Viewing page 3 of 9 pages. Previous 1 2 3 4 5 6 7 8 9 Next
THE TORTURE TEST
We've been running a complete set of tests for PSUs for many years with minor
adjustments here and there. You can read all the details the tests in SPCR
Power Supply Test Rig, v.4.1. It can be summarized thus:
- A PSU load tester customized to provide up to ~1,000W loading, with three
separate load circuits for 12V lines.
- Both current and voltage are monitored on each output line and manually
summed to obtain the power load.
- We employ the same loading formula that Intel and 80 Plus
uses for PSU testing, and multiple load test points are checked from a low
of 20W to full rated power, with 15~20 minutes of operation at each test point.
- Voltage regulation, power factor, noise & ripple, efficiency, low AC
input performance are tested at every test point. Efficiency at 220~240VAC
is also checked for a few test points.
- Noise and temperature are closely monitored, the latter with the enture
test rig in a 10~11 dBA ambient anechoic chamber.
- The heat from the load is in the box where the PSU is mounted, with a single
120mm fan used for cooling exhaust.
For this roundup, the test has been modified:
- Low power tests (under 250W) conducted quickly to establish efficiency,
power factor, electronic noise, voltage regulation and other parameters. Only
120VAC input used.
- 250W load test run for 15 hours continuously while monitoring all
parameters periodically, with no exhaust fan in the hot box. The basic rationale
for this test is that any fanless PSU rated for 400W or higher maximum load
should be able to handle 250W indefinitely. 250W is a fairly typical maximum
peak demand for the kind of system most DIY builders would use with a 400W
- Full load test run for 15 hours continuously while monitoring all
parameters periodically, with no exhaust fan in the hot box.
- If Over Temperature Protection (OTP) shutdown occurs, wait till the power
supply can be powered up again, then run the test with the Nexus 120
exhaust fan, starting at 800 RPM. If OTP shutdown recurs, retry with exhaust
fan at 1060 RPM (maximum speed of Nexus 120). If OTP shutdown still occurs,
then try a Scythe Gentle Typhoon 120mm fan at full speed (rated for
1450 RPM and more than double the airflow of Nexus 120). If OTP shutdown,
- Don't run the test in the anechoic chamber, which is not ventilated, but
in the open adjacent room to ensure some air circulation.
- Place the PSU for best airflow/cooling, ie, either right side up or upside
down. The Seasonic has an advantage with its open vented side facing up here,
so let's give all the others a bit of an edge by optimizing cooling for them
in our test setup, even if the PSU does not have the screw mount holes to
allow upside down mounting. Remember, this is not a realistic test anyway,
it's plain torture.
The OTP function in the tested PSUs could be critical to successful completion
of the torture test, one measure of which is to survive it, at least. If the
components in the power supply get too high, failure can occur. Hopefully, without
drama, explosions, sparks or flames. OTP is supposed to halt operation before
excessively high temperature is reached. In modern PSUs, normal operation is
usually restored after OTP shutdown when the internal temperature drops to a
safe level, and the power switch is reset. One simple reason for the seemingly
arbitrary 15 hour testing time: Safety. It's about as long as we could stand
to be hovering near the test rig to ensure that some catastrophic failure would
not occur unwitnessed.
The SPCR PSU tester is a Frankenstein mix of a commercial load tester
and a home-brew hotbox. The box simulates a typical ATX tower case, with
the PSU positioned just above the rear panel exhaust vent. The bottom,
back and one side of the PSU is exposed to the heat generated in the box,
which comes from banks of loading resistors in the box and in the commercial
load tester. The PSU has the thermal benefit of being open to the outside
air on the top and one side, unlike in a case. With no exhaust fan at
the exhaust vent, the air temperature directly below the PSU can exceed
60°C at 400W load. This is hotter than any computer enthusiasts would
tolerate in their own PCs.
Four 80mm fans running slowly at 5V keep the DBS-2100 commercial load
tester cooled; they also blow the heat into the hotbox. Resistor banks
in the hotbox account for more than 50% of the heat when the test load
is at or above ~400W.
Vancouver's climate is relatively mild, and over the years, most PSU tests
have been conducted in 20~22°C ambient temperature. July and August were
hot in 2010, with just trace amounts of rain and peak temperature on most days
reaching 25~32°C. That meant the ambient temperature in the test room was
typically 23-24°C, and often several degrees warmer. It also made OTP shutdown
much more likely than at lower ambient temperatures. No tests were run when
the ambient temperature in the room (at several points around the load tester)
was higher than 24°C... which is one of the reasons this project took a
couple of weeks. The ambient noise in the room measured 14~15 dBA quiet
enough to hear any excessive electronic squealing, buzzing or whine (from the
PSU under test) worth measuring or recording in the anechoic chamber.
|Help support this site, buy from one of our affiliate retailers!|