PC wont turn on after adding new cpu

[Paul]

One thing did happen during the course of the cpu transplant.
There are 4 or 5 capacitors in a row next to the heatsink/fan
The side of my hand accidentally applied a little pressure to the one
on the end of the row when I was negotiating with the HSF retaining
clips.
It seems so minor that I had all but dismissed it.
Perhaps I dislodged or loosened one of it''s connectrs
I can check it, but I'll have to remove the motherboard and look at
the back - perhaps resolder it.
I saw this, it says either remove the power line or reset the cmos.
But is it safe to reset the CMOS while the power line is attached?
If not, surely one implies the other.
if it is I'll try this/
The items I've looked at to check temps are in the Bios PC Health
page.
Also a program called Aida32, which I think is an earlier version of
Everest. The idle temps are about 34C, while about 46 under load.
I've touched the side of the HSF it's generally cool to the touch.
It is certianly making contact with the processor.
I don't see the cpu overheating, perhaps the sensor is bad?
aida32 doesn't report this. What else can I check it with?

I don't have the necessary expertise, knowledge or experience to do
much more of this diagnosis that you and w_tom have suggested.

At this point, I'm getting ready to put back the old cpu and see if it
still occurs. If it does, at least then I'll know that either the
mobo or power supply is bad( I think)
However, what if I put in the old cpu and it's working
perfectly?

I think it is worthwhile to put the old processor back and
repeat the test. So you can see if the behavior stays or
leaves.

With the incident with the capacitor, you can break where the wire
connects to the capacitor, if you applied enough pressure. I doubt
you'd pull the solder loose. If you damaged the capacitor, the
most observable result would happen if the plates of the capacitor
shorted inside the cap. If the capacitor failed open circuit, chances
are you wouldn't see any effect. There are multiple capacitors on
the primary and secondary sides, and it is possible the circuit will
still work just the same (plus or minus a bit of ripple). I wouldn't
worry about it too much, unless you can visually see damage (like
one side lifted higher off the board than the other, implying something
broke inside the capacitor).

(Pictures of an electrolytic)
http://www.microphotonics.com/skyscan/1076/electronics_3.html

In the spirit of swapping stuff, if you get tired of this motherboard,
there are always others.

Paul

 

[w_tom]

I saw this, it says either remove the power line or reset the cmos.
But is it safe to reset the CMOS while the power line is attached?
If not, surely one implies the other.
if it is I'll try this/

All those other suspicions - loose capacitor next to CPU, excessive
heat, reset CMOS - all are not involved. Long before any of those can
create a problem ... first the power supply 'system' must establish
power.

Grey wire - also called "power good" - is the message to everything
to start working. No 'power good'? Then CMOS does nothing. Heat
never causes a problem. Voltage from that capacitor has no
functioning CPU to halt.

Long before anything happens, the power supply controller must order
its power supply ON. Only after power is stable and "power good"
signal created - only then does everything else start working.

Ignore all the 'try this and try that' speculation. Without those
voltages, then nothing else even tries to work. None of those other
functions hinders a 'power on' function. Your problem is best defined
by what happens on purple, green, gray, and power switch wires.
Nothing will work if those voltages fail.

 

[Ed Medlin]

On Tue, 22 May 2007 07:51:06 +1000, Franc Zabkar


Thanks for the info Franc.
My case is slightly different; it occurred after I upgraded the cpu.
Some have said well perhaps the load increased, but both cpu's
consume 89W according to AMD.

Also, mine occurs when ever the PC is shutdown.
If I recycle my AC, it works first time.

I sent a message to Antec to see if they have any clues.

Also left a message with the CPU vendor, but haven't heard
from them yet
I have purchased a digital multimeter and I'm going to measure
the voltages as mentioned in an earlier post.
Dave

It always seems that most PSU failures for me have happened after I have
made some change in components and then power back up. There is really no
reason for it, it just seems to happen that way in my own experience. w_tom
is correct as far as taking voltage readings is concerned. The 'power on'
circuitry is where I have found most problems and if there is any problem
there, it is time to replace the PSU. Of course there could be a MB issue,
but I would definately look to the PSU as the first culprit. All your
symptoms seem to point in that direction.

Ed

 

[dave xnet]

All those other suspicions - loose capacitor next to CPU, excessive
heat, reset CMOS - all are not involved. Long before any of those can
create a problem ... first the power supply 'system' must establish
power.

Grey wire - also called "power good" - is the message to everything
to start working. No 'power good'? Then CMOS does nothing. Heat
never causes a problem. Voltage from that capacitor has no
functioning CPU to halt.

Long before anything happens, the power supply controller must order
its power supply ON. Only after power is stable and "power good"
signal created - only then does everything else start working.

Ignore all the 'try this and try that' speculation. Without those
voltages, then nothing else even tries to work. None of those other
functions hinders a 'power on' function. Your problem is best defined
by what happens on purple, green, gray, and power switch wires.
Nothing will work if those voltages fail.

When the PC is shutdown from within windows and it closes down and
to all intents and purposes it looks as if you pulled the plug - but
it's not off is it?

For example, if I look at the back the nic lights are still flashing.
So some part of the motherboard is still receiving power.

Would you still expect the gray wire to still provide it's
*power ready* voltage when you press the button on the front?
I think I set my meter to 1 significant digit. It read 0,0 on the
gray and I thought this was the problem. Perhaps not?

I've started reading Scott Muellers information here:
http://www.informit.com/articles/article.asp?p=31105&rl=1
It has a lot of info for somebody in learning mode.

appreciate your help -
Dave H.

 

[w_tom]

Again, we are breaking a problem down into parts. Only concentrate
on that part of the problem. Your posts indicate distraction on
things irrelevant. If a power supply 'system' does not work and does
not issue a "Power good" (gray wire) signal to CPU, then nothing
works. Then everything read elsewhere becomes irrelevant.

Yes, computer is powered constantly by purple wire. Apply the
multimeter on that purple wire and monitor voltage while doing
anything. Only time that purple wire voltage is zero is when AC power
cord is disconnected. Adding or removing hardware when purple wire
voltage exists may harm hardware. Just another reason why
'shotgunning' can complicate a problem. Just another example of
knowing 'why'.

Until we have numbers from meter readings, as detailed previously,
then anything else may appear defective - remains unknown. Do not be
distracted by anything else since everything else is not relevant to
this one part of the problem. AND everything else may be good but act
defective if THIS part (a power supply system) is not 'known good'.
Not 'suspected good'. It must be 'known good'. To have 'known good',
first the power supply 'system' must be 'known good'. Not 'suspected
good'; 'known good'.

Posted is measuring two inputs to the power supply controller: purple
wire and power switch. If these work just fine when power supply does
not power up, and if power supply controller output (green wire) does
not work (per previous instructions), then power supply controller is
defective. Other ideas in Scott Mueller's book then would be
completely irrelevant.

BTW, why do we fix things? Certainly not to save money. We do it
to learn. Notice how this concept of breaking a problem down into
parts and then concentrating only on that separate problem -
'following the evidence' - requires 'hands on experience' to
appreciate. It is the difference between those who shotgun verses
those who demand irrefutable facts.

 

[dave xnet]

Again, we are breaking a problem down into parts. Only concentrate
on that part of the problem. Your posts indicate distraction on
things irrelevant. If a power supply 'system' does not work and does
not issue a "Power good" (gray wire) signal to CPU, then nothing
works. Then everything read elsewhere becomes irrelevant.

Yes, computer is powered constantly by purple wire. Apply the
multimeter on that purple wire and monitor voltage while doing
anything. Only time that purple wire voltage is zero is when AC power
cord is disconnected. Adding or removing hardware when purple wire
voltage exists may harm hardware. Just another reason why
'shotgunning' can complicate a problem. Just another example of
knowing 'why'.

Until we have numbers from meter readings, as detailed previously,
then anything else may appear defective - remains unknown. Do not be
distracted by anything else since everything else is not relevant to
this one part of the problem. AND everything else may be good but act
defective if THIS part (a power supply system) is not 'known good'.
Not 'suspected good'. It must be 'known good'. To have 'known good',
first the power supply 'system' must be 'known good'. Not 'suspected
good'; 'known good'.

Posted is measuring two inputs to the power supply controller: purple
wire and power switch. If these work just fine when power supply does
not power up, and if power supply controller output (green wire) does
not work (per previous instructions), then power supply controller is
defective. Other ideas in Scott Mueller's book then would be
completely irrelevant.

BTW, why do we fix things? Certainly not to save money. We do it
to learn. Notice how this concept of breaking a problem down into
parts and then concentrating only on that separate problem -
'following the evidence' - requires 'hands on experience' to
appreciate. It is the difference between those who shotgun verses
those who demand irrefutable facts.

I reread your tiny url and here are some more readings, this time
2 decimal places.

After shutting down from windows:
purple 5.06
green 5.05
grey 0.06
orange 0.00
red 0.07
yellow 0.00

After pressing the front "on' switch pc didn't start
purple 5.06
green 5.03 > 5.05 This changed while the probe was on it. The green
LED in the front went out by itself. put the probe on grey, press
"on" LED relights,
grey stays at 0.06
orange, red & yellow as above.

Reset at the back and actually start the computer, these readings
taken after the pc started:
purple 5.05
green 0.12
grey 4.19
orange 3.36
red 5.11
yellow 12.02

 

[w_tom]

Those power supply voltages are quite healthy and well. Power
supply, as measured when connected to computer, is perfectly good.
Only voltage that indicates failure is green wire when switch is
pressed. Power supply controller does not drive down that green wire
voltage when switch is pressed.

When you reset, I assume you mean remove AC power cord. Therefore
purple wire (+5VSB or plus five volts stand by) becomes 0 volts. This
resets the power supply controller. Then green wire does exactly as
it should - drops to below 0.8 volts and power supply turns on.
Again, orange, red, and yellow wires are all healthy.

Your problem is within the power supply controller - a function on
motherboard. If I remember correctly, this occured when you changed
the CPU. Can this same failure occur with the old CPU? It should
(but then I would perform that experiment just to confirm
experimentally what is currently only known theoretically).

At this point, a power switch and power supply are 'known good'. No
reason to 'shotgun' parts that are 'known good'. Visual inspecton of
the motherboard concentrated in an area where power supply controller
is located (and I cannot tell you where it is on that board - where
green wire's PC trace goes to) for nicks, shorts, metal fragment, bent
over and touching part, bulging parts, etc might discover something.
But that inspection is simply a long shot. Most failures leave no
visual indication.

I don't remember what started this problem. However if it occured
when you changed hardware, then it might be a result of static
electric discharge. Just one of so many reasons and one that can
create mysterious failures immediately or days/weeks later.

 

[dave xnet]

Those power supply voltages are quite healthy and well. Power
supply, as measured when connected to computer, is perfectly good.
Only voltage that indicates failure is green wire when switch is
pressed. Power supply controller does not drive down that green wire
voltage when switch is pressed.

When you reset, I assume you mean remove AC power cord. Therefore
purple wire (+5VSB or plus five volts stand by) becomes 0 volts. This
resets the power supply controller. Then green wire does exactly as
it should - drops to below 0.8 volts and power supply turns on.
Again, orange, red, and yellow wires are all healthy.

Your problem is within the power supply controller - a function on
motherboard. If I remember correctly, this occured when you changed
the CPU. Can this same failure occur with the old CPU? It should
(but then I would perform that experiment just to confirm
experimentally what is currently only known theoretically).

At this point, a power switch and power supply are 'known good'. No
reason to 'shotgun' parts that are 'known good'. Visual inspecton of
the motherboard concentrated in an area where power supply controller
is located (and I cannot tell you where it is on that board - where
green wire's PC trace goes to) for nicks, shorts, metal fragment, bent
over and touching part, bulging parts, etc might discover something.
But that inspection is simply a long shot. Most failures leave no
visual indication.

I don't remember what started this problem. However if it occured
when you changed hardware, then it might be a result of static
electric discharge. Just one of so many reasons and one that can
create mysterious failures immediately or days/weeks later.

Thanks for the further info.
Here's a link to MB if you'd like a view.
http://www.newegg.com/Product/ShowI...38-269-16.jpg&Depa=1&Description=BIOSTAR+6100

It did first occur after a cpu was changed.
I'll put back the old cpu tomorrow and report back.
Thanks for the insights!

Dave

 

[dave xnet]

Thanks for the further info.
Here's a link to MB if you'd like a view.
http://www.newegg.com/Product/ShowI...38-269-16.jpg&Depa=1&Description=BIOSTAR+6100

It did first occur after a cpu was changed.
I'll put back the old cpu tomorrow and report back.
Thanks for the insights!

Dave

I put back the old CPU, a single core AMD64 3200+

The PC is behaving absolutely fine, all trace of the problem
has disappeared. Which really brings me back to my
original thought - is there something wrong with the new cpu?


Dave

 

[w_tom]

I put back the old CPU, a single core AMD64 3200+

The PC is behaving absolutely fine, all trace of the problem
has disappeared. Which really brings me back to my
original thought - is there something wrong with the new cpu?

Only demonstrated is an incompatibility between CPU and
motherboard. Is the incompatibility between that model CPU and that
modem motherboard? Or between that unique CPU and unique
motherboard? We don't know. Again, follow the evidence. All we know
is that new CPU and that power supply controller don't work when
together.

Does motherboard manufacturer claim that CPU (of that date code) is
compatible with that motherboard? If not listed as compatible, then
that is (maybe) the problem.

Then it gets even more interesting. See all those little square
chips? They are bypass capacitors. Enough must exist so that voltage
differences between IC do not exist. But when new parts draw more
power, then more of those little squares must be installed or
relocated. This is simply one example of why some motherboards don't
work with some processors designed after the motherboard was sold.
Just an example of why two completely compatible parts might not work
- and why we repair things simply to learn these 'statistical'
concepts.

Meanwhile, we can address something mentioned long ago. You thought
a CPU adjacent capacitor (those upright cylinders) might be loose?
Those capacitors provide power to the CPU. If disconnected, then one
lead is broken; capacitor will rotate around the other lead. I doubt
it is busted since old CPU still works.

To beat that dog again, problem was traced down to specific parts by
'following the evidence'. Since your posts provided technical details
(especially numbers), then useful replies were possible. Mixing
fundamental electrical concepts (theory) with experimental evidence
has defined the problem down to a specific compatibility issue.

Where do we go from here? First check that that unique processor
date code is compatible with that motherboard. Then maybe exchange
that new CPU for another. However in the 'clone computer' business,
observed are many components that are supposed to be compatible but do
not work. Three months later, the manufacturer had quietly
(mysteriously) terminated that part.

 

[dave xnet]

Only demonstrated is an incompatibility between CPU and
motherboard. Is the incompatibility between that model CPU and that
modem motherboard? Or between that unique CPU and unique
motherboard? We don't know. Again, follow the evidence. All we know
is that new CPU and that power supply controller don't work when
together.

Does motherboard manufacturer claim that CPU (of that date code) is
compatible with that motherboard? If not listed as compatible, then
that is (maybe) the problem.

Then it gets even more interesting. See all those little square
chips? They are bypass capacitors. Enough must exist so that voltage
differences between IC do not exist. But when new parts draw more
power, then more of those little squares must be installed or
relocated. This is simply one example of why some motherboards don't
work with some processors designed after the motherboard was sold.
Just an example of why two completely compatible parts might not work
- and why we repair things simply to learn these 'statistical'
concepts.

Meanwhile, we can address something mentioned long ago. You thought
a CPU adjacent capacitor (those upright cylinders) might be loose?
Those capacitors provide power to the CPU. If disconnected, then one
lead is broken; capacitor will rotate around the other lead. I doubt
it is busted since old CPU still works.

To beat that dog again, problem was traced down to specific parts by
'following the evidence'. Since your posts provided technical details
(especially numbers), then useful replies were possible. Mixing
fundamental electrical concepts (theory) with experimental evidence
has defined the problem down to a specific compatibility issue.

Where do we go from here? First check that that unique processor
date code is compatible with that motherboard. Then maybe exchange
that new CPU for another. However in the 'clone computer' business,
observed are many components that are supposed to be compatible but do
not work. Three months later, the manufacturer had quietly
(mysteriously) terminated that part.

Hi w-tom,
My Biostar motherboard has a one year warranty, while the power supply
has at least that much. The CPU only has the 30 day OEM warranty,
but I'm just within that. I did speak to Antec the PS people,
and they're quite willing to give me an RMA so I can send it in for
possible replacement.
I have put the cpu in a padded mailer and returned it today.
They'll attempt to replace if thy've some left, otherwise refund.
(It's the socket 939 AMD suddenly dropped this a few months ago,
so nothing new being made.)
It's been very interesting., in theory, I could get each part replaced
assuming they're still available.

I did try to gently rotate the knocked capacitor - it didn't turn,
so I assume it's still attached. You might be right about some
underlying incompatability (especially if the replacement
cpu does the same thing) But it's not something I've been unable to
find any evidence on - it would have to be isolated cases like mine.
The parts are certainly *meant* to work together.
The mtherboard was produced while the chip was in existance.
And has full BIOS recognition. I do have the cpu serial number,
perhaps I can initiate some digging there - although
I'm not sure if AMD would help me, seeing as it's an OEM part.

Regarding the possisbility that the new cpu may "draw" more power
than the old, thus the power supply is simply too small - both CPU's
are supposed to consume 89W according to AMD. I took this to mean
both cpu's had the same demand power-wise.
Lets wait and see what happens when the new cpu arrives. If it's
still not working, I could contact the mobo vendor next!

I love these kind of problems myself, but I don't have the
experience/education to have much *deep* electrnics
knowledge. Still it hasn't stopped me with some relatively basic
jobs. I've fixed a few vcr's, changed transistors in amplifiers,
replaced parts in the vertical circuit of a Panasonci TV,
but in all those instances, I had help in that the bad parts were
visible, or I had advice on what exact parts to inspect/change.

 

[w_tom]

Regarding the possisbility that the new cpu may "draw" more power
than the old, thus the power supply is simply too small - both CPU's
are supposed to consume 89W according to AMD. I took this to mean
both cpu's had the same demand power-wise.
Lets wait and see what happens when the new cpu arrives. If it's
still not working, I could contact the mobo vendor next!

Numbers from multimeter completely exonerate the power supply. Not
only were numbers 'good'. Power supply numbers were well within
'good'.

Some CPUs have a date code or stepping number. Slight differences
in same CPU made at different times or in different fabs.

Now we have to wait for the thrilling 'season ending' conclusion -
motherboard or CPU. And nobody gets to vote.