Specifications

MAXIMUMPC JANUARY 200554
Ultra Products X-Connect
A sexy-looking power supply with an ugly disposition
The Ultra X-Connect is like a bad fi rst date. Its sleek, intricate fan grille and
shiny silver fi nish are immediately impressive, but after about 20 minutes, it
clearly comes up short on inner beauty.
Ultra claims the X-Connect can continuously deliver 500 watts of power,
but our load test proved otherwise. Everything seemed fi ne at fi rst, with the X-
Connect typically drawing around 350 watts from the outlet. But a few minutes
into the test, we witnessed extreme artifacting in 3DMark03—a sign of an
unclean power stream. At times, a few moving white lines were all that could be
seen on an otherwise-black screen. Soon thereafter, the power supply started to
emit a strange and rather worrisome burning smell.
Concerned, we immediately closed 3DMark03 and shut down the computer,
noticing that the Windows desktop was also badly distorted. After letting the X-
Connect cool off for a few minutes, we rebooted and the graphical distortion had
vanished. Fire extinguishers in hand, we tried running the load test a second time
and quickly reproduced our earlier results.
In the voltage sag test, the X-Connect was a mixed bag. We measured an
initial voltage of 12.57V on the 12V power rail—the biggest deviation of any
PSU in this roundup, and dangerously close to the 12.6V maximum tolerance
of the ATX spec. It’s conceivable this could damage cheap or poorly designed
hardware over time. On the other hand, when we dropped the input voltage to
60V, the X-Connect maintained an output voltage of 11.54V. That’s high enough to
avert a system crash.
The X-Connect ships with sturdy shielded power cables that glow under UV
light, and extra cables can
be disconnected to reduce
case clutter. But active power
factor correction—the ability
to smooth out distortions in
the current being drawn from
your wall outlet— is absent,
and the Ultra’s power factor
rating of 63 percent was the
lowest in the roundup.
Sturdy, modular shielded cables and exceptional
fit and finish.
CLASSIC ROCK
ELECTRIC SHOCK
Poor voltage accuracy, no active power factor
correction, and it failed our load test.
$100, www.ultraproducts.com
4
MAXIMUMPC
VERDICT
STACK ‘EM UP
BLOW
‘EM
OUT
BEFORE WE BEGIN…
How We Test Power Supplies:
Torture Is the Name of
the Game
Our challenge was to craft feasible tests
that would provide useful and accurate
information about how different power
supplies hold up under various usage
conditions. Ultimately, we settled on a
series of rigorous tests that subjected the
contestants to a degree of stress beyond
what they would realistically be expected
to endure. Our rationale: Any PSU that can
cope with such an intense strain should
have no problem handling the day-to-day
demands of a typical power user.
First up was a load test. We built a
computer packed with tons of power-
hungry components—a 3.4GHz Prescott
Pentium 4 CPU on an Abit IC7-MAX3
mobo, 3GB of Crucial PC3200 DDR
memory, a 256MB GeForce 6800 Ultra
AGP videocard, two IBM 75GXP 7,200rpm
hard drives, two Seagate Cheetah
15,000rpm hard drives plugged into
an Adaptec PCI SCSI card, CD-RW and
DVD+RW drives, a Sound Blaster Live!
PCI soundcard, a Netgear 10/100Mbps
Ethernet card, and two 120mm case fans.
We connected each power supply to
this monstrosity and then attempted to
maximize power usage by simultaneously
performing all of the following tasks for
about an hour:
➤ Running the Iometer disk benchmark
(www.iometer.org) on both SCSI hard drives
➤ Scanning the primary 75GXP hard
drive with AVG Anti-Virus
(www.grisoft.com)
➤ Repeatedly copying a data DVD to the
secondary 75GXP drive
➤ Running CPU Burn-in (users.bigpond.net.
au/cpuburn) with error-checking disabled to
tax the CPU
➤ Looping 3DMark03 (www.futuremark.
com) Games 2 and 4 at 1600x1200 with 6x
antialiasing and all visual-quality settings
maxed to strain the videocard
Because power supplies operate
less effi ciently at high temperatures, we
conducted our tests in a poorly ventilated
room heated to around 100 degrees
Fahrenheit to simulate a hot summer day.
We used a Seasonic PowerAngel
wattmeter to measure each PSU’s power
factor—an indicator of effi ciency—and
energy consumption, and plugged the
whole setup into a surge protector. Note
that we measured the number of watts
drawn from the electrical outlet, whereas
power supplies are rated by the wattage
they can provide to the computer. This is
generally 60-70 percent of what is drawn
from the outlet, as some energy is lost to
heat, electromagnetic radiation, and other
environmental effects.
For our second test, we connected
each PSU to a loading device drawing
about 150 watts, roughly equivalent to a
typical PC under moderate load. We used
a voltmeter to check each power supply’s
initial voltage on the 12V line—the closer
our reading was to the 12V spec, the
better.
Next, we simulated a severe power
fl uctuation. We used a voltage regulator
to decrease the voltage of the electricity
being sent from the outlet to the PSU
from a starting point of 110V to a low of
about 60V. We then measured the new
output voltage on the 12V line. Ideally, this
would register little or no change from
the initial value. To be fair, you’re unlikely
to experience a 50-volt sag in real life.
A brownout usually reduces voltage to
between 90 and 100 volts, which all the
PSUs in this roundup were able to handle.
As such, the power supplies that failed at
60V aren’t necessarily bad. Rather, those
that survived are particularly good.
Now, on with the results.