Actual chips have been known to burn up. The ICH5/ICH5R Southbridge
can go into latchup, and burns a spot on the top of the chip. That
chip can also fail more silently, and just lose the use of the USB
ports (which means the burning stopped once it disconnected USB
power to the USB logic block or pad area).
The failure mode shown here, will cause the machine to fail to POST.
This is the more violent form of latchup, where the USB pads
stay connected to the power long enough to fry the chip.
http://www.abxzone.com/forums/intel...-ich5r-chip-p4p800-ed.html?highlight=usb+port
While latchup can be caused by ESD (static discharge) into the
USB ports, at least some of the failure scenarios don't involve
static electricity at the time of failure. It is possible this
is a design fault with the chip, or a problem with the motherboard
design. But no company has claimed responsibility, and the warranty
returns continue.
Latchup is the formation of a parasitic PNPN junction in the silicon.
The PNPN is the same device type as the SCR or silicon controlled
rectifier. Like an SCR, if the gate voltage is high enough, the SCR
turns on. In this case, shorting the power rails through the silicon
die and heating it. The SCR will not stop, until power is removed,
and the voltage drops low enough for the SCR to switch off. Since
the user does not know this is happening, there is no opportunity
for the power to be switched off, and the device heats until something
else happens, like a burnout of the power connections.
Paul