Boot Problem: Mobo/Power?

[Jim Garrison]

System: Intel DG965WH in an Antec Sonata case with
450w PS.

Symptoms:

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing
connected to the PS is the motherboard. Is this more
likely to be a power supply problem or a motherboard
problem?

 

[JAD]

hows the temps?

 

[gfretwell]

System: Intel DG965WH in an Antec Sonata case with
450w PS.

Symptoms:

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing
connected to the PS is the motherboard. Is this more
likely to be a power supply problem or a motherboard
problem?

First inspect the system board for a foreign object shorting something
out, turn over the case and shake it. Try again.
No joy?
Unplug the system board, plug in one drive and ground pin 14 (usually
a green wire) to a black one, see if it will spin up the drive. That
should be enough load for the supply. If that holds you are probably
back to the system board. If you have a meter you could check the
voltages while it is spinning the drive, just to be sure the supply is
really working.

 

[Paul]

System: Intel DG965WH in an Antec Sonata case with
450w PS.

Symptoms:

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing
connected to the PS is the motherboard. Is this more
likely to be a power supply problem or a motherboard
problem?

Between the DIMM slots, and the main power connector,
is the Standby Power Indicator LED. When the PSU is
switched on at the back, the LED should light, and it
should not wink or blink. It should remain on steady,
even when you press the power button on the front of
the computer.

If the Standby Power LED is flashing, it means +5VSB on the
power supply is cutting out. Either the supply is
weak, or the motherboard is overloading that rail of
the supply. The tie breaker, would be a clamp-on
DC ammeter reading on +5VSB. Without a meter, a
swap of the power supply might help.

Also, check the switch on the back of the PSU, that
sets whether the supply runs at 115V or 230V. Wavering
could be that switch set wrong. Be careful with that
switch! One moment's inattention, flipping it to the
wrong position, could mean kaboom, depending on what
country you are in.

Paul

 

[Rod Speed]

System: Intel DG965WH in an Antec Sonata case with 450w PS.

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing connected
to the PS is the motherboard. Is this more likely to be a power
supply problem or a motherboard problem?

Can be either. Obviously easier to try another power supply.

 

[Rod Speed]

First inspect the system board for a foreign object shorting
something out, turn over the case and shake it. Try again.

That shouldnt produce that recycling effect, it should shut down cleanly
and only restart once the mains is unplugged to reset the power supply.

No joy?
Unplug the system board, plug in one drive and ground pin 14
(usually a green wire) to a black one, see if it will spin up the drive.

Thats rather dangerous if the power supply is dying.

 

[RussellS]

System: Intel DG965WH in an Antec Sonata case with
450w PS.

Symptoms:

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing
connected to the PS is the motherboard. Is this more
likely to be a power supply problem or a motherboard
problem?

---------------------------------------------------------
Jim,

What memory do you have installed? That board requires 1.8V memory and is
very picky about memory in general. I saw identical symptoms on another
P965 chipset board when 2V memory was installed (constant power on-power
off, every 2 seconds or so, with only fans spinning for a moment.) It was
the memory causing the issue. See
http://www.intel.com/products/motherboard/ddr2/index.htm.

 

[gfretwell]

That shouldnt produce that recycling effect, it should shut down cleanly
and only restart once the mains is unplugged to reset the power supply.

That is an awful broad statement without knowing what brand machine
this is or what power supply is installed.

 

[Arno Wagner]

System: Intel DG965WH in an Antec Sonata case with
450w PS.

When I apply power to the system, the fans start but then
everything shuts down for about 1 second. This repeats
ad infinitum, and of course the sytem doesn't even get
close to running POST.

I have disconnected all peripherals, so the only thing
connected to the PS is the motherboard. Is this more
likely to be a power supply problem or a motherboard
problem?

Power supply. Very likely. The way PSU overload protection
works is that it checks the output voltages. If they are too low,
the PSU is shut down. Since the volgates need some time to stabilize
during power-up, this protection is activated with a small delay.
If one of the output voltages is faulty, then the PSU willstart
and stop as you observe.

You can test the PSU without the MB to verify this:

http://www.ee.surrey.ac.uk/Workshop/advice/conns/connPSU.html

lists the contacts in the connector. To start the PSU, shorten
out pin 14 to ground, e.g. to pin 13 or pin 15. The PSU shoud run
as long as pin 14 is grounded. If it does the start-stop thing
you observed, then the PSU is broken. If it comes up and runns,
then it is unclear. Might still be the PSU or might be the MB
in the second case.

Arno

 

[Rod Speed]

(e-mail address removed) wrote

That is an awful broad statement without knowing what
brand machine this is or what power supply is installed.

Nope, thats what the ATX specs say should happen if it sees a short on one of its rails.

 

[Arno Wagner]

On Mon, 5 Feb 2007 17:00:14 +1100, "Rod Speed"

That is an awful broad statement without knowing what brand machine
this is or what power supply is installed.

The cycling effect could well be an interaction between mainboard and
PSU. Of course the PSU should shut down completely on a short and
require unplugging of the mains voltage to restart it. However that
might go over some peoples heads and so a PSU manifacturer could well
have choosen to accept a new (or even the old one) negative flank on
the /POWER_ON line. It might also save a few cents in components
and with the margins today cutting cost in this fashion would not
surprise me.

Arno

 

[DaveW]

The only way to tell which is bad is to install a known working PSU and see
if that fixes it. (Much easier than first replacing the motherboard.)

 

[Jim Garrison]

Thanks much to all of you who responded to my question.

hows the temps?

This happens starting from a cold system.

Between the DIMM slots, and the main power connector,
is the Standby Power Indicator LED. When the PSU is
switched on at the back, the LED should light, and it
should not wink or blink.

Solid as a rock

Jim,
What memory do you have installed?
http://www.intel.com/products/motherboard/ddr2/index.htm.

Kingston KVR667D2N5K2/2G (1.8v DDR2-667). I forgot to
mention that the system has been working for several
months and just started doing this recycling behavior.
Good question, though.

The only way to tell which is bad is to install a known working
PSU and see if that fixes it. (Much easier than first replacing
the motherboard.)

I guess I'll have to go buy a power supply.... :-(

 

[Jim Garrison]

OK, I measured the voltages on the PSU connector with
/PS_ON shorted to ground. I used the pinout chart
at http://pinouts.ru/Power/atx_v2_pinout.shtml for
reference, and found one potential bad voltage: Pin
14 should be -12, but on my PSU it's -10.5.

Is that enough to cause the observed behavior?

Everything else is within 2% of spec (3.3v reads 3.35,
12v reads 12.25, 5v reads 5.01 or 4.99).

 

[JAD]

Thanks much to all of you who responded to my question.


This happens starting from a cold system.

thats not really answering the question. Within seconds the CPU can over
heat if... lets say the heat sink was bumped loose or improperly installed
to start.

 

[gfretwell]

OK, I measured the voltages on the PSU connector with
/PS_ON shorted to ground. I used the pinout chart
at http://pinouts.ru/Power/atx_v2_pinout.shtml for
reference, and found one potential bad voltage: Pin
14 should be -12, but on my PSU it's -10.5.

Is that enough to cause the observed behavior?

Everything else is within 2% of spec (3.3v reads 3.35,
12v reads 12.25, 5v reads 5.01 or 4.99).

In most machines the -12 is only used by the serial port.

 

[w_tom]

OK, I measured the voltages on the PSU connector with
/PS_ON shorted to ground. ...
Everything else is within 2% of spec (3.3v reads 3.35,
12v reads 12.25, 5v reads 5.01 or 4.99).

Even defective power supplies can read 'good' voltages without a
load. Best load for testing is that motherboard.

-12 volts is ignored by voltage monitor circuits; would not cause
power supply cycling. And without some loading, -12 volts may read
incorrect (unstable) anyway. However -12 volts so far out of spec
could be a valuable symptom. We need more useful facts.

More relevant facts are necessary - numbers obtained by a meter. For
example, what does green wire voltage do when supply is power
cycling? IOW is supply cycling on its own or is it cycling because
power supply controller orders cycling? Green wire voltages would
report that necessary information.

Also important are voltages on any one of orange, red, purple, and
yellow wires. With a load, each must obtain minimum voltages of 3.23,
4.87, and 11.7 (numbers defined, in part, because of how meters
work). Purple wire voltage must meet specs even when computer is not
powered.

I suspect one of those voltages is on the low side or does not even
rise sufficient in first seconds after power switch is pressed. If
not, then gray wire voltage will not rise well above 2.4 volts within
seconds after power switch is pressed.

Numbers from these various measurements (without disconnecting
anything) will in but minutes either identify the failure or provide
information to provide useful replies. Disconnecting power supply from
motherboard (to jumper the green wire) simply does not provide useful
information other than 'power supply does start'. Even when on, it
could still be defective. No way to know without measuring voltages
under load - connected to motherboard.

 

[Arno Wagner]

OK, I measured the voltages on the PSU connector with
/PS_ON shorted to ground. I used the pinout chart
at http://pinouts.ru/Power/atx_v2_pinout.shtml for
reference, and found one potential bad voltage: Pin
14 should be -12, but on my PSU it's -10.5.

Is that enough to cause the observed behavior?

Not really. -12V is hardly used on mainboards. The only usage I
know of is for the serial interface. There everyting between -5V
and -15V is acceptable.

Arno

Everything else is within 2% of spec (3.3v reads 3.35,
12v reads 12.25, 5v reads 5.01 or 4.99).

 

[Rod Speed]

Even defective power supplies can read 'good' voltages without a
load. Best load for testing is that motherboard.

-12 volts is ignored by voltage monitor circuits; would not cause
power supply cycling. And without some loading, -12 volts may read
incorrect (unstable) anyway. However -12 volts so far out of spec
could be a valuable symptom. We need more useful facts.

More relevant facts are necessary - numbers obtained by a meter. For
example, what does green wire voltage do when supply is power
cycling? IOW is supply cycling on its own or is it cycling because
power supply controller orders cycling? Green wire voltages would
report that necessary information.

Also important are voltages on any one of orange, red, purple, and
yellow wires. With a load, each must obtain minimum voltages of 3.23,
4.87, and 11.7 (numbers defined, in part, because of how meters work).

Pig ignorant drivel. Nothing like the ATX specs and this
clown hasnt got a clue about how multimeters work.

Purple wire voltage must meet specs even when computer is not powered.

I suspect one of those voltages is on the low side or does not even
rise sufficient in first seconds after power switch is pressed. If
not, then gray wire voltage will not rise well above 2.4 volts within
seconds after power switch is pressed.

Numbers from these various measurements (without disconnecting
anything) will in but minutes either identify the failure or provide
information to provide useful replies.

Not necessarily, particularly when a basic multimeter is used.

 

[Jim Garrison]

It was the power supply. I bought another one, put it
in, and it booted fine. Looks stable now. Thanks for
all the input and information.