Will not power up

[Bob H]

I have an E-Machines T2742 that will not power up after last shutdown. I
changed the power supply and checked out the front panel push button switch.
Is there a relay or something on the main board that controls that function
that I would be able to change?

 

[kony]

I have an E-Machines T2742 that will not power up after last shutdown. I
changed the power supply and checked out the front panel push button switch.
Is there a relay or something on the main board that controls that function
that I would be able to change?

Not exactly, and what there is isn't so likely to fail nor
is user replaceable, though you might try using the Clear
CMOS jumper or pull the battery for a few minutes instead
(either of these while AC power is disconnected from PSU).

Did you confirm that the old PSU had failed for certain or
is it possible it is ok and some other part had failed?
Examine the board for failed capacitors. Take PSU readings
with a multimeter if you have one. Check that all cables
are plugged in like the one for the CPU if present, and that
no cards/etc were disturbed while installing the PSU. Make
sure the PSU input voltage switch is correct per your region
(if a switch for that is present, usually on the rear of the
PSU).

 

[powersupplycables]

I have an E-Machines T2742 that will not power up after last shutdown. I
changed the power supply and checked out the front panel push button switch.
Is there a relay or something on the main board that controls that function
that I would be able to change?

For trouble shooting, I always check one thing first. Physical
connection.
Check to see if your cables from psu to motherboard are securely
connected. Then your psu may not be big enough to handle the
application. Check psu wattage.

jennifer
http://www.powersupplycables.com/

 

[kony]

For trouble shooting, I always check one thing first. Physical
connection.
Check to see if your cables from psu to motherboard are securely
connected. Then your psu may not be big enough to handle the
application. Check psu wattage.

jennifer
http://www.powersupplycables.com/

.... and yet, a machine that "will not power up after last
shutdown", not having been moved nor having sat around for a
LONG time, won't tend to have the cables jumping out of
their sockets all by themselves.

The PSU may indeed not be of high enough capacity or quality
to handle the system, but this system flaw only revealed
itself in longer term use instead of an immediate failure to
run when brand new. IOW, premature aging of the PSU by
running at higher load than would deliver acceptible
lifespan. Then again, it could still be the motherboard or
something else instead.

 

[Bob H]

For trouble shooting, I always check one thing first. Physical
connection.
Check to see if your cables from psu to motherboard are securely
connected. Then your psu may not be big enough to handle the
application. Check psu wattage.

jennifer
http://www.powersupplycables.com/

.... and yet, a machine that "will not power up after last
shutdown", not having been moved nor having sat around for a
LONG time, won't tend to have the cables jumping out of
their sockets all by themselves.

The PSU may indeed not be of high enough capacity or quality
to handle the system, but this system flaw only revealed
itself in longer term use instead of an immediate failure to
run when brand new. IOW, premature aging of the PSU by
running at higher load than would deliver acceptible
lifespan. Then again, it could still be the motherboard or
something else instead.

I have removed the battery for a few minutes but nothing has changed. I have
no way to check voltages to determine if the old and/or new supply is bad.
The front panel switch is good as looked at with an ohmmeter where it plugs
into the main board. Where it goes from there I haven't a clue but it does
something to control the application of 115 volts to the PSU. There is no 115
volts delivered so I have no way to determine the status of either the old or
new supply. Everything is plugged in correctly and all other stuff is in
order. Am I to assume that the computer is a throwaway because of this
strange problem?

 

[Paul]

... and yet, a machine that "will not power up after last
shutdown", not having been moved nor having sat around for a
LONG time, won't tend to have the cables jumping out of
their sockets all by themselves.

The PSU may indeed not be of high enough capacity or quality
to handle the system, but this system flaw only revealed
itself in longer term use instead of an immediate failure to
run when brand new. IOW, premature aging of the PSU by
running at higher load than would deliver acceptible
lifespan. Then again, it could still be the motherboard or
something else instead.

I have removed the battery for a few minutes but nothing has changed. I have
no way to check voltages to determine if the old and/or new supply is bad.
The front panel switch is good as looked at with an ohmmeter where it plugs
into the main board. Where it goes from there I haven't a clue but it does
something to control the application of 115 volts to the PSU. There is no 115
volts delivered so I have no way to determine the status of either the old or
new supply. Everything is plugged in correctly and all other stuff is in
order. Am I to assume that the computer is a throwaway because of this
strange problem?

There are a couple power supply specs here. This will show you what voltages
are on which pins.

24 pin
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf

20 pin
http://www.formfactors.org/developer/specs/atx/ATX12V_1_3dg.pdf

ATX supplies are "soft off". The AC is connected, when you flip the switch
on the back. There isn't much to check there anyway, so no reason to go near
the AC. As long as you see your room lights flicker, when the back switch is
used, that proves you are seeing the inrush current charging up the main
capacitor.

This is not intended to fix your problem, but is a reference, in case you
are curious what is inside the supply. The main capacitors, would be C5
and C6. Inrush current is limited by NTCR1, a device whose resistance
changes when it gets hot.

http://www.pavouk.comp.cz/hw/en_atxps.html

The first thing the supply does, when switched on at the back, is put out
the +5VSB voltage. The motherboard needs that voltage, to run the logic
chip that connects to the front panel switch. When a momentary contact
is seen on the front panel power switch, the motherboard logic chip
latches the condition, and drives a steady zero volt logic level on the
PS_ON# signal. When the supply sees PS_ON# go low, that tells the
supply to switch on the rest of the outputs. So that is the soft
power function. The +5VSB runs all the time, while the other
voltages are delivered to make fans spin, power the processor
and so on.

So, with a meter, you'd check that +5VSB was present. (On Asus motherboards,
there is usually a green LED connected directly to +5VSB, so you can tell
when that voltage is present.) Then, the next thing you'd check, is
what state the PS_ON# signal is in, and whether PS_ON# is close
to 5V when the supply is supposed to be "soft off" and the signal
on PS_ON# is close to zero volts, when the computer is supposed to
be running.

That is where I'd start.

Since all the voltages on the 20 or 24 pin connector are relatively
low voltages, there isn't a shock hazard to speak of. The only thing
you'd want to avoid, is shorting any of the output rails. On a
multimeter, that means avoiding the use of the current measurement
feature of your meter (i.e. don't plug the leads into the current
measurement holes - the internal fuse will blow before anything
really nasty happens).

Paul

 

[Bob H]

... and yet, a machine that "will not power up after last
shutdown", not having been moved nor having sat around for a
LONG time, won't tend to have the cables jumping out of
their sockets all by themselves.

The PSU may indeed not be of high enough capacity or quality
to handle the system, but this system flaw only revealed
itself in longer term use instead of an immediate failure to
run when brand new. IOW, premature aging of the PSU by
running at higher load than would deliver acceptible
lifespan. Then again, it could still be the motherboard or
something else instead.

I have removed the battery for a few minutes but nothing has changed. I have
no way to check voltages to determine if the old and/or new supply is bad.
The front panel switch is good as looked at with an ohmmeter where it plugs
into the main board. Where it goes from there I haven't a clue but it does
something to control the application of 115 volts to the PSU. There is no 115
volts delivered so I have no way to determine the status of either the old or
new supply. Everything is plugged in correctly and all other stuff is in
order. Am I to assume that the computer is a throwaway because of this
strange problem?

There are a couple power supply specs here. This will show you what voltages
are on which pins.

24 pin
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf

20 pin
http://www.formfactors.org/developer/specs/atx/ATX12V_1_3dg.pdf

ATX supplies are "soft off". The AC is connected, when you flip the switch
on the back. There isn't much to check there anyway, so no reason to go near
the AC. As long as you see your room lights flicker, when the back switch is
used, that proves you are seeing the inrush current charging up the main
capacitor.

This is not intended to fix your problem, but is a reference, in case you
are curious what is inside the supply. The main capacitors, would be C5
and C6. Inrush current is limited by NTCR1, a device whose resistance
changes when it gets hot.

http://www.pavouk.comp.cz/hw/en_atxps.html

The first thing the supply does, when switched on at the back, is put out
the +5VSB voltage. The motherboard needs that voltage, to run the logic
chip that connects to the front panel switch. When a momentary contact
is seen on the front panel power switch, the motherboard logic chip
latches the condition, and drives a steady zero volt logic level on the
PS_ON# signal. When the supply sees PS_ON# go low, that tells the
supply to switch on the rest of the outputs. So that is the soft
power function. The +5VSB runs all the time, while the other
voltages are delivered to make fans spin, power the processor
and so on.

So, with a meter, you'd check that +5VSB was present. (On Asus motherboards,
there is usually a green LED connected directly to +5VSB, so you can tell
when that voltage is present.) Then, the next thing you'd check, is
what state the PS_ON# signal is in, and whether PS_ON# is close
to 5V when the supply is supposed to be "soft off" and the signal
on PS_ON# is close to zero volts, when the computer is supposed to
be running.

That is where I'd start.

Since all the voltages on the 20 or 24 pin connector are relatively
low voltages, there isn't a shock hazard to speak of. The only thing
you'd want to avoid, is shorting any of the output rails. On a
multimeter, that means avoiding the use of the current measurement
feature of your meter (i.e. don't plug the leads into the current
measurement holes - the internal fuse will blow before anything
really nasty happens).

Paul

aul
Thank you for your comments but I must tell you that there are NO voltages
present on any pins in the supply chain. I believe the basic 115 Volts is not
presented to the supply and I feel the signal (or whatever it is) is not
getting to the PSU to apply and power it. Again I must state that the front
panel switch, which initiates the power up process checks out as active when
depressed and checked with an ohmmeter where it interfaces the main board.
This is true of two supplies, one new and 300W and the original which is
250W. The system is dead as it can be. Something somewhere takes that front
panel switch closure when depressed and initiates a sequence that applies the
115 Volts to the supply (in my opinion). It is not hapening and I am helpless
without a schematic to follow the path.

 

[Paul]

Thank you for your comments but I must tell you that there are NO voltages
present on any pins in the supply chain. I believe the basic 115 Volts is not
presented to the supply and I feel the signal (or whatever it is) is not
getting to the PSU to apply and power it. Again I must state that the front
panel switch, which initiates the power up process checks out as active when
depressed and checked with an ohmmeter where it interfaces the main board.
This is true of two supplies, one new and 300W and the original which is
250W. The system is dead as it can be. Something somewhere takes that front
panel switch closure when depressed and initiates a sequence that applies the
115 Volts to the supply (in my opinion). It is not hapening and I am helpless
without a schematic to follow the path.

Nope. Doesn't work that way.

Plug in the ATX power supply. Flip the switch on the back to the ON
position. Take a multimeter. Measure between +5VSB and COM. If there is
no +5VSB present, the power supply is dead.

The +5VSB powers the motherboard (logic level power, having nothing to
do with 115VAC). The motherboard has a logic chip. The logic chip is
controlled by the front panel, momentary contact switch. The logic chip
sends a signal which is either zero volts or five volts, to the
PS_ON# pin. That is how the rest of the power supply is controlled.

Sometimes, the motherboard is at fault, or you have a bad momentary contact
switch. Or the driver on the motherboard has become weak, and is unable
to sink a couple milliamps, and bring the level on the PS_ON#
to the close-to-zero volts needed to turn the power supply on.

Once you bring PS_ON# to zero volts (can do that by shorting to COM),
the other outputs should turn on. That won't work, if the +5VSB part
is not working.

The power supply is a switching unit. It has a couple transistors that
switch the high voltage DC rectified on the primary side, to drive the
transformer. The base drive to the transistors, is relatively low level.
It is a simple matter, to turn off the signals to the switching
transistors, and that is what "shuts off the supply". The supply is
actually still running, at least the +5VSB portion is. You should
review the description on the Pavouk site I quoted, as the author
of that site as gone to some amount of trouble to teach us. (The schematic
of a representative ATX supply, was hand copied by examining the PCB
of the supply. The schematic did not come from the PSU manufacturer.)

Notice the thing "second power supply" in the lower left corner of the
power supply schematic. That is what supplies +5VSB, and that runs as
soon as AC is available. That "second power supply" is almost exactly
what you would have found in a wall wart DC adapter 10 years ago.
It consists of transformer, rectifier, filter cap, and linear 5V
regulator. It is not very efficient, but should be quite reliable.

The TL494 provides base drive for the transistors. If the TL494 is
suppressed, by control coming from PS_ON# on the lower left of the
page, then it doesn't provide any signal, and the two transistors
just sit there silent. The transistors have to alternate, like a
two cylinder engine, to make any output. Both the base drive (T2),
and the output (T3), are transformer isolated. That is how the hazard
from the AC is avoided. Transformers separate the user from the AC.
The transformers are "hi pot tested" before leaving the factory,
to a level of 1000V or more (I'd have to look up a typical number).
Thus, they easily withstand the normal AC level applied, without
breaking down, and making a stink.

So start simple - plug in, turn on at the back, and check for
+5VSB.

If the fuse is blown (F1, upper left of Pavouk diagram),
then neither the +5VSB chunk, nor the main supply, will work.
If the fuse blew, it blew for a good reason, like a serious fault
in the main bridge rectifier or the big caps. You'd need to
replace more than the fuse, to make it work again.

If +5VSB is not present, chuck the supply. Your life is worth
more, than poking around inside the supply, so stay out of
there.

Paul

 

[Bob H]

... and yet, a machine that "will not power up after last
shutdown", not having been moved nor having sat around for a
LONG time, won't tend to have the cables jumping out of
their sockets all by themselves.

The PSU may indeed not be of high enough capacity or quality
to handle the system, but this system flaw only revealed
itself in longer term use instead of an immediate failure to
run when brand new. IOW, premature aging of the PSU by
running at higher load than would deliver acceptible
lifespan. Then again, it could still be the motherboard or
something else instead.

I have removed the battery for a few minutes but nothing has changed. I have
no way to check voltages to determine if the old and/or new supply is bad.
The front panel switch is good as looked at with an ohmmeter where it plugs
into the main board. Where it goes from there I haven't a clue but it does
something to control the application of 115 volts to the PSU. There is no 115
volts delivered so I have no way to determine the status of either the old or
new supply. Everything is plugged in correctly and all other stuff is in
order. Am I to assume that the computer is a throwaway because of this
strange problem?

There are a couple power supply specs here. This will show you what voltages
are on which pins.

24 pin
http://www.formfactors.org/developer/specs/ATX12V_PSDG_2_2_public_br2.pdf

20 pin
http://www.formfactors.org/developer/specs/atx/ATX12V_1_3dg.pdf

ATX supplies are "soft off". The AC is connected, when you flip the switch
on the back. There isn't much to check there anyway, so no reason to go near
the AC. As long as you see your room lights flicker, when the back switch is
used, that proves you are seeing the inrush current charging up the main
capacitor.

This is not intended to fix your problem, but is a reference, in case you
are curious what is inside the supply. The main capacitors, would be C5
and C6. Inrush current is limited by NTCR1, a device whose resistance
changes when it gets hot.

http://www.pavouk.comp.cz/hw/en_atxps.html

The first thing the supply does, when switched on at the back, is put out
the +5VSB voltage. The motherboard needs that voltage, to run the logic
chip that connects to the front panel switch. When a momentary contact
is seen on the front panel power switch, the motherboard logic chip
latches the condition, and drives a steady zero volt logic level on the
PS_ON# signal. When the supply sees PS_ON# go low, that tells the
supply to switch on the rest of the outputs. So that is the soft
power function. The +5VSB runs all the time, while the other
voltages are delivered to make fans spin, power the processor
and so on.

So, with a meter, you'd check that +5VSB was present. (On Asus motherboards,
there is usually a green LED connected directly to +5VSB, so you can tell
when that voltage is present.) Then, the next thing you'd check, is
what state the PS_ON# signal is in, and whether PS_ON# is close
to 5V when the supply is supposed to be "soft off" and the signal
on PS_ON# is close to zero volts, when the computer is supposed to
be running.

That is where I'd start.

Since all the voltages on the 20 or 24 pin connector are relatively
low voltages, there isn't a shock hazard to speak of. The only thing
you'd want to avoid, is shorting any of the output rails. On a
multimeter, that means avoiding the use of the current measurement
feature of your meter (i.e. don't plug the leads into the current
measurement holes - the internal fuse will blow before anything
really nasty happens).

Paul

Well Paul I have found some egg on my face when reviewing the
supply data you reference. I did find that the +5VSB is indeed
present on pin 9 (purple wire) on my 20 pin connector. Having
previously checked the front panel momentary switch is working
I can then proceed to short pin 14 (PS_ ON) green wire to ground
momentarily and the supply should come on. If indeed that is the case
(I haven't tried it yet) I have a situation that is not readily repairable
but
a work around could be rigged up so I can get back on line and at least
save things I would otherwise lose.

OK so I shorted PS_ON to ground and up comes the supply as you
indicated. Progress - progress. However I find it is not a momentary
thing and it shuts down when I remove the short. So if I rig up a switch
that will do that I can probably proceed.

I cannot possibly thank you enough for your assistance. I should have
gone this route in the first place instead of assuming a bad supply and
just going out and buying one.

Bob H

 

[w_tom]

You are now learning that the power supply 'system' is far more than
a power supply. The purple wire provides power to another component
called the power supply controller. There was no reason to disconnect
anything (even power switch wire) to appreciate what is happening.

First, that +5VSB (purple wire) voltage is there all the time. So
what exactly is that numbers. Does it exceed 4.87 volts?

Not take a measurement of the green wire (what you shorted with a
paper clip). What are those numbers before and when power switch is
pressed. And then what happens to the gray wire before and when power
switch is pressed. When? Yes, does it rise and then fall off? Or
does the voltage rise (or drop) and stay there?

As you provide those numbers, then what actually exists is
understood AND the next reply tells you more about what you are seeing
and how the power supply 'system' works. But you must provide
numbers. Not just that the +5VSB voltage exists. But what is the
number.

 

[Bob H]

On my not very exotic meter the purple wire reads 4.9 volts. The green wire
reads 2.7 volts both before and after presing the power switch. The gray wire
reads nothing at all either way.


You are now learning that the power supply 'system' is far more than
a power supply. The purple wire provides power to another component
called the power supply controller. There was no reason to disconnect
anything (even power switch wire) to appreciate what is happening.

First, that +5VSB (purple wire) voltage is there all the time. So
what exactly is that numbers. Does it exceed 4.87 volts?

Not take a measurement of the green wire (what you shorted with a
paper clip). What are those numbers before and when power switch is
pressed. And then what happens to the gray wire before and when power
switch is pressed. When? Yes, does it rise and then fall off? Or
does the voltage rise (or drop) and stay there?

As you provide those numbers, then what actually exists is
understood AND the next reply tells you more about what you are seeing
and how the power supply 'system' works. But you must provide
numbers. Not just that the +5VSB voltage exists. But what is the
number.

 

[Paul]

On my not very exotic meter the purple wire reads 4.9 volts. The green wire
reads 2.7 volts both before and after presing the power switch. The gray wire
reads nothing at all either way.

Purple is +5VSB, so that one looks good. The green (PS_ON#) should either
float to 5V, or be pulled low to 0.4V or less. The fact that it is 2.7V both
before and after pressing the front button, means the motherboard logic chip is
not driving it properly.

Inside the power supply, is a pullup resistor. That is what pulls PS_ON# up to
5V (the +5VSB voltage level). All the motherboard normally has, is a transistor
for shorting the PS_ON# signal to ground. So the motherboard wouldn't normally
be actively driving the pin to +5V. The pullup in the power supply should do
that. Such an interconnect method is called "Open Collector", and it allows the safe
shorting to ground of PS_ON#, even if the motherboard is connected.

You can hook a wire to PS_ON# and to COM, and put a full-time (not momentary)
switch across it. The only slightly disturbing thing, is when you select
"Shutdown" in Windows, you might see the "It is safe to turn off your computer",
because Windows will have told the logic chip to flip the PS_ON# signal,
but of course nothing will happen. Then, you'd flip your manual switch to OFF
to finish things.

In terms of what is broken, that is still uncertain in my mind. The motherboard
could be the problem (driver not working on the PS_ON# line). Another test for
the power supply, would be to disconnect the main connector from the motherboard,
and see if PS_ON# is pulled up to +5V or not. If PS_ON# was still 2.7V, I might
suspect the power supply. The supply should not run with the main connector
unplugged, as the PS_ON# should be pulled up to the inactive +5V logic state.

You can continue to run it with a manual switch if you want. The +5VSB supply
has limited current flow, and even if there was an issue between the +5VSB
supply and the PS_ON# signal, no more than a couple amps can flow. The +5VSB
regulator should be thermally protected (shuts down on overheat), and is not
nearly as dangerous as shorting one of the main rails.

Paul

 

[UCLAN]

On my not very exotic meter the purple wire reads 4.9 volts. The green wire
reads 2.7 volts both before and after presing the power switch. The gray wire
reads nothing at all either way.

Sounds to me like the MB is failing to tell the PSU to turn on. Either bad MB
logic, or a bad switch. Bypass switch by momentarily shorting together pins to
which switch is attached. If MB logic then pulls green wire low, then switch
is bad. If not, then MB logic is *likely* bad.

 

[w_tom]

On my not very exotic meter the purple wire reads 4.9 volts. The green wire
reads 2.7 volts both before and after presing the power switch. The gray wire
reads nothing at all either way.

The purple wire is above 4.87 volts (assuming your meter is 3.5
digits; has two digits after the decimal point with is standard for
any $20 meter) meaning purple wire voltage is probably good. Power
supply controller has proper (and clean) power to control the supply.
Green wire voltage must exceed 2.0 volts to keep power supply off. In
most cases, this voltage should be higher; closer to the purple wire
voltage. OK. It's in spec but marginal. For now we just move on.

When power switch is pressed, power supply controller should drop
green wire to below 0.8 volts. It does not. Only other relevant
controller input is the power supply switch. Previously switch was
working according an ohm meter. But better is to measure it without
being disconnected - where it connects to motherboard. Voltage where
it connects to motherboard should be a high voltage - something
approaching purple wire voltage. When switch is pressed, then voltage
between two switch wires should drop to near zero - well below 0.8
volts. This better measurement where switch connects to motherboard
reports what the controller sees.

If switch voltage drops to zero when pressed, then power supply
controller has proper inputs and should be commanding the power supply
on. Your green wire number says 'not'. Therefore power supply
controller is defective.

Now perform some visual inspections concentrating on that controller
area. Carefully remove the motherboard to locate where PC traces from
the green wire and from the power switch connection converge. Inspect
for metal fragments (especially wedged between IC pins), a sliced or
cracked pc board surface (top or bottom), or a standoff that has
somehow shorted to an IC pin or printer circuit (pc) board trace.
Chances are no visual defect will be found. Most all defects have no
visual indication. Therefore the power supply controller is
defective; motherboard must be replaced.

As promised, what those signal wires do. Purple wire always powers
the controller (which is why nothing is added or removed from a
computer without its AC power cord completely disconnected). Then
when power switch shorts its wires together (switch is pressed), then
controller grounds (drops green wire voltage from above 2 volts to
near zero volts) to order power supply on. Supply would put voltages
on orange, red, and yellow wires. Power supply controller watches the
gray wire for maybe two seconds. If that gray wire voltage does not
rise above 2.4 volts, then power supply controller assumes a failure
and shuts off power supply.

You never saw the gray wire voltage rise because power supply was
never told to power on. Had it powered on and provided controller
with the gray wire 'Power Good' signal, then power supply controller
would have released the CPU to start working. Another controller
function is to hold (stop) CPU from running until all voltages are
sufficient. But again, we never got that far because power supply
controller never told power supply to power on.

Solution: replace a defective power supply controller that never
orders power supply on.

One final note. Sometimes (almost never), an internal power supply
problem causes too much current so that power supply controller cannot
order power supply on. Measure this green wire current with the
meter. Disconnect the power supply from motherboard. Then restore AC
power cord. Put meter in DC Current measurement mode at maybe 2
amps. On some meters, the red wire probe is moved to a current
measurement hole. Connect the meter to any power supply black wire
and (red probe) to the green wire. Meter will short the green wire
just like the paper clip did. This meter connection should cause
power supply to power on. Switch the meter setting lower from 2 amps
until a valid current measurement is read. Current should be well
less than 20 milliamps or 0.02 amps (probably about 1 milliamp). If
yes, then any working power supply controller can order that power
supply on by shorting out that less than 20 milliamp current.

Again, this last test is only for a rare (almost impossible) type
failure inside power supply. So rare and considering that 2.7 volts,
then test is probably unnecessary. However, this is a learning
process. The other tests measured voltage. This test taught (safely)
how to measure current.

 

[Bob H]

On my not very exotic meter the purple wire reads 4.9 volts. The green wire
reads 2.7 volts both before and after presing the power switch. The gray wire
reads nothing at all either way.

The purple wire is above 4.87 volts (assuming your meter is 3.5
digits; has two digits after the decimal point with is standard for
any $20 meter) meaning purple wire voltage is probably good. Power
supply controller has proper (and clean) power to control the supply.
Green wire voltage must exceed 2.0 volts to keep power supply off. In
most cases, this voltage should be higher; closer to the purple wire
voltage. OK. It's in spec but marginal. For now we just move on.

When power switch is pressed, power supply controller should drop
green wire to below 0.8 volts. It does not. Only other relevant
controller input is the power supply switch. Previously switch was
working according an ohm meter. But better is to measure it without
being disconnected - where it connects to motherboard. Voltage where
it connects to motherboard should be a high voltage - something
approaching purple wire voltage. When switch is pressed, then voltage
between two switch wires should drop to near zero - well below 0.8
volts. This better measurement where switch connects to motherboard
reports what the controller sees.

If switch voltage drops to zero when pressed, then power supply
controller has proper inputs and should be commanding the power supply
on. Your green wire number says 'not'. Therefore power supply
controller is defective.

Now perform some visual inspections concentrating on that controller
area. Carefully remove the motherboard to locate where PC traces from
the green wire and from the power switch connection converge. Inspect
for metal fragments (especially wedged between IC pins), a sliced or
cracked pc board surface (top or bottom), or a standoff that has
somehow shorted to an IC pin or printer circuit (pc) board trace.
Chances are no visual defect will be found. Most all defects have no
visual indication. Therefore the power supply controller is
defective; motherboard must be replaced.

As promised, what those signal wires do. Purple wire always powers
the controller (which is why nothing is added or removed from a
computer without its AC power cord completely disconnected). Then
when power switch shorts its wires together (switch is pressed), then
controller grounds (drops green wire voltage from above 2 volts to
near zero volts) to order power supply on. Supply would put voltages
on orange, red, and yellow wires. Power supply controller watches the
gray wire for maybe two seconds. If that gray wire voltage does not
rise above 2.4 volts, then power supply controller assumes a failure
and shuts off power supply.

You never saw the gray wire voltage rise because power supply was
never told to power on. Had it powered on and provided controller
with the gray wire 'Power Good' signal, then power supply controller
would have released the CPU to start working. Another controller
function is to hold (stop) CPU from running until all voltages are
sufficient. But again, we never got that far because power supply
controller never told power supply to power on.

Solution: replace a defective power supply controller that never
orders power supply on.

One final note. Sometimes (almost never), an internal power supply
problem causes too much current so that power supply controller cannot
order power supply on. Measure this green wire current with the
meter. Disconnect the power supply from motherboard. Then restore AC
power cord. Put meter in DC Current measurement mode at maybe 2
amps. On some meters, the red wire probe is moved to a current
measurement hole. Connect the meter to any power supply black wire
and (red probe) to the green wire. Meter will short the green wire
just like the paper clip did. This meter connection should cause
power supply to power on. Switch the meter setting lower from 2 amps
until a valid current measurement is read. Current should be well
less than 20 milliamps or 0.02 amps (probably about 1 milliamp). If
yes, then any working power supply controller can order that power
supply on by shorting out that less than 20 milliamp current.

Again, this last test is only for a rare (almost impossible) type
failure inside power supply. So rare and considering that 2.7 volts,
then test is probably unnecessary. However, this is a learning
process. The other tests measured voltage. This test taught (safely)
how to measure current.


So thanks to all for an excellent education regarding the care and feeding of
computer power supplies. My only regret is having immediately gone out and
purchasing a new supply instead of coming here first. I did upgrade to a 300W
supply and now I have the original as a spare. Being an old ham radio
operator I will have no problem working with a wire and switch hanging out of
the computer until I get around to upgrading. My old Win 98 SE machine has
helped me thru this and to think I was going to can it. Thanks again!
Bob H

 

[JB]

The purple wire is above 4.87 volts (assuming your meter is 3.5
digits; has two digits after the decimal point with is standard for
any $20 meter) meaning purple wire voltage is probably good. Power
supply controller has proper (and clean) power to control the supply.
Green wire voltage must exceed 2.0 volts to keep power supply off. In
most cases, this voltage should be higher; closer to the purple wire
voltage. OK. It's in spec but marginal. For now we just move on.

When power switch is pressed, power supply controller should drop
green wire to below 0.8 volts. It does not. Only other relevant
controller input is the power supply switch. Previously switch was
working according an ohm meter. But better is to measure it without
being disconnected - where it connects to motherboard. Voltage where
it connects to motherboard should be a high voltage - something
approaching purple wire voltage. When switch is pressed, then voltage
between two switch wires should drop to near zero - well below 0.8
volts. This better measurement where switch connects to motherboard
reports what the controller sees.

If switch voltage drops to zero when pressed, then power supply
controller has proper inputs and should be commanding the power supply
on. Your green wire number says 'not'. Therefore power supply
controller is defective.

Now perform some visual inspections concentrating on that controller
area. Carefully remove the motherboard to locate where PC traces from
the green wire and from the power switch connection converge. Inspect
for metal fragments (especially wedged between IC pins), a sliced or
cracked pc board surface (top or bottom), or a standoff that has
somehow shorted to an IC pin or printer circuit (pc) board trace.
Chances are no visual defect will be found. Most all defects have no
visual indication. Therefore the power supply controller is
defective; motherboard must be replaced.

As promised, what those signal wires do. Purple wire always powers
the controller (which is why nothing is added or removed from a
computer without its AC power cord completely disconnected). Then
when power switch shorts its wires together (switch is pressed), then
controller grounds (drops green wire voltage from above 2 volts to
near zero volts) to order power supply on. Supply would put voltages
on orange, red, and yellow wires. Power supply controller watches the
gray wire for maybe two seconds. If that gray wire voltage does not
rise above 2.4 volts, then power supply controller assumes a failure
and shuts off power supply.

You never saw the gray wire voltage rise because power supply was
never told to power on. Had it powered on and provided controller
with the gray wire 'Power Good' signal, then power supply controller
would have released the CPU to start working. Another controller
function is to hold (stop) CPU from running until all voltages are
sufficient. But again, we never got that far because power supply
controller never told power supply to power on.

Solution: replace a defective power supply controller that never
orders power supply on.

One final note. Sometimes (almost never), an internal power supply
problem causes too much current so that power supply controller cannot
order power supply on. Measure this green wire current with the
meter. Disconnect the power supply from motherboard. Then restore AC
power cord. Put meter in DC Current measurement mode at maybe 2
amps. On some meters, the red wire probe is moved to a current
measurement hole. Connect the meter to any power supply black wire
and (red probe) to the green wire. Meter will short the green wire
just like the paper clip did. This meter connection should cause
power supply to power on. Switch the meter setting lower from 2 amps
until a valid current measurement is read. Current should be well
less than 20 milliamps or 0.02 amps (probably about 1 milliamp). If
yes, then any working power supply controller can order that power
supply on by shorting out that less than 20 milliamp current.

Again, this last test is only for a rare (almost impossible) type
failure inside power supply. So rare and considering that 2.7 volts,
then test is probably unnecessary. However, this is a learning
process. The other tests measured voltage. This test taught (safely)
how to measure current.


So thanks to all for an excellent education regarding the care and feeding
of
computer power supplies. My only regret is having immediately gone out and
purchasing a new supply instead of coming here first. I did upgrade to a
300W
supply and now I have the original as a spare. Being an old ham radio
operator I will have no problem working with a wire and switch hanging out
of
the computer until I get around to upgrading. My old Win 98 SE machine has
helped me thru this and to think I was going to can it. Thanks again!
Bob H

Did you do like Kony said and reset the cmos.? Removing the cmos battery
for a few minutes is not good enough. Use the jumper to clear cmos if you
have not tried it and make sure the machine is unplugged when you do it.

 

[kony]

Did you do like Kony said and reset the cmos.? Removing the cmos battery
for a few minutes is not good enough. Use the jumper to clear cmos if you
have not tried it and make sure the machine is unplugged when you do it.

Removing the battery usually works, if left out long enough
(most people seem to suggest up to 10 minutes) but certainly
doing both would be more thorough. It's a shame motherboard
manufacturers don't more thoroughly document their board
designs.

 

[UCLAN]

In terms of what is broken, that is still uncertain in my mind. The
motherboard
could be the problem (driver not working on the PS_ON# line). Another
test for
the power supply, would be to disconnect the main connector from the
motherboard,
and see if PS_ON# is pulled up to +5V or not. If PS_ON# was still 2.7V,
I might
suspect the power supply. The supply should not run with the main connector
unplugged, as the PS_ON# should be pulled up to the inactive +5V logic
state.

Not totally accurate. The PS_ON signal level is TTL compatible. There is a
voltage divider of sorts in most PSUs, not just a pull-up to +5vsb. 2.7v is
fine for a TTL high signal, which should keep the PSU off. Anything less than
0.8v is low enough to turn PSU on.

You can continue to run it with a manual switch if you want. The +5VSB
supply
has limited current flow, and even if there was an issue between the +5VSB
supply and the PS_ON# signal, no more than a couple amps can flow.

The current that the PS_ON circuit can provide is limited by the same resistor
that is part of the pull-up voltage divider, normally at least 1K ohm. So
we're talking about milliamps here, at most.

 

[Bob H]

How do I locate the Clear CMOS jumper pins?

Did you do like Kony said and reset the cmos.? Removing the cmos battery
for a few minutes is not good enough. Use the jumper to clear cmos if you
have not tried it and make sure the machine is unplugged when you do it.

Removing the battery usually works, if left out long enough
(most people seem to suggest up to 10 minutes) but certainly
doing both would be more thorough. It's a shame motherboard
manufacturers don't more thoroughly document their board
designs.

 

[JB]

How do I locate the Clear CMOS jumper pins?




Removing the battery usually works, if left out long enough
(most people seem to suggest up to 10 minutes) but certainly
doing both would be more thorough. It's a shame motherboard
manufacturers don't more thoroughly document their board
designs.

Check your motherboard manual. There should be a layout diagram. If you
don't have a manual, go to the manufacturer's site and see if you can
download one.

If you can't find it on a layout diagram, look for an array of three pins
with one jumper located near the battery. The array should be labled "CMOS
CLEAR" or something similar to that.

The jumper position determines whether the CMOS is in the normal ( run) mode
or the clear mode..

If the jumper is on the left two pins, move it to the right two pins for a
few seconds to clear cmos, then move it back to it's normal position. If
you find it initially on the right two pins do the opposite of that.