M
mynick
What you should get between those pins of a good PC mobo when you
test with ohm-meter
test with ohm-meter
A
Adrian C
Normally, something other than a dead short.
M
mynick
are those directly connected to 'green and black wire' on atx power
connector on motherboard
A
Adrian C
No, there normally is a transistor to switch that - and that is after a
circuit powered by the standby 5V supply (for a typical ATX rig) that's
involved in other power monitoring stuff. What you can measure as
resistance across the contacts could be anything, and not really conclusive.
What's the problem?
Maybe a read of the following may help
http://www.aitechsolutions.net/pchwtrblsht.html
P
Paul
mynick said:
Not exactly.
First, look at the power on button on your PC.
It is a normally open, momentary contact switch.
When you push the button, it creates a pulse.
The logic input on the motherboard, has a pullup to +5VSB,
and when you push the button, the logic signal is shorted
to ground. A current of a milliamp or less may flow
through the switch. (So the front panel switch can be
a flimsy piece of crap, and still work. The front panel
switch doesn't need a multi-amp current rating or anything.)
The signal enters one of the motherboard ICs, and is
conditioned. It is eventually converted into an active low signal
called PS_ON#, driven by an open collector driver. The
motherboard IC means there is no direct path, from front
panel power on switch, to the PS_ON# signal. The motherboard
IC doing the conditioning, is running off +5VSB at the time
you push the button. If +5VSB is not available, then the
signal from the switch can't be conditioned, and can't be
acted upon.
(Note - in the following, I'm illustrating the principle
of cause and effect, not the timing. I didn't verify that
the timing looks exactly like this. But it gets the idea
across.)
PWR -----+ +------ Momentary pulse, active low
(Mobo | |
Header) +---+
PS_ON# -----+ OFF
(main |
20 pin) +---------- ON
The decoupling is more apparent, if you attempt to turn off
the PC, after the PC is booted. They have an option to check
a timer, which validates the state change on the PWR switch. You
have to press the front panel switch for at least 4 seconds,
to get the PC to power off. And the switch can be set in the BIOS,
to either do a controlled shutdown of the PC, or do a power off
instead. In the following diagram, I'm showing the "immediate"
power down option in action (it's how my PC is set up right now
in the BIOS). So after the four seconds is up, the power just
goes OFF, without warning the OS. This gives a "dirty" shutdown,
and potentially needs a CHKDSK later, to fix the file system.
PWR -----+ +------
(Mobo | |
Header) +---------------+
PS_ON# <- 4 sec -> |--------- OFF
(main |
20 pin) -----------------+ ON
For a sample motherboard schematic, you can take a look at
this old 440BX design.
http://www.intel.com/design/chipsets/designex/BXDPDG10.PDF
On page 32, B_SUSC drives pin 14 on the ATX power connector. That
is the green wire (PS_ON# signal) in the ATX standard. B_SUSC
stands for "buffered SUSC signal".
On page 18, you can see the creation of the B_SUSC (PS_ON#) signal.
A 74F07 open collector driver is used. That is a beefy OC driver,
with lots of current sink capability to ground. Modern PCs
are probably using something a bit weaker than that. To operate
PS_ON#, probably requires sinking a milliamp or two (I don't
know the exact figure right off hand). It shouldn't need a lot
of beef, but the beauty of the 74F07, is it is more likely to
survive all insults. Occasionally, on modern motherboards, the
equivalent to the 74F07 function, fails to sink properly to
ground (logic 0).
The SUSC# signal is coming from the Southbridge. So that is where the
"conditioned" control signal, comes out of the motherboard chipset in
this case.
Now, still on page 18, you can see in the Power Management section
of the Southbridge IC, they have a "PWRBT#" (Power Button) signal,
which is active low. That is the signal the Southbridge is going
to be looking for a pulse on. The power button circuit is back on page 32.
And on page 32, they kinda ruined my explanation. They chose to use
a momentary high pulse from the switch (switch pulls to 3VSB), plus
a CMOS Schmitt trigger/inverter to clean up the edge of the signal. The
74LVC14 turns that signal upside-down again, so as the PWRBT#
signal leaves page 32, it is an active low pulse. But as far as
I know, modern motherboards don't have that additional step.
The switch would be set up to pulse low, so the 74LVC14 would
not be present.
Using an ohmmeter, on the PWR/GND pair on the motherboard panel
header, should have little to do with the PS_ON# signal on the
main 20 or 24 pin cable, as they're separated by the logic in
at least one chip. In the Intel schematic, that was the Southbridge.
So ohming from PWR to PS_ON# wouldn't be telling you anything.
What you want to do, is check the voltage level on PS_ON# (green wire),
while you're fiddling with the front panel power button. If the
motherboard open collector driver, pulls the PS_ON# signal towards
ground (zero volts), then you should be seeing the power supply
fan come on and the main rails pop up.
On the input side, you'd monitor the voltage between PWR/GND
pair, when you push the front panel button. PWR should
drop to zero volts, for as long as the front power button
is pushed. Alternately, you can connect the front panel PWR
switch to your ohmmeter, and see if it reads zero ohms, when
the button is pushed. Sometimes, the flimsy button breaks,
and when you push the button, it no longer makes a proper
momentary contact.
HTH,
Paul
M
mynick
thanks for great explanation
so in modern mobos the on switch grounds a pulled up line, straight to
southbridge
(Possibly there is a Schmitt trigger/inverter in between the two)
J
JW
The term southbridge is pretty much dead now. Today it's usually referred
to as the ICH or I/O controller hub.
P
Paul
mynick said:
+5VSB +5VSB
| |
Pullup Pullup
Resistor Resistor
| PS_ON# |
PWR X----+---- Motherboard logic ---- Open -------------------+- ...
Collector (to
GND X----+ Driver ATX +
| supply) |
(Front GMD GND
Panel
Switch)
Using an ohmmeter, between PWR and PS_ON#, doesn't tell you anything.
There is a silicon chip in the way.
Paul
P
Paul
John said:
When I google on Veriton M460, I'm seeing a higher than normal number of
problems there.
http://forums.techguy.org/virus-other-malware-removal/798570-bsod-page-error-malware-related.html
http://forums.majorgeeks.com/showthread.php?p=1418793
Things you'd need to list in your query:
1) Was the system in S3 suspend to RAM, or S4 Hibernate, or just shutdown from the menu ?
If you were starting up in the morning from S3, then the RAM could have an issue.
(I have one machine here, that won't reliably start from S3 the next day. The RAM
is good, so that's a motherboard problem. The motherboard can't retail memory contents
overnight.)
If S4, then perhaps the drive isn't "becoming ready" within the timeout interval.
If you're starting from a complete shutdown, that would be about the same scenario
as Hibernate. Does the hard drive diagnostic that you can download for it, pass or not ?
Seagate and Western Digital offer diagnostic programs. (Make sure you've
burned the recovery media for your PC, in case the hard drive needs to be
replaced at some point. Backups to an external hard drive would be nice as well.)
2) Have you tested with some other boot scenario ? For example, set the wake up time,
do a shutdown from the Windows menu. Then, plug a floppy diskette with a copy of
memtest86+ on it, into the floppy drive. When the system starts the next time, as
long as the floppy is first in the boot order, it'll boot from the floppy. The purpose
of this kind of test, is to try to remove the hard drive from the picture. Even better,
would be to unplug the hard drive data cable (so the system can't get hung up,
while probing the hard drive).
http://www.memtest.org (scroll half way down, get the download, use the program
to prepare a boot floppy. After prep, the floppy cannot
be listed - there is no conventional file system on it.)
Have you done a visual inspection of the motherboard recently ? Are there any
bulging or leaking capacitors on the motherboard ?
http://www.badcaps.net/images/caps/kt7/kt3.html
If I could see some root cause, listed in the postings I can see for Veriton M460,
that would give a better direction to look in. It sounds like it could be
a motherboard issue, but I suspect the machine may have been sold, with more
than one motherboard type installed in it. (Which means, some versions of
the machine might have more problems than others, but the users wouldn't
list the motherboard details for us to know.)
You eliminated the power supply, so that's a start. They're a high runner, in
terms of causing problems.
Failing power supplies, also give little hints about their health. For example,
you may hear a muffled "sizzling" sound at startup. Or see a small puff of smoke
go out the back of the machine at startup. Those are signs of failing (leaking)
capacitors inside the supply. Another indicator I use for my personal machines,
is when I notice a fixed speed 12V fan, is beginning to go "off pitch" on its
fan sound. That can be an indicator of impending failure. It implies the
moment to moment drift of the 12V rail voltage, is larger than it used to
be. I used that to predict the impending failure of my very first supply.
That supply still "works" today, but the output voltage on the 12V rail
drops to 7V, with even a single 0.1 amp cooling fan connected to it.
So now, the supply is as "weak" as is physically possible. It puts out
less power, than a digital camera adapter
But technically, the powersupply still works, as under no load, all voltages are present and it's
cooling fan (internal one) still spins. It just can't take any load.
And internally, all the caps are flat, bright and shiny.
Not every failure condition has visual symptoms - but when offered a
visual symptom, go with it.
Paul
P
Paul
John said:
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
P
Paul
mynick said:
Then you'd simplify the setup a bit, and retest.
At this point, are you hearing any BIOS beep pattern or not ?
Possible answers are:
1) No beeps - implies processor can't run very much BIOS code.
RAM is bad in the low memory area. Processor 12V is unplugged.
BIOS code got erased by user accident.
2) One beep - implies processor has completed POST and is booting
the hard drive. Check for hard drive light flashing.
If the flashing "looks like Windows is booting normally",
it could be that only the video card output signal is
defective.
3) Multiple beeps - now we know the BIOS has detected some kind of
subsystem failure. Either video or RAM. If you
have two video devices, remove one and retest.
If you have two sticks of RAM, retest one stick
in place at a time.
When it comes to the display, your computer has multiple video connectors.
The output signal could be on a different connector, than the one
you're currently connected to. Alternately, the system is using the
built-in video, and wasn't able to detect the add-in video card.
Moving the monitor cable from connector to connector,
may be sufficient to sort that out.
You can remove the video card, and connect the monitor to the motherboard video.
If still no luck, power off the system, and use the Clear CMOS jumper.
This is on the theory, that the default BIOS settings include
enabling the built-in video. The system must be powered off, because
some motherboards get damaged, if you "Clear CMOS" with any power present.
I always unplug the computer, when doing this. After Clear CMOS,
if you can see a video screen, you'll need to re-enter the BIOS,
select the boot device and so on.
The fact that your keyboard flashes, implies you're not in (1) above.
So now it's a question of how many beeps you hear, is the hard
drive LED flashing a normal boot sequence, and so on. If could be
just a video issue of some sort.
Paul
W
William Sommerwerck
On closer inspection, I do see a bulging cap near the
Surround the area with absorbent material, then crush the capacitor. You
shold be able to tack-solder a replacement to the protruding leads.
Surround the area with absorbent material, then crush the capacitor. You
shold be able to tack-solder a replacement to the protruding leads.
B
Baron
John Keiser Inscribed thus:
Shush don't tell anyone, the secret is to preheat that area of the
board before you use the solder sucker... ;-)
Having said that some boards have so little clearance in the hole around
the pins that the crushing trick is the only way to free the pin so
that it will come out when heated from the underside, with the
soldering iron.
Shush don't tell anyone, the secret is to preheat that area of the
board before you use the solder sucker... ;-)
Having said that some boards have so little clearance in the hole around
the pins that the crushing trick is the only way to free the pin so
that it will come out when heated from the underside, with the
soldering iron.
B
baron
John Keiser Inscribed thus:
Yes, that works ! If a little untidy.
Yes, that works ! If a little untidy.