PCI bus problem - D865PERL

Earlier, I posted about various problems with a D865PERL based system,
including:
- sometimes the computer stops dead after coming out of standby or
hibernation, display frozen, no keyboard (including ctrl-alt-del) or
mouse input works. The only thing that works is reset button.

I bought a PCI bus monitor and POST diagnostic card. Here's what
happens. (this card refers to IRDY - device ready, and DATA - data transfer)
1) I don't have any PCI cards -- except now the monitor.
2) Normally, IRDY is constant off (probably blinking so fast it doesn't
register), and DATA flashes regularly, about 1 per second.
3) When the computer hangs up, IRDY and DATA cease normal activity, and
so far I have seen the following constant states:
IRDY DATA
dim off
on on
off off
off off

From looking at the MB block diagram, It seems most any other component
could cause this.
Any suggestions as to the most likely culprit, or what to do next?

Thanks.

George Orwell wrote:
If you look in the BIOS ACPI Submenu where you set the suspend state,
there's the caution that the S3 suspend state "consumes less power, but
some drivers may not support this state." Try setting the suspend state to
S1 and see if the machine is more stable coming out of standby.

You might also try reseating the DDR modules, maybe swapping positions if
you have a matched pair. And again, there's also the possibility you have
a power supply problem. Even if the hardware monitor says all the voltages
are within spec, something may be happening in the sleep state.

Even if you were able to trace the possible fault to a specific
surface-mount capacitor, resistor, or other chip, do you have the soldering
skills to replace such a component (assuming you could even find the proper
replacement part)?

If suspend to RAM is an important feature for you, try setting to the S1
state, and see if the problem goes away.

Il mittente di questo messaggio|The sender address of this
non corrisponde ad un utente |message is not related to a real
reale ma all'indirizzo fittizio|person but to a fake address of an
di un sistema anonimizzatore |anonymous system
Per maggiori informazioni |For more info
https://www.mixmaster.it


Thanks. I have reseated all components, RAM, AGP card, and CPU (that's
all there is) plus unplugged and re-plugged all disk connectors. Yes -
I've built several pc boards with solder and even wire wrap, 20 years ago.

How would a specific surface mount component change behavior depending
on reboot, or wake up. Theoretically, there's no difference - is there?

I'll try S1, but I believe that leaves the power supply on, which I
don't want -- but that could show us something. I post what happens.

Of MB, AGP card, CPU, and RAM -- which do you think is most probably to
cause a repeatable hangup? (again, the problem is eventually after one
or more S3 wake ups, my computer hangs completely dead -- only reset
works, and PCI bus apparently hung somehow as described before) I don't
see how a MB fault would cause different behavior whether fresh boot, or
wake up -- is that wrong? . Seems like a RAM problem would cause
variable errors. CPU stop dead on it's own - not likely?? That leaves
the AGP card, and it's memory as my guess. But that's why I'm asking for
help.

"cew" wrote in message ...
Thanks. I have reseated all components, RAM, AGP card, and CPU (that's
all there is) plus unplugged and re-plugged all disk connectors. Yes -
I've built several pc boards with solder and even wire wrap, 20 years ago.

How would a specific surface mount component change behavior depending
on reboot, or wake up. Theoretically, there's no difference - is there?

I'll try S1, but I believe that leaves the power supply on, which I
don't want -- but that could show us something. I post what happens.

Of MB, AGP card, CPU, and RAM -- which do you think is most probably to
cause a repeatable hangup? (again, the problem is eventually after one
or more S3 wake ups, my computer hangs completely dead -- only reset
works, and PCI bus apparently hung somehow as described before) I don't
see how a MB fault would cause different behavior whether fresh boot, or
wake up -- is that wrong? . Seems like a RAM problem would cause
variable errors. CPU stop dead on it's own - not likely?? That leaves
the AGP card, and it's memory as my guess. But that's why I'm asking for
help.

When you use S3, the chipset and RAM is running on 5V standby.
This is not as good (tolerence), especially if the MB needs more than spec.
Check if the voltage is much less than 5V (can't remember if it's 5 or 10%).

Also, the DRAM is refreshed a lower rates. I've seen corruption after wake.

On Sun, 18 Nov 2007 14:24:09 -0800, "Eric Gisin" wrote:

"cew" wrote in message ...
Thanks. I have reseated all components, RAM, AGP card, and CPU (that's
all there is) plus unplugged and re-plugged all disk connectors. Yes -
I've built several pc boards with solder and even wire wrap, 20 years ago.

How would a specific surface mount component change behavior depending
on reboot, or wake up. Theoretically, there's no difference - is there?

I'll try S1, but I believe that leaves the power supply on, which I
don't want -- but that could show us something. I post what happens.

Of MB, AGP card, CPU, and RAM -- which do you think is most probably to
cause a repeatable hangup? (again, the problem is eventually after one
or more S3 wake ups, my computer hangs completely dead -- only reset
works, and PCI bus apparently hung somehow as described before) I don't
see how a MB fault would cause different behavior whether fresh boot, or
wake up -- is that wrong? . Seems like a RAM problem would cause
variable errors. CPU stop dead on it's own - not likely?? That leaves
the AGP card, and it's memory as my guess. But that's why I'm asking for
help.

When you use S3, the chipset and RAM is running on 5V standby.

This is not as good (tolerence), especially if the MB needs more than spec.
Check if the voltage is much less than 5V (can't remember if it's 5 or 10%).

Also, the DRAM is refreshed a lower rates. I've seen corruption after wake.

DRAM hasn't operated from 5V in many years, and neither have chipsets. They
are running from onboard regulators that *source* the standby voltage, but
there should be plenty of headroom, assuming one has a properly rated power
supply.

As for memory refresh rates, it is highly unlikely that the chipset will
arbitrarily decide to apply lower rates during standby (and that's assuming
the chipsets aren't taking advantage of self-refresh mode).

That said...every read operation to DRAMs results in a refresh to the active
row - it's the nature of DRAMs that reading a row depletes cell charge, which
in turn must be restored at the end of the read operation. This can "hide"
excessive leakage in a cell, which in turn may indeed manifest errors coming
out of standby operation.

As for supply rail specs: DRAM vendors guarantee operation at nominal voltage
+/- 10%. What shows up at the dimms is a function of the on-board regulator
design, which if properly designed can easily withstand a 5% drop at their
source inputs...

Cheers

/daytripper

Cheers

/daytripper

"daytripper" wrote in message
...
On Sun, 18 Nov 2007 14:24:09 -0800, "Eric Gisin" wrote:

When you use S3, the chipset and RAM is running on 5V standby.

This is not as good (tolerence), especially if the MB needs more than spec.
Check if the voltage is much less than 5V (can't remember if it's 5 or 10%).

Also, the DRAM is refreshed a lower rates. I've seen corruption after wake.

DRAM hasn't operated from 5V in many years, and neither have chipsets. They
are running from onboard regulators that *source* the standby voltage, but
there should be plenty of headroom, assuming one has a properly rated power
supply.

The 5VSB may be below 3.5V then, or the MB regulation is screwed.
ATX 5VSB comes in 1A and 2A version, and the MB demand may exceed the PSU.

This circuit is complete separate from switching 12/5/3.3V one.
I could not find how 5VSB connects to the Intel 8XX chipsets in their docs.
It does mention DRAM going into self-refresh for S3&S4.

Eric Gisin wrote:
"daytripper" wrote in message
...
On Sun, 18 Nov 2007 14:24:09 -0800, "Eric Gisin"
wrote:

When you use S3, the chipset and RAM is running on 5V standby.

This is not as good (tolerence), especially if the MB needs more
than spec.
Check if the voltage is much less than 5V (can't remember if it's 5
or 10%).

Also, the DRAM is refreshed a lower rates. I've seen corruption
after wake.

DRAM hasn't operated from 5V in many years, and neither have
chipsets. They
are running from onboard regulators that *source* the standby
voltage, but
there should be plenty of headroom, assuming one has a properly rated
power
supply.

The 5VSB may be below 3.5V then, or the MB regulation is screwed.
ATX 5VSB comes in 1A and 2A version, and the MB demand may exceed the
PSU.

This circuit is complete separate from switching 12/5/3.3V one.
I could not find how 5VSB connects to the Intel 8XX chipsets in their
docs.
It does mention DRAM going into self-refresh for S3&S4.


Understood daytripper and Eric. I also have a PS tester than indicate
+5SB is OK -- but the OK indication is meanly an LED on, no specifics.
WRT RAM, wouldn't RAM corruption cause variable, random errors? This
problem is completely repeatable in effect (system lockup and PCI bus
RDY and DATA signals messed up), and only variable in how long after
wake up, and number of wake up cycles, before it happens. It just
happened while typing this - about 2 minutes after wake up from
overnight S3 state.
Also, this problem has never happened between reboot and the first S3
state. And once it does occur, it does not clear on it's own within 2 hours.
My PS is an Antec Smart Power 500 watt, with 2A +5SB.
Not arguing with anyone, I appreciate the help. And maybe I'm not
thinking straight - but trying to understand what is conclusively
causing this.
Questions:
can a malfunctioning AGP card hang the PCI bus?
can RAM errors hang the PCI bus?
can anything else but a malfunctioning MB component hang the PCI bus?

cew wrote:
Earlier, I posted about various problems with a D865PERL based system,
including:
- sometimes the computer stops dead after coming out of standby or
hibernation, display frozen, no keyboard (including ctrl-alt-del) or
mouse input works. The only thing that works is reset button.

I bought a PCI bus monitor and POST diagnostic card. Here's what
happens. (this card refers to IRDY - device ready, and DATA - data
transfer)
1) I don't have any PCI cards -- except now the monitor.
2) Normally, IRDY is constant off (probably blinking so fast it
doesn't register), and DATA flashes regularly, about 1 per second.
3) When the computer hangs up, IRDY and DATA cease normal activity,
and so far I have seen the following constant states:
IRDY DATA
dim off
on on
off off
off off

From looking at the MB block diagram, It seems most any other
component could cause this.
Any suggestions as to the most likely culprit, or what to do next?

Thanks.
OK, S1 doesn't work either. Same result. Does that prove anything.
I'm looking for best guess at to what can hang up the PCI bus as described.
1) AGP card, 2) SATA, PATA, DVD, floppy drive, 3) DRAM, 4) MB problem,
5) CPU?
I don't understand how DRAM problems could cause the identical,
repeatable PCI bus hang. Can it?
And drives don't seem very likely either.
Thanks for looking --

George Orwell wrote:

On Thu, 22 Nov 2007 07:16:07 -0700, cew wrote:



OK, S1 doesn't work either. Same result. Does that prove anything.



Please try one more thing. Remove your current hard drive and replace it with a spare. Install a completely new Windows XP with just enough drivers to make your system functional. Usually you won't see the option to put your system into standby until video drivers are installed, unless the OS has them built-in. With the fresh install, see if your system can go into standby, either S1 or S3. If the system no longer hangs, then the problem was with the OS or some driver, which I suspect is the case. If the system still has problems, well who knows where the fault resides. I had a bad AGP card cause all sorts of odd behavior. Il mittente di questo messaggio|The sender address of this non corrisponde ad un utente |message is not related to a real reale ma all'indirizzo fittizio|person but to a fake address of an di un sistema anonimizzatore |anonymous system Per maggiori informazioni |For more info https://www.mixmaster.it

Already bought a cheap, new Seagate PATA drive, and will do a fresh install soon - probably this week end. Also bought a cheap, new, more capable ATI AGP card to try if the install doesn't do it. Will post results, and thanks.

cew wrote:
Earlier, I posted about various problems with a D865PERL based system,
including:
- sometimes the computer stops dead after coming out of standby or
hibernation, display frozen, no keyboard (including ctrl-alt-del) or
mouse input works. The only thing that works is reset button.

I bought a PCI bus monitor and POST diagnostic card. Here's what
happens. (this card refers to IRDY - device ready, and DATA - data
transfer)
1) I don't have any PCI cards -- except now the monitor.
2) Normally, IRDY is constant off (probably blinking so fast it
doesn't register), and DATA flashes regularly, about 1 per second.
3) When the computer hangs up, IRDY and DATA cease normal activity,
and so far I have seen the following constant states:
IRDY DATA
dim off
on on
off off
off off

From looking at the MB block diagram, It seems most any other
component could cause this.
Any suggestions as to the most likely culprit, or what to do next?

Thanks.
I booted this morning, and for the first time, a few minutes later, the
display got jumbled and the computer hung up. On reboot, BIOS custom
settings and time were lost. (have new MB battery) Sounds like bad MB now.

George Orwell wrote:

On Thu, 22 Nov 2007 07:16:07 -0700, cew wrote:



OK, S1 doesn't work either. Same result. Does that prove anything.



Please try one more thing. Remove your current hard drive and replace it with a spare. Install a completely new Windows XP with just enough drivers to make your system functional. Usually you won't see the option to put your system into standby until video drivers are installed, unless the OS has them built-in. With the fresh install, see if your system can go into standby, either S1 or S3. If the system no longer hangs, then the problem was with the OS or some driver, which I suspect is the case. If the system still has problems, well who knows where the fault resides. I had a bad AGP card cause all sorts of odd behavior. Il mittente di questo messaggio|The sender address of this non corrisponde ad un utente |message is not related to a real reale ma all'indirizzo fittizio|person but to a fake address of an di un sistema anonimizzatore |anonymous system Per maggiori informazioni |For more info https://www.mixmaster.it

New Seagate disk. Fresh install. Same problem.(reaching for hammer--)