Small Form-factor Power Supply Failure

I've got a Dell Inspiron 531S -- one of those slimline desktops -- that
will not power up, nor show any sign of life.

Pulling the power supply and hooking up a known good unit permits the
machine to boot normally. Putting one of those power supply testers on
the suspect PSU shows normal readings.

I know that those PSU testers are not infallible, but I'm surprised that
a unit that is so dead *in* a machine, will show any readings on the
tester, much less nominal ones.

The reasonable move from here, I guess, is to replace the power supply,
but I invite you to speculate as to what might be going on to give these
apparently discordant readings.

Thanks.

Grinder wrote:
I've got a Dell Inspiron 531S -- one of those slimline desktops -- that
will not power up, nor show any sign of life.

Pulling the power supply and hooking up a known good unit permits the
machine to boot normally. Putting one of those power supply testers on
the suspect PSU shows normal readings.

I know that those PSU testers are not infallible, but I'm surprised that
a unit that is so dead *in* a machine, will show any readings on the
tester, much less nominal ones.

The reasonable move from here, I guess, is to replace the power supply,
but I invite you to speculate as to what might be going on to give these
apparently discordant readings.

Thanks.

You need +5VSB to run the front power switch. If the motherboard overloads
+5VSB, and the power supply shuts it off, then you won't be able to start
the computer.

Asus motherboards have an ideal feature for this issue. They have a green
LED wired directly to +5VSB, for monitoring purposes. If your motherboard
had that LED, you could immediately identify whether the +5VSB was present
before, during, and after you press the front panel power button. The green
LED should remain steadily lit, and not glitch while doing the test.

Other brands of motherboards lack that convenient LED. (My Asrock doesn't
have the LED, and I miss it.)

The power supply tester likely doesn't have a dummy load for +5VSB. It
has a dummy load for a rail like +5V or perhaps +12V. And as such,
since the power supply tester puts next to no load on +5VSB, then it
isn't a reliable test that the supply is ready to go.

In addition to delivering power, a power supply also delivers the
Power_Good signal. You can have fans spinning, and solid meter
readings on +3.3V, +5V, and +12V, and yet the motherboard won't
actually start. Without Power_Good asserted, the motherboard
stays in reset. But this isn't your situation, because you can't
even get the fans to spin. So you're missing +5VSB - that's my guess.
And if there was one of those green LEDs present, it makes it
dead easy to check out +5VSB. If you don't have the LED, you'd need
a multimeter.

Paul

On 6/22/2011 1:09 PM, Paul wrote:
Grinder wrote:
I've got a Dell Inspiron 531S -- one of those slimline desktops --
that will not power up, nor show any sign of life.

Pulling the power supply and hooking up a known good unit permits the
machine to boot normally. Putting one of those power supply testers on
the suspect PSU shows normal readings.

I know that those PSU testers are not infallible, but I'm surprised
that a unit that is so dead *in* a machine, will show any readings on
the tester, much less nominal ones.

The reasonable move from here, I guess, is to replace the power
supply, but I invite you to speculate as to what might be going on to
give these apparently discordant readings.

Thanks.

You need +5VSB to run the front power switch. If the motherboard overloads
+5VSB, and the power supply shuts it off, then you won't be able to start
the computer.

Asus motherboards have an ideal feature for this issue. They have a green
LED wired directly to +5VSB, for monitoring purposes. If your motherboard
had that LED, you could immediately identify whether the +5VSB was present
before, during, and after you press the front panel power button. The green
LED should remain steadily lit, and not glitch while doing the test.

Other brands of motherboards lack that convenient LED. (My Asrock doesn't
have the LED, and I miss it.)

The power supply tester likely doesn't have a dummy load for +5VSB. It
has a dummy load for a rail like +5V or perhaps +12V. And as such,
since the power supply tester puts next to no load on +5VSB, then it
isn't a reliable test that the supply is ready to go.

In addition to delivering power, a power supply also delivers the
Power_Good signal. You can have fans spinning, and solid meter
readings on +3.3V, +5V, and +12V, and yet the motherboard won't
actually start. Without Power_Good asserted, the motherboard
stays in reset. But this isn't your situation, because you can't
even get the fans to spin. So you're missing +5VSB - that's my guess.
And if there was one of those green LEDs present, it makes it
dead easy to check out +5VSB. If you don't have the LED, you'd need
a multimeter.

Thanks, I'll check it no with my multitester (no LED on the board.)

Is it a reasonable presumption that the power supply is the fault if
another known good power supply can get the system to boot? I would had
to get a replacement PSU only to find out the fault is in fact in the
motherboard.

Grinder wrote:


Thanks, I'll check it no with my multitester (no LED on the board.)

Is it a reasonable presumption that the power supply is the fault if
another known good power supply can get the system to boot? I would had
to get a replacement PSU only to find out the fault is in fact in the
motherboard.

The fault isolation isn't that easy. A motherboard can fail to
drive PS_ON# low, when you press the button. So a failure to start
can be a motherboard defect. First you check for +5VSB, and then
you contemplate "dead motherboard" and an inability to drive PS_ON#
properly.

The power supply itself can be "deaf" to PS_ON#. Again, with
your meter, you check the voltage on PS_ON# both before
and after you push the button. Before you push the button,
the voltage on PS_ON# should be close to the same value as
you observe on +5VSB. After you push the power button on the
front of the computer, the level on PS_ON# should drop to
logic 0. This would be in the vicinity of 0.4 to 0.8V or
so (it can't be exactly zero, due to the drop across the
transistor driving the signal).

So now you have two things to check.

1) Voltage on +5VSB before, during, and after pushing the button.
Should stay "solid" 5V the whole time. On an Asus motherboard,
just use the green LED, as a "power quality monitor". The LED
should stay solid, indicating +5VSB is stable at all times.

2) Voltage on PS_ON#. If it stays at 5V, before, during and after
the button push, then the motherboard may be failing to drive
the signal. Perhaps a bad driving transistor or chip. If the
voltage is 5V before pushing the button, and close to zero volts
after pushing the button, then the motherboard is sending
the correct signal to the power supply, but the power supply is
not listening.

When the power supply has an internal fault, it may "latch" the fault
condition, until all power is removed. If the power supply is
"not listening", you turn off power at the main switch, wait
two minutes, and turn it on again. Then try your test again.
Watch for a "fan twitch", which is symptomatic of the supply
halting operating early after T=0.

If the power supply appears not to be listening (is getting good
levels on PS_ON#), change it out and try another. If the motherboard
can't seem to drive out a proper PS_ON# signal, then another
motherboard is probably the ticket. There are yet other fault
possibilities, and for those, I might include the power supply
swap test first, before putting in another motherboard. I'd do that,
because the power supply swap is easy to do, and since you'll be
tearing it apart anyway, probably isn't a lot of incremental extra work.
(You tear out half the cabling, to make room for the power supply
swap. You won't be wiring all the PSU cables for a quick test,
just the main and ATX12V. And if it doesn't work, it's just more
tearing to get the motherboard out.)

If you need a waveform diagram, to put the above words in context,
try page 25 "Figure 7. Power Supply Timing" here. There is information
on levels and such, on subsequent pages.

http://www.formfactors.org/developer...public_br2.pdf

Paul

On 6/22/2011 10:30 PM, Paul wrote:
Grinder wrote:


Thanks, I'll check it no with my multitester (no LED on the board.)

Is it a reasonable presumption that the power supply is the fault if
another known good power supply can get the system to boot? I would
had to get a replacement PSU only to find out the fault is in fact in
the motherboard.

The fault isolation isn't that easy. A motherboard can fail to
drive PS_ON# low, when you press the button. So a failure to start
can be a motherboard defect. First you check for +5VSB, and then
you contemplate "dead motherboard" and an inability to drive PS_ON#
properly.

The power supply itself can be "deaf" to PS_ON#. Again, with
your meter, you check the voltage on PS_ON# both before
and after you push the button. Before you push the button,
the voltage on PS_ON# should be close to the same value as
you observe on +5VSB. After you push the power button on the
front of the computer, the level on PS_ON# should drop to
logic 0. This would be in the vicinity of 0.4 to 0.8V or
so (it can't be exactly zero, due to the drop across the
transistor driving the signal).

So now you have two things to check.

1) Voltage on +5VSB before, during, and after pushing the button.
Should stay "solid" 5V the whole time. On an Asus motherboard,
just use the green LED, as a "power quality monitor". The LED
should stay solid, indicating +5VSB is stable at all times.

2) Voltage on PS_ON#. If it stays at 5V, before, during and after
the button push, then the motherboard may be failing to drive
the signal. Perhaps a bad driving transistor or chip. If the
voltage is 5V before pushing the button, and close to zero volts
after pushing the button, then the motherboard is sending
the correct signal to the power supply, but the power supply is
not listening.

When the power supply has an internal fault, it may "latch" the fault
condition, until all power is removed. If the power supply is
"not listening", you turn off power at the main switch, wait
two minutes, and turn it on again. Then try your test again.
Watch for a "fan twitch", which is symptomatic of the supply
halting operating early after T=0.

If the power supply appears not to be listening (is getting good
levels on PS_ON#), change it out and try another. If the motherboard
can't seem to drive out a proper PS_ON# signal, then another
motherboard is probably the ticket. There are yet other fault
possibilities, and for those, I might include the power supply
swap test first, before putting in another motherboard. I'd do that,
because the power supply swap is easy to do, and since you'll be
tearing it apart anyway, probably isn't a lot of incremental extra work.
(You tear out half the cabling, to make room for the power supply
swap. You won't be wiring all the PSU cables for a quick test,
just the main and ATX12V. And if it doesn't work, it's just more
tearing to get the motherboard out.)

If you need a waveform diagram, to put the above words in context,
try page 25 "Figure 7. Power Supply Timing" here. There is information
on levels and such, on subsequent pages.

http://www.formfactors.org/developer...public_br2.pdf

Thanks Paul, abundant as always.

Grinder wrote:

Is it a reasonable presumption that the power supply is the fault if
another known good power supply can get the system to boot?

Yes, unless the mobo has something like a solder crack at one of the
header pins where the power switch connects, and the pin just happened
to make contact with the mobo when the second PSU was tried.

I know that those PSU testers are not infallible, but I'm surprised that
a unit that is so dead *in* a machine, will show any readings on the
tester, much less nominal ones.

Assuming the PSU tester measured the voltages accurately, then maybe
the PSU has a bad diode pair or bad capacitor connected to it, as that
could make the voltage from them seem OK under light load but droop
under the mobo's load. OTOH it's not reasonable to assume the tester
is accurate, unless it gives a numerical readout of the voltages. I
tried a CompUSA brand tester equipped only with simple lights, and it
said my PSU was fine, despite its +12V rail putting out only about
10.6V, not enough to make hard disks spin.

As a web designing company, we offer complete web design in affordable price in India and throughout the globe, we provide you the best web designing package in a very low budget, if you are searching for cheapest web designing packages, your search completes here. We will take care for your low budget package with a eye catching website design.