermometer.
A second thermistor is placed at the back of the PSU.
Above: The PSU is on, but idle and it’s first thing in the morning so it’s only 20°C in the room. On the left is the temperature of the air coming into the case and on the right is the temperature of the air coming out of the power supply. Obviously, there’s hardly any heat being generated by the load tester and most of that heat there is not going through the power supply.
After the load tester has been running for about 15 minutes at about a 150W load, the temps rise to 41°C. The temperature of the air coming out of the PSU are now 38°C.