John Keiser wrote:
Same result when booting to a CD: power light comes on but process proceeds
no further. Not a HD issue.

To clarify, the PC is set to awake from fully off using the RTC alarm, not
suspend or hibernate. This worked until a few weeks ago.


Have you added any hardware recently ? Like any USB devices ?
A USB card reader ?

A computer which crashes, before the BIOS can run any significant
number of instructions, won't beep (the computer speaker) at all.
The implication would be, that the default startup conditions
are no longer sufficient to allow the processor to start the
bootstrap process.

Some computers, they do a "double start" at startup. The BSEL
(bus select pins on the processor) request a particular FSB
setting (for processors that still have an FSB). The BIOS starts
to run, using nominal processor voltage and frequency. If the
motherboard is an enthusiast type, the BIOS can then set up
things according to any BIOS modifications and start the system
again. After this, you'd hear the single beep, which tells you
the machine has finished POST, and is about to start the OS boot
sequence.

If you're missing the single beep, and you're not getting an
error beep pattern, then the system is getting stuck early
on. Maybe not even one BIOS instruction has run in your case.

There is a second kind of failure for your situation, namely
the video display shows, but there is a flashing cursor in the
upper left hand corner. But if you had video, you probably would
have reported that here. ("Stuck with flashing cursor").

There are some simple things that can stop a board from
starting. The Power_Good signal coming from the power supply,
is a gating term in the reset chain. If the power supply
signaled that it wasn't at full voltage, by keeping Power_Good
de-asserted, that prevents the BIOS from booting. It's the
equivalent of pressing and holding the front panel reset button.
But since your system starts, after you've reset it, the problem
is not likely to be a power supply withholding Power_Good.

The motherboard adds terms to the Power_Good chain. Additional
status bits, can come from the onboard regulators. If an onboard
regulator withholds its own version of Power_Good, that stops
the startup process as well.

About all you can do at this point, is attempt a "clear CMOS" using
the Clear_RTC jumper. You do that with the power OFF and the computer
unplugged. The reason for that warning, is some older motherboards
will be damaged, if you use Clear_RTC while *any* power is present.
Usually, the user manual for the motherboard, warns about doing
a Clear CMOS with the system still powered. The purpose of
doing the Clear CMOS, followed by restoring all the BIOS settings
manually, is on the theory that some "phantom bit" in the 256 byte
CMOS RAM area, is causing the mis-programming of the motherboard
early in the BIOS startup/bootstrap sequence.

Other than that, its a good question as to what other persistent
storage on the motherboard, is remembering something it shouldn't.

Motherboards have "backfeed cut" circuits, intended to prevent
power from flowing in the wrong places, in the various shut off
states the motherboard can have. Such circuits can be fooled,
and not know the computer has been turned on, and then they prevent
the proper operation of the computer. (In one case, this was caused
by leakage current coming down the monitor cable - the computer would
start, if the user unplugged the monitor cable.) Backfeed circuit design
is kinda adhoc, so you're relying on the skills of the individual
motherboard designer, as to how well it works. Some aspects
of computer design are rather automated, or have design discipline
that ensures a high probability of success. But any time a designer
fiddles with analog power issues, there is room for error to crop
in, and the result is flaky behavior at startup. In some
cases, the behavior can even be temperature sensitive ("my
computer won't start, when the room is cold"). And for
all of those kinds of problems, I've got no answer. The
issue is complicated enough, that even with a schematic,
I'd be hard pressed to trace all those paths with a multimeter,
and determine what was going on.

Paul