John wrote:
> My daughter had a power surge bad enough to get through the surge
> protector and hit the computer. After the surge, the computer would
> boot but had no internet or USB. Thinking that the mobo was fried,
> she replaced it and that seemd to cure the problem. But a day later,
> ir came back -- no internet, no USB. If she turns it off for a while
> and back on, it's okay, at least for a while.
>
> She's asking my opinion of what else could be wrong and I'm thinking
> it sounds like PSU. Do others agree? If not, then what else might it
> be?
>
>
> Reply-to address is real
> -- John
One question would be about the surge itself. Were other devices in the
house affected ? Was the surge protector damaged ? Some surge protectors
have a status LED that lights up when the surge protector itself is fried.
(Of several I own, one has the LED.)
Naturally, the power supply is the first thing to feel the effects of the
surge. And with the limited feedback range on the control loop, it's also
possible the overvoltage could be transferred through to the outputs.
(The transformer within the power supply likely has a breakdown rating
of more than 1100 volts, and lightning could easily jump across there
and go right into the DC outputs. The sticker on the ATX supply that
says "hi pot tested", verifies that the transformer can take at
least 1100 volts across the insulation inside it. That is intended to
stop certain kinds of power company issues going straight into the PC,
but can't stop lighting, because nothing stops a direct lightning
hit.)
Now, say the primary rails 3.3V, 5V, 12V are out of spec. The power supply
itself has a signal Power_Good on the main cable, which is supposed to
detect when the voltages are close to the final values, at start up. The
motherboard won't start running unless there is Power_Good asserted by
the power supply. The motherboard stays in reset, until Power_Good is
asserted. If the rails were substantially low, then the computer would
sit there with a black screen, fans running.
To have your current symptoms then, you'd need a rail to drop to a low
voltage, Power_Good to remain asserted, and other circuits on the motherboard
to still "feed good enough" for the computer itself not to crash. Which
seems a bit strange.
I think there is more to it than power supply, but swapping in another
supply for a test would be a good thing to try. If someone has a
multimeter handy, they can take the side off the PC, clip the black
ground lead to a screw on the I/O plate on the back of the computer,
then take the red lead, and probe the exposed metal within each hole
of the nylon shell of the main connector. In rough round numbers, the
primary rails should be within 5% tolerance. So if you probed where the
yellow wire went into the nylon shell of the PSU connector, the meter
would read 12.00 on the 20V range when you touched the metal inside where
the yellow wire meets the motherboard. A 5% tolerance, means any value
between 11.4V and 12.6V is acceptable.
Power supply specs, oldest to newest... Pinout tables...
http://web.archive.org/web/200304240...12V_PS_1_1.pdf (page 27)
http://www.formfactors.org/developer...X12V_1_3dg.pdf (page 30)
http://www.formfactors.org/developer...public_br2.pdf (page 37)
The rails used for Internet and USB, could be different as well. USB
at the interface pads, may be using the +5VSB rail. And if that was
out of spec enough, the PC would shut off because there isn't enough
voltage to run supervision logic. The Internet would rely on several voltages,
such as 3.3V or an onboard regulator that converts 3.3V into an even
lower voltage for the core logic of the LAN chip. (Many LAN chips use
two rails, one rail being internal on the motherboard.) And I really have
to wonder why the PC is still running, if multiple rails are out of
whack. You'd think something would crash or shut down.
Another question you'd have to ask, is whether the hard drive(s) were
damaged. The hard drive has surge protection devices on +5V and +12V.
A sustained overvoltage on either rail, causes the surge protector
device on the drive to get burned. Such a situation may arise on a
computer with a 250W Bestec brand supply, and a failure there
comes from the supply output rising on its own, to a higher than
normal value.
I'm not liking the symptoms here, because I don't have a good explanation
for them. You can try another supply, especially if you don't have
a multimeter. Swapping is easy to do, even for a novice, as long as
they take pictures with a digital camera and put everything back the
way they found it. If you aren't there personally to help, you can
use pictures from this web page, to aid in discussions about
connectors (such as "why are these Molex 1x4 connectors on the
hard drives, so hard to remove" :-) ). When removing the Molex 1x4
from a drive, you should use your fingers, as attempting to apply
tools to it, the tools slip off. Wiggling it is what I attempt
to do. Modern supplies have a compression section on either
side of the Molex, which when compressed, pushes against the
drive and eases the connector out. But the older connectors
don't have that feature, and it takes a fair bit of finger
strength to get them to budge.
http://www.playtool.com/pages/psucon...onnectors.html
(Hard on the fingers...)
http://www.playtool.com/pages/psucon...ipheralpic.jpg
Good luck, and post back if the symptoms aren't resolved. I
feed there is more at work here than just supply. Maybe a
USB device is damaged and overloading something ? With all
excess USB devices disconnected, does the LAN still run ?
Paul
Similar Threads
Linear Mode
