XP Powermanagement

[Noone]

Win XP Home and Pro (both on numerous 'puters with same issue). All Dell P4
Northwoods and Prescotts with 865 and 915 chipsets. S3 and S4
powermanagement works fine in these boxes until I use a USB keyboard and
configure it to be able to wake the computer from sleep. S4 (Hibernation is
not affected). Then the PSU fan and the PROC fan won't shut down but keeps
spinning until the timeout for S4 occurs. Then Hibernation operates as it
should. Note, the HDD spins down and the monitor goes black but the fans
spin on. This does not happen with a PSU keyboard. And only happens with a
USB keyboard when configured to have the ability to wake the computer from
sleep. Diabling this feature allows the computer to enter S3 properly
complete with fans spinning down. But then the 'puter must be awakened by
the front panel power button because the keyboard's funtion in this regard
is disabled.

So I learned that Pin #9 is +5 volts VSB (Voltage Stand By). And it's
adjacent pin, #19 (on a 20 pin connector) and #21 (on a 24 pin connector) is
a +5 volt full time voltage. Some MOBO's have a jumper to allow choice
between +5 volt VSB and +5 VOLT continous voltage. This choice, as I
understand, allows the keyboard and mice or whatever is plugged into the
rear USB (main) ports to either have continous power or Voltage Stand By.
The problem is that USB keyboards require continous voltage to have ability
to use the wake-up function in their powermanagement tab and still allow the
motherboard (i.e. read...Fans) to shut down with S3 powermanagement. Problem
is my Dell keyboard does not have this jumper. So my question for any of you
smart people out there is: Can I switch the MOBO connectors listed above and
achieve the same thing as if I had a jumper to choose which power flowed to
the USB ports. I will appreciate any logical reply.

 

[Paul]

Win XP Home and Pro (both on numerous 'puters with same issue). All Dell P4
Northwoods and Prescotts with 865 and 915 chipsets. S3 and S4
powermanagement works fine in these boxes until I use a USB keyboard and
configure it to be able to wake the computer from sleep. S4 (Hibernation is
not affected). Then the PSU fan and the PROC fan won't shut down but keeps
spinning until the timeout for S4 occurs. Then Hibernation operates as it
should. Note, the HDD spins down and the monitor goes black but the fans
spin on. This does not happen with a PSU keyboard. And only happens with a
USB keyboard when configured to have the ability to wake the computer from
sleep. Diabling this feature allows the computer to enter S3 properly
complete with fans spinning down. But then the 'puter must be awakened by
the front panel power button because the keyboard's funtion in this regard
is disabled.

So I learned that Pin #9 is +5 volts VSB (Voltage Stand By). And it's
adjacent pin, #19 (on a 20 pin connector) and #21 (on a 24 pin connector) is
a +5 volt full time voltage. Some MOBO's have a jumper to allow choice
between +5 volt VSB and +5 VOLT continous voltage. This choice, as I
understand, allows the keyboard and mice or whatever is plugged into the
rear USB (main) ports to either have continous power or Voltage Stand By.
The problem is that USB keyboards require continous voltage to have ability
to use the wake-up function in their powermanagement tab and still allow the
motherboard (i.e. read...Fans) to shut down with S3 powermanagement. Problem
is my Dell keyboard does not have this jumper. So my question for any of you
smart people out there is: Can I switch the MOBO connectors listed above and
achieve the same thing as if I had a jumper to choose which power flowed to
the USB ports. I will appreciate any logical reply.

No, you can't do that.

If you rewire the main power supply harness, you'll be swapping
a 5V @ 20 to 30 amp rail for a 5VSB @ 2 amp rail. The computer
will overload the (weak) 2 amp rail, and the computer will shut off instantly.
You won't be able to keep it running for more than about 100 milliseconds.
Your rewiring job may also violate some motherboard design rules
about supply sequencing or the like (i.e. upset the reset circuit).
That is not a trivial change.

You could implement the change at the 2x5 header or connector level,
but I still wouldn't recommend it. If done at the header, you should
include fuse protection, with a Polyfuse. Or, if a Polyfuse is
evident already, the position where the Polyfuse is installed,
may be a good point to make a circuit change. (Tip the Polyfuse up on
end with your soldering irons, and wire +5VSB to one end.) But you
really need a motherboard schematic, before committing such butchery.

There are three options for USB circuits.

1) Permanently wired to +5V. The USB port only has power when the
computer is running with fans spinning. You have to use the front
power button, to start a sleeping or hibernated computer. An older
computer might be set up that way (Pentium 3 era?).

2) Motherboard has a +5V/+5VSB 3 pin jumper block. This allows
selecting to power a pair of USB ports, from either of those
rails. Selecting +5VSB via the jumper, means the keyboard *may*
be a candidate for waking the computer. You must enter the BIOS
and look in the appropriate "Wake On..." page to ensure that USB
devices can wake the computer. In some cases, the BIOS even includes
a setting to determine the key press sequence that will wake the
computer. So there can be some fun options in there.

The very same details may apply to PS/2 ports as well. A jumper
for +5V/+5VSB, BIOS settings for "Wake On..." and so on.

Looking in the BIOS first, is the very first thing you should
be doing. Maybe a setting there will change the computer behavior ?

3) Motherboard wired to +5VSB. This means the computer is always
powering the USB ports (even in sleep). This option has been
used for about the last three to five years or so. On the minus side,
if enough heavy USB loads are connected (four or more Alcatel Frog
modems), it could overload the +5VSB rail. On the plus side, there
are no jumpers to play with.

The jumper scheme in (2) is actually preferable, as you can operate only
the keyboard/mouse USB ports from +5VSB, and leave the other ports set
to +5V. The "modern" option in (3) is a step backwards, but it
is probably saving $0.10 per motherboard they build.

If you can't get it working, or the infrastructure isn't there
for the feature, you can always use the front power switch. While
I've tested kayboard/mouse waking, I always leave the final
machine configuration controlled by the front power switch. It
means the computer won't come on by accident if it is bumped.

HTH,
Paul

 

[Noone]

Paul, thanks for your reply. Then does this mean that PS2 ports are routed
through +5 volts and USB ports are routed through +5 volts VSB? And if so
does this mean in order to prevent the fans from continous spinning that the
2 USB ports for input devices needs to be run on the +5 volts rail? I don't
have this problem with Vista or Win 7. Are newer MOBOS reconfigured in
regard to USB as compared to my generation of boards. As well would the
auxilliary USB ports on the front panel wire to +5 volts or +5 volts VSB? I
have almost no control in the BIOS for power management on these boxes. My
only choices were S1, S3, and S4. Other than that I am relegated to control
using the OS. Currently nothing is enabled to wake the computer and it
sleeps just fine. I have to use the front panel power button to resume from
S3 just as if it were S4. I can live with it. Most likely I will revert to
PS2 keyboards in order to have my cake and eat it to. I did use
"DUMPPO.EXE". If I understood it correctly this executable forces an S1 to
S3 modification if needed. The before showed I was in S1 and of course the
after showed I was in S3 but nothing changed. So I then understood it was
simply a hardware issue.

 

[Paul]

Paul, thanks for your reply. Then does this mean that PS2 ports are routed
through +5 volts and USB ports are routed through +5 volts VSB? And if so
does this mean in order to prevent the fans from continous spinning that the
2 USB ports for input devices needs to be run on the +5 volts rail? I don't
have this problem with Vista or Win 7. Are newer MOBOS reconfigured in
regard to USB as compared to my generation of boards. As well would the
auxilliary USB ports on the front panel wire to +5 volts or +5 volts VSB? I
have almost no control in the BIOS for power management on these boxes. My
only choices were S1, S3, and S4. Other than that I am relegated to control
using the OS. Currently nothing is enabled to wake the computer and it
sleeps just fine. I have to use the front panel power button to resume from
S3 just as if it were S4. I can live with it. Most likely I will revert to
PS2 keyboards in order to have my cake and eat it to. I did use
"DUMPPO.EXE". If I understood it correctly this executable forces an S1 to
S3 modification if needed. The before showed I was in S1 and of course the
after showed I was in S3 but nothing changed. So I then understood it was
simply a hardware issue.

At least some of this "volts" stuff, you can verify by looking at
a PS/2 optical mouse or a USB optical mouse, and seeing if the LED
light is on. At least my optical mouse, still uses red LED lighting.
Some other mice, may use infrared. I use my optical mouse, as
a quick way to verify whether power is available in sleep or hibernate
or whatever. (Note - PS/2 is not hot plug compatible. If the mouse
light is on, you must turn off the computer completely, before
making connector changes. USB doesn't have that issue, due to the
usage of "advanced power/ground" pins. All the pins on the PS/2 connector
are the same length, and that is not a hot plug compatible design.)

The PS/2 ports have their own power jumper, on the motherboards that have
jumpers for everything.

I've seen at least one motherboard, where to save money on
jumpers, they wired a whole bunch of ports to a single jumper block.
On my Asus motherboards, there could be as many as five three pin blocks,
for setting the power source. (One jumper block for PS/2, perhaps four
more for pairs of USB and so on.)

On the newer motherboards, you may have to use the "optical mouse test",
to determine the power scheme. It would make sense (from a philosophy
point of view), for them all to be connected to +5VSB, both USB and PS/2
ports. Even the front ports. If they're not going to use jumpers, then
making all ports "wake compatible" by using +5VSB, makes sense. You can try
various tests with the USB optical mouse, because it is hot plug compatible,
and you can try the mouse with the system powered, with the system sleeping,
or watch what happens to the LED on the mouse, as the computer goes to sleep.

I don't understand all the power states on USB. I've had USB stuff, where
the LEDs stay on when the computer sleeps. My USB to RS232 adapter dongles
stay lit all the time (they keep the room lit at night). Sleep doesn't kill
them. The optical mouse on the other hand, the LED goes off when the computer
sleeps. So that complicates things a bit, when it comes to testing. Maybe on
an older computer, the same mouse LED would stay on. It seems if the computer
has the right software to issue commands to the mouse (or other devices),
they can get it to power off. This is particularly an issue with some NIC ports
on computers now, where they've been so clever, you can no longer do Wake On LAN,
because the software turned off the LAN LEDs and PHY :-( In fact, if one
OS turns off the LAN port in that way, multi-booting a second OS may not
be able to turn it back on. Very clever.

*******

Regarding dumppo.

Say you originally did an ACPI OS install (i.e. Windows didn't complain).
You did a good job, and set the BIOS to S3. Everything was working.

Then one day, the BIOS battery goes flat, all the BIOS setting are
lost, and the BIOS defaults set ACPI to S1 only.

When the computer boots, it notices the new table entry, specifically
that S3 is gone. The OS makes a registry change.

Now, you've got no S3.

The job of "dumppo", is to do an administrative override. If the
hardware again supports S3 (as determined by correcting the BIOS entry),
then you use dumppo to force the OS to recognize the good news. Otherwise,
without it, the OS won't revert to using S3 on its own.

That's my understanding of the purpose of dumppo. To do an
"administrative override", when the OS won't do anything about
the S3 coming back.

If you had a non-ACPI install, like the installer CD detects the
motherboard is not ACPI compatible, then dumppo can't clean up
a mess like that. Dumppo is for a good ACPI install, that has
gone slightly wrong. It won't correct a major foulup.

Paul

 

[Noone]

Good stuff Paul. I really appreciate your knowledge on the subject. Before I
went looking for help I read all I could find on the subject. I came back
here after years of absence because I could find no one in my association
that could explain the circumstance. The MOBO's are Dell PIV 2003-2005. I
will hot plug the mouse to each and every USB port but what you say makes
sense when you state that they would have made all ports +5 volts VSB and be
done with it. Vista was around 2007, yes? So it seems that maybe somewhere
around then the MOBO manufacturers changed the architecture to accomidate
the new direction hardware was going. Remember, my PS2 input devices will
allow S3 so the fans shut down. The USB input devices keep the fans blowing
which seems to say that they only allow S1...until of course the timeout for
S4 occurs which works just fine. I found these 6 Dells at numerous random
yard sales for a token. Some worked, some didn't...but my investment was
next to nothing overall. And now of course I have 6 fast dependable boxes.
They all run XP and of course have COA's so all is good. I hate Vista (my
son's Alienware laptop) and I hate Win7 (my wife's laptop). I don't like Mac
but Linux is still too much work. Most likely we will migrate to Mac knowing
XP won't be usable much longer. I think the time is ripe for a new Steve
Jobs or Bill Gates. I could be wrong but I don't think Linux can fill that
bill. But to be long in tooth, I am still unclear as to why PS2 will enter
S3 fine and USB will stop at S1...if the input devices in question are set
to wake the system. Is it positively because one is on +5 volts VSB and the
other is on +5volts? Or is it simply the MOBO is from a generation of
hardware where the USB protocol was not clearly implicated in hardware? The
Dells have other issues but none so far that have put me in the weeds.
Expecially for the price. Between 2002 and 2007 Dell sold a grand number of
these boxes and now they can be had for little money for the price conscious
and tech savvy tinkerer. However most tech savvy tinkerers prefer hardware
more attuned to their manipulation efforts. And understandably so. I'm just
a cheapskate. So is it simply the voltage rails of the MOBO or is it the
architecture? TIA!

 

[Paul]

Good stuff Paul. I really appreciate your knowledge on the subject. Before I
went looking for help I read all I could find on the subject. I came back
here after years of absence because I could find no one in my association
that could explain the circumstance. The MOBO's are Dell PIV 2003-2005. I
will hot plug the mouse to each and every USB port but what you say makes
sense when you state that they would have made all ports +5 volts VSB and be
done with it. Vista was around 2007, yes? So it seems that maybe somewhere
around then the MOBO manufacturers changed the architecture to accomidate
the new direction hardware was going. Remember, my PS2 input devices will
allow S3 so the fans shut down. The USB input devices keep the fans blowing
which seems to say that they only allow S1...until of course the timeout for
S4 occurs which works just fine. I found these 6 Dells at numerous random
yard sales for a token. Some worked, some didn't...but my investment was
next to nothing overall. And now of course I have 6 fast dependable boxes.
They all run XP and of course have COA's so all is good. I hate Vista (my
son's Alienware laptop) and I hate Win7 (my wife's laptop). I don't like Mac
but Linux is still too much work. Most likely we will migrate to Mac knowing
XP won't be usable much longer. I think the time is ripe for a new Steve
Jobs or Bill Gates. I could be wrong but I don't think Linux can fill that
bill. But to be long in tooth, I am still unclear as to why PS2 will enter
S3 fine and USB will stop at S1...if the input devices in question are set
to wake the system. Is it positively because one is on +5 volts VSB and the
other is on +5volts? Or is it simply the MOBO is from a generation of
hardware where the USB protocol was not clearly implicated in hardware? The
Dells have other issues but none so far that have put me in the weeds.
Expecially for the price. Between 2002 and 2007 Dell sold a grand number of
these boxes and now they can be had for little money for the price conscious
and tech savvy tinkerer. However most tech savvy tinkerers prefer hardware
more attuned to their manipulation efforts. And understandably so. I'm just
a cheapskate. So is it simply the voltage rails of the MOBO or is it the
architecture? TIA!

I recommend documenting the exact model number of the Dell computers,
so you can search for identical symptoms. Just knowing they're
vintage Pentium 4 systems from 2003-2005 is not enough.

Generally speaking, if the chipset used by the processor is Intel,
they're pretty good when it comes to ACPI. The BIOS is also a part of
ACPI, and on occasion, there are errors in the ACPI tables. But then,
we'd see some evidence out there, with respect to that model, if
that were the case. (The Linux people spot more of these errors
and rough edges, as part of the development process of doing
ACPI for Linux.)

Your symptoms don't match anything I've run into, which is why I'd
start with the exact models, and see if someone else has seen it.

Apple computers have had problems like that, where the addition
of a PCI card, will lead to a failure to enter sleep state. But
that's not your situation either, as you seem to be in S1 to start,
and somehow switch to S3 later via timer.

I've seen a claim, with respect to "dumppo", that the utility will
print the name of a driver which doesn't support a particular
ACPI state. I've yet to see such a report in the wild, where
a driver name was actually printed out.

The setupapi.log file, in C:\WINDOWS, keeps track of device
installation. It would be one place to look for hardware specifics,
but in this case, I wouldn't really expect to see any hints.
It's not particularly readable, but at least it's not a binary
file and a text editor works.

The Event viewer logs things too, but I've stopped recommending
people look there, because I seldom if ever get any help tracking
down a problem by looking there. It's filled with mainly bogus
messages about "such and such a service successfully started".
When I need specifics, they're never there.

Paul

 

[Noone]

Paul, let me see if I can simplify this. What rail is PS2 on? It would seem
to me that it is on +5 volts vsb since it allows the fans to shut down. And
then, what rail is USB on? It would seem it is on +5 volts since it won't
allow the fans to stop running.

 

[Noone]

Paul, I have just read, and it took a very long time to find this one
paragraph that stated what was on the +5 volt VSB rain, that +5 volt VSB
powers mouse, keyboard, WOL, WOR, etc. This proves why USB will not allow
S3.

 

[Paul]

Paul, I have just read, and it took a very long time to find this one
paragraph that stated what was on the +5 volt VSB rain, that +5 volt VSB
powers mouse, keyboard, WOL, WOR, etc. This proves why USB will not allow
S3.

On a modern motherboard (last three to five years), USB and PS/2 are on
+5VSB. When they removed the jumpers for choosing the power source, they
left those items powered by +5VSB. If the USB ports are powered by +5VSB,
then if a wake condition is programmed into the USB device, the USB
device will be able to wake the computer.

The OS uses information it gets from the driver. If a driver writer
programs into his code, that the hardware device isn't S3 compatible,
then what should happen, is the computer should simply remove S3
from any menu that happens to offer sleeping. Then, it's a matter
of gathering "votes" from all the subsystems, as to whether the
entire computer is S1, S3, or S4 compatible. The "dumppo" program
is supposed to be able to flag (print out) the names of drivers that fail
to offer ACPI states, but I've never seen anyone note that
when they were using dumppo.

Paul

 

[Paul]

Paul, I have just read, and it took a very long time to find this one
paragraph that stated what was on the +5 volt VSB rain, that +5 volt VSB
powers mouse, keyboard, WOL, WOR, etc. This proves why USB will not allow
S3.

I don't think the software is clever enough to know what
rail the hardware is powered by. There is no measurement
capability like that.

The OS can know, whether the port is completely unpowered.
The OC# status signal per port, provides that info (it drops
to logic zero on a power failure). If a USB fuse opens, the
OS can find out about it. But there is no hardware function,
to report power is present, and it's a particular flavor like +5VSB.

Paul

 

[Noone]

So I just read that somewhere around 2009 the VSB rail was increased to 2A
from a previous 10ma. Is that how a modern MOBO can charge a USB device like
a phone or iPod and still enter S3 or S4? Does that mean that all USB
devices on modern MOBO's run on the newly powered VSB rail. As well, if USB
was rerouted via hardware to the VSB rail instead of the +5 volt rail, as it
was prior to the new specification, then I guess that explains why my Vista
and Win7 boxes will enter S3 with a USB keyboard enabled to wake the 'puter.
So now I'm back to enabling my old MOBO to switch USB from +5 volt to the
VSB rail. I mean this not logical or economical...I understand this. But,
could a jumper be installed in these boards? Oh well, at least now I
understand why all this is the way it is.