Re: PC clock wrong after a few minutes- always on

On Wed, 24 Oct 2007 19:01:33 -0400, Paul <nospam@xxxxxxxxxx> wrote:

Michael Hawes wrote:
"RobV" <robv@xxxxxxxxxxxxxxx> wrote in message
news:471e92e9$0$16485$4c368faf@xxxxxxxxxxxxxxxxx
kony wrote:
On Tue, 23 Oct 2007 19:26:49 -0400, "RobV"
<robv@xxxxxxxxxxxxxxx> wrote:

kony wrote:
On Tue, 23 Oct 2007 18:14:44 -0400, "RobV"
<robv@xxxxxxxxxxxxxxx> wrote:

Bull*** Baffles Brains wrote:
I realize that the motherboard battery strength affects the
Windows XP clock time but even while constantly on and in Windows
mode, after a few minutes the clock is off by several seconds if
not more. If the PC is always on, shouldnt' the motherboard be
getting it's power from the AC to run the clock?
No, the computer does not get timing from AC 50/60Hz, since the PSU
supplies only DC voltages.
He might've mean, AC->PSU->5VSB power.
5VSB is still DC.
Yes, but can't you see the logic in skipping a step in
mentioning the power origin? It wouldn't make any sense to
write "DC" would it, since the active-on PSU outputs DC as
well as a DC battery.
The OP obviously thought that the line frequency (50/60Hz AC) was used to
keep the time correct. When you wrote "He might've mean, AC->PSU->5VSB
power.", that could be interpreted as 5VSB has an AC component to it. At
least, that's the way I was concerned the OP might take it. I was just
pointing out that it is DC as well, so the OP would not get confused. ;-)
Does clock gain or lose? Windows clock runs from CPU clock, scaled down,
I think it is called a 'tick' and is 1/60th sec. If clocki loses, may be
badly behaved software.

Mike.


Windows keeps time, by incrementing a memory location, based on the clock
tick interrupt. If the clock tick interrupt is not serviced in time, a
tick can be lost, and the Windows clock would run slow.

An example of a chipset with a bug in this respect, is Nforce2. The conditions
are not exactly reproducible, as some people experience problems and
some don't (that's me). But apparently, for whatever reason, switching
to PIC mode, rather than using APIC, in the BIOS, was enough to fix it for some
people. The problem has something to do with interrupts, but I've never seen
a proper technical description or admission there is a problem.

The clock tick interrupt, would be based on the 14.3MHz clock coming from the
clock gen chip, and its external 14.3MHz quartz crystal. By comparison,
the CMOS clock (inside the Southbridge, and backed up by the CMOS battery),
uses a 32768 Hz quartz crystal, very similar to what is in your digital watch.
Both of those crystals should be able to maintain a constant frequency,
much better than the OPs symptoms would suggest. So my guess would be,
the problem is related to the clock tick interrupt not getting
serviced properly.

The time-of-day, is transferred from the RTC to a Windows memory location,
when Windows starts. So the CMOS clock, maintains the time when the OS is
not there to do it, and the OS keeps time otherwise. I'm not sure of the
details, as to what happens when Windows shuts down - whether the
CMOS clock is entrusted with the long term time, or whether the last
value of the Windows clock is written into the CMOS clock registers.
Since all my time keeping works OK here, I've never had a chance to
observe a busted implementation.

You can observe what happens to the clock, when you come out of standby.
Initially, the time in the lower corner of the screen, will equal the
value of the clock when Windows went to sleep. But in a few seconds,
Windows gets a copy of the RTC value of the time, and then the time
in the right hand corner of the screen, corrects itself.
Yes.
The basic sequence is that the RTC, keeps time all the time, and is
_not_ changed to match the 'Wndows' value, unless you use something
(like the 'time' setting program), which does this.
While Windows is running, the RTC value is ignored. On_server_
systems, the main PC time, is updated to match the RTC time at
intervals (usually about 1 hour).
A number of things can interfere with the system time. As outlined
above, some things can cause a tick to be missed. However much worse,
some applications, reprogram the chipset, and change the tick
frequency. Things that particularly affect the time system are:
XP multi-media timers. If enabled, and especially if set towards their
highest rate, will cause timing errors.
Some old 'DOS' applications, if run in the command window.
Some games.
Dell have a timekeeping utility, that rn on a desktop machine,
performs a resync to the RTC at intervals.
Asimilar general purpose utility, is 'clockmon', which is free.
Better though if a network connection is available, is to sync to a
network time server.

Best Wishes
.