motherboards ignoring power_good (grey wire) ?!

[jameshanley39]

I was experimenting with the wires - power_on(green) ,
power_good(grey), and 5VSB(purple)

I had read on a pcguide article, something along the lines of, the
PSU sends 5V down power_good , and 5VSB. And when the power button is
pushed, the motherboard then turns the power supply on (presumably
connecting the green and black)


Oddly though, I have noticed that for the few computers I have, if I
cut the power_good wire on my atx psu , it still works fine. The
motherboard is not checking the power_good.

any thoughts why it isn`t checking it?

TIA

 

[Guest]

I was experimenting with the wires - power_on(green) ,
power_good(grey), and 5VSB(purple)

I had read on a pcguide article, something along the lines of, the
PSU sends 5V down power_good , and 5VSB. And when the power button is
pushed, the motherboard then turns the power supply on (presumably
connecting the green and black)


Oddly though, I have noticed that for the few computers I have, if I
cut the power_good wire on my atx psu , it still works fine. The
motherboard is not checking the power_good.

any thoughts why it isn`t checking it?

It's possible the Power_good input on the motherboard floats up to 5V
when nothing is connected to it.

I remember some very old motherboards, cheap knockoffs of the original
vintage-1981 IBM PCs, that ignored the Power_good signal and instead
generated their own power-on reset signal through an onboard resistor
and capacitor. That may have been done because many power supplies
from that era generated improper Power_good signals, using nothing but
an R-C delay or even a direct connection to the +5V line, rather than
the fairly elaborate circuit found in IBM brand supplies.

 

[Paul]

I was experimenting with the wires - power_on(green) ,
power_good(grey), and 5VSB(purple)

I had read on a pcguide article, something along the lines of, the
PSU sends 5V down power_good , and 5VSB. And when the power button is
pushed, the motherboard then turns the power supply on (presumably
connecting the green and black)


Oddly though, I have noticed that for the few computers I have, if I
cut the power_good wire on my atx psu , it still works fine. The
motherboard is not checking the power_good.

any thoughts why it isn`t checking it?

TIA

Try this reference schematic from Intel.

http://download.intel.com/design/chipsets/designex/BXDPDG10.PDF

Flip to page 32. Look at "ATX Power Connector". On pin 8 of the
connector, there is a signal called "PG1". Notice there is a
8.2K ohm pullup, resistor R216.

If you were to cut the wire at the connector for pin 8, the pullup is
still present on the motherboard side. With the wire disconnected,
the pullup is saying "power still good".

To see whether the motherboard pays attention to power_good,
you'd have to do two things. Cut the wire feeding pin 8.
Then, use a ground wire, to ground pin 8 on the motherboard
side. That will send the "power no good" condition to
the rest of the motherboard.

What should happen, is the motherboard should not POST. The
fans should spin (because all rails would be present), but the
board shouldn't be able to come out of reset.

That motherboard is a dual slot 1, and in addition to checking the
state of PSU power_good, the logic on page 32 also checks that
VRM #1 and VRM #2 are fully operational. On some of those
old boards, the VRM was a plugin module with robust edge
connector. On others, it might have been part of the motherboard.

Paul

 

[jameshanley39]

Try this reference schematic from Intel.

http://download.intel.com/design/chipsets/designex/BXDPDG10.PDF

Flip to page 32. Look at "ATX Power Connector". On pin 8 of the
connector, there is a signal called "PG1". Notice there is a
8.2K ohm pullup, resistor R216.

If you were to cut the wire at the connector for pin 8, the pullup is
still present on the motherboard side. With the wire disconnected,
the pullup is saying "power still good".

To see whether the motherboard pays attention to power_good,
you'd have to do two things. Cut the wire feeding pin 8.
Then, use a ground wire, to ground pin 8 on the motherboard
side. That will send the "power no good" condition to
the rest of the motherboard.

What should happen, is the motherboard should not POST. The
fans should spin (because all rails would be present), but the
board shouldn't be able to come out of reset.

That motherboard is a dual slot 1, and in addition to checking the
state of PSU power_good, the logic on page 32 also checks that
VRM #1 and VRM #2 are fully operational. On some of those
old boards, the VRM was a plugin module with robust edge
connector. On others, it might have been part of the motherboard.

I don't know about electronics beyond using a multimeter..
I can cut the grey wire - and have done..

In layman's terms, How would I ground it "at the motherboard side"- to
use your terminology?

if I look at the atx cable as 2 sides, above the connector is PSU
side, below the connector, is MBRD side.
I cut the wire above the connector.. But below the connector.. things
are a bit cramped!! Considering that the connector is plugged into
the MBRD!


tx

 

[Paul]

I don't know about electronics beyond using a multimeter..
I can cut the grey wire - and have done..

In layman's terms, How would I ground it "at the motherboard side"- to
use your terminology?

if I look at the atx cable as 2 sides, above the connector is PSU
side, below the connector, is MBRD side.
I cut the wire above the connector.. But below the connector.. things
are a bit cramped!! Considering that the connector is plugged into
the MBRD!


tx

There are a couple ways to think about it. If you look at the motherboard
by itself, pin 8 would be exposed on the solder side. You could make
contact there.

But, you're missing another opportunity.

Cut the gray wire in half. That disconnects the Power_Good signal from
the PSU. Then, grab the half of the wire still connected to the motherboard.
Ground that end, which should pull down the signal on the motherboard.

------+ gray pin 8
PSU |------X X---------------X Motherboard
------+ ^ Connector
| connect
| end to GND
|
Ground

HTH,
Paul

 

[Rod Speed]

Theres two power supplies in a computer PSU.... the 5V that is
always on as long as there is power there, this is the 5V standby
supply that switches the main power unit when commanded...

Thats comprehensively mangling what actually happens.

 

[jameshanley39]

There are a couple ways to think about it. If you look at the motherboard
by itself, pin 8 would be exposed on the solder side. You could make
contact there.

But, you're missing another opportunity.

Cut the gray wire in half. That disconnects the Power_Good signal from
the PSU. Then, grab the half of the wire still connected to the motherboard.
Ground that end, which should pull down the signal on the motherboard.

thanks for the explanation.. I did it by using a long wire, stripped
both ends, and connected/hooked one end to the metal within the
case(on the 5.25 enclosure), and I held the other end of the long
wire, onto that that bit you mention, the bit of wire now "sprouting"
from the MBRD atx connector.

And it did have an effect..

For the duration that it was grounded, it went into a state, I can
describe as,
PSU fan on, case fan plugged into PSU, was on. or CPU fan plugged into
PSU, was on.
No Video, ,the LED on the monitor was orange not green.

If the computer was off, and I pushed the on/off button, it would turn
on but into that state.
If the computer was on, it would go into that state.

Of course , if the computer was off, ungrounding it would do nothing.
But if the computer was "on" in that state, with the wire grounded.
And it was ungrounded, then it would reset.. You get the flash of
lights on the ps2 keyboard as it gets its power, you get the BIOS,
e.t.c.
So when it went into the state I described, while grounded, it was a
bit like a "partial" reset.

If the computer was running/"on" i.e. running regularly, or in that
state, e.g. fans, then as soon as it was ungrounded, it would reset..

The reason why I wanted to know more about power_good, was besides the
obvious reason that it is useful generally to know how things work.
I know somebody that had a problem with their PSU and for some reason
or other, did not want to replace it yet, but wanted to do every test
he could on the PSU. I wanted to generalise that problem.. Say I
only have one PSU, or I have others, but still, I want to know exactly
what it was..

I know of testing 12V 5V , with a multimeter.
I know of trying to turn it on by connecting green and a black

His computer would not turn on..when he pushed the button.
I was wondering , what other tests he could do,

and was wondering, maybe his PSU was not giving a power_good signal,
and maybe that was why his MBRD was not turn on, and if he could test
power_good, then maybe he would see that.
Though with my initial test - cutting the power_good wire, it makes no
difference whether it gives it or not!
it seems to me that your test shows that the motherbaord prodcues its
own (and it can be seen with a multimeter).

so in doing that, I didn't learn anything that would help to test a
power supply(testing easily testable areas to see where the problem
lies)!

I have heard of an electronic test to see if there is a response from
the power switch, a test one might do if a computer does not turn on.
I saw it mentioned of on another ng somewhere. But no further
description. Any idea?

thanks

 

[Paul]

thanks for the explanation.. I did it by using a long wire, stripped
both ends, and connected/hooked one end to the metal within the
case(on the 5.25 enclosure), and I held the other end of the long
wire, onto that that bit you mention, the bit of wire now "sprouting"
from the MBRD atx connector.

And it did have an effect..

For the duration that it was grounded, it went into a state, I can
describe as,
PSU fan on, case fan plugged into PSU, was on. or CPU fan plugged into
PSU, was on.
No Video, ,the LED on the monitor was orange not green.

If the computer was off, and I pushed the on/off button, it would turn
on but into that state.
If the computer was on, it would go into that state.

Of course , if the computer was off, ungrounding it would do nothing.
But if the computer was "on" in that state, with the wire grounded.
And it was ungrounded, then it would reset.. You get the flash of
lights on the ps2 keyboard as it gets its power, you get the BIOS,
e.t.c.
So when it went into the state I described, while grounded, it was a
bit like a "partial" reset.

If the computer was running/"on" i.e. running regularly, or in that
state, e.g. fans, then as soon as it was ungrounded, it would reset..

The reason why I wanted to know more about power_good, was besides the
obvious reason that it is useful generally to know how things work.
I know somebody that had a problem with their PSU and for some reason
or other, did not want to replace it yet, but wanted to do every test
he could on the PSU. I wanted to generalise that problem.. Say I
only have one PSU, or I have others, but still, I want to know exactly
what it was..

I know of testing 12V 5V , with a multimeter.
I know of trying to turn it on by connecting green and a black

His computer would not turn on..when he pushed the button.
I was wondering , what other tests he could do,

and was wondering, maybe his PSU was not giving a power_good signal,
and maybe that was why his MBRD was not turn on, and if he could test
power_good, then maybe he would see that.
Though with my initial test - cutting the power_good wire, it makes no
difference whether it gives it or not!
it seems to me that your test shows that the motherbaord prodcues its
own (and it can be seen with a multimeter).

so in doing that, I didn't learn anything that would help to test a
power supply(testing easily testable areas to see where the problem
lies)!

I have heard of an electronic test to see if there is a response from
the power switch, a test one might do if a computer does not turn on.
I saw it mentioned of on another ng somewhere. But no further
description. Any idea?

thanks

You can use the ohms scale on the multimeter, connect the leads to the
case switch, and test the switch. The computer case power switch is
momentary contact and normally open. The multimeter on the ohms scale,
will read zero ohms, for as long as the switch is depressed.

Connecting green and black, grounds PS_ON#. That signal is active low,
and grounding it causes the power supply to come on. The reason you
can ground it, while the motherboard logic signal is still (attempting)
to drive it, is because the logic is open collector. Open collector
logic only pulls strong to ground, and relies on a pullup resistor
(weak) to make the logic one on the wire. With open collector logic,
a number of devices can drive to ground, without any of them getting
damaged. Which is why, as long as the motherboard continues to use
an open collector drive for PS_ON#, you can use "green to black" as a
test of the power supply.

Connecting green to black (PS_ON# to COM) is not latched anywhere, so
if the motherboard signal driving PS_ON# is defective, the motherboard
would then run for only as long as PS_ON# is connected to COM. If, in
Windows, you did a shutdown, while PS_ON# is connected to COM, the
motherboard cannot do a normal shutdown. So Windows should stop running,
but the fans would continue to spin. Thus "PS_ON# to COM" doesn't mesh
nicely with the way everything works - it isn't sufficient as a workaround.
But the method is good as a test that the PSU works, and perhaps the motherboard
is having a problem driving PS_ON#.

Power_good, to me at least, doesn't have much proactive usage. You can
connect your multimeter, and check the state of the signal. If the power
supply is sending a logic 0 on the power_good, then you know the power
supply is not happy internally. If there was a short on the motherboard,
cutting the wire would not change anything in that case. So power_good
is something you check, for its logic level, but otherwise should be
allowed to function normally.

And using your multimeter, to check all the output voltages, also has
diagnostic value. If you hear the fans spin, but the board won't start,
you can check the 3.3V, 5V, 5VSB and so on, and check their voltage
levels.

Paul