No Power (DELL)

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
A few days ago a storm came through and kicked the power off for a few
hours. There was no lightening strike involved close by to the best of
my knowledge mostly wind.

 I have one pc {Dell Inspiron i537-3492} hooked up to an expensive ass
a.p.c. 500 ( Its surge protection plus battery back up) unit
 and the other hooked up to a cheap ass bar unit surge
protecter only.

Go figure the "Dell" is the one that will not power on now. The reset
button only slides from left to right as far as I can tell. I've un
plugged it and let it sit for a day and tried to power on with the
reset in both positions and it will not power on. There is a green
light just below the reset button. It's on when you have it plugged
into the wall and goes off several seconds after you unplug it.

I have the side panel off. I see no visible fuses or breakers. I'm
guessing taking out the power supply is next but I am waiting for some
advise before going any further.
I hope I can fix this my self with a little help from you guys. :-)


and yes... when testing I am pluging directly into a "confirmed
working outlet".

Re: No Power (DELL)

Doomed Soul wrote:
Quoted text here. Click to load it

I don't see a lot in the way of controls on that thing.

And I don't see a "reset" slider here. There is a jumper for clearing
the password. And another jumper for clearing the rest of the CMOS RAM
(256 bytes of parameters, which store BIOS settings of one sort or another).


The green LED on the motherboard, should be connected to +5VSB from the
power supply. The power supply consists of two parts (like, two separate

      AC ---- HVDC ------ 5VSB regulator
                   ------ Regulator for 3.3V, 5V, 12V main rails

If you want to see a PSU in more detail, to see what I'm talking about,
there's a full schematic of one, here.

So if the green LED comes on, that means the fuse isn't blown
in the ATX supply. And a portion of the supply (+5VSB) is working.

For the motherboard to turn on the system, it has the following.

      PSU --- +5VSB -----------|             |
                               | Motherboard |--- PS_ON# ---> back to PSU
      Power_button_on_front ---|             |

If the motherboard was severely damaged some how, such that it couldn't
send PS_ON#, that would prevent it from starting.

If the power supply is damaged on the PS_ON# input, then that
could stop it. Some power supplies become "deaf" to their PS_ON#
input, and it doesn't matter whether logic 0 or logic 1 is
sent to the supply - nothing happens. And that would be a
dead supply.

To test the supply, we'd normally pull it (making a diagram of where
all the wires go, and what color wire is on which side of the connector).

Then, using a paper clip, we'd connect PS_ON# to COM. That would be
performing the same function as the motherboard was performing,
namely, pulling the PS_ON# signal towards ground or zero volts.
If the PS_ON# signal is floating, it floats up to a 5V potential.
And to turn on the main rails, the motherboard "tugs down" on the
PS_ON# signal, sinking a couple milliamps, and the logic low level
on PS_ON# tells the supply to start, spin the power supply fan, and
so on.

The wiring information for a 24 pin supply is here, page 37. You could
test the power supply, stand alone, by connecting pin 15 to 16
(black and green wires). Any black wire would do, but 15 is pretty
close to 16. The "best practices" recommendation, is to have a
dummy load connected to the supply, when connecting pin 15 to pin 16.
If you have an old disk drive, that would be an example of a dummy load.
(If I don't mention that, someone will only complain :-) ) The idea
of the dummy load, is to draw an ampere or two from some rail on the
supply, as the supply is not push-pull in design, only pushes, and
a slight load performs the pull function and helps keep the supply
regulated. I've not had a problem here, testing without a dummy load,
but I have to mention the dummy load for sake of completeness. Some
supplies, on the label, actually mention they have a minimum load spec.
I no longer see supplies with such a crappy design feature.

So your progress so far, is

1) Green light = good. Means +5VSB is present. That should power
    enough of the motherboard, for the motherboard to assert PS_ON# to
    ground level, when the front bezel power button is pressed.

2) Either the power supply is not responding to PS_ON#, or the motherboard
    is no longer capable of sending PS_ON# to the supply. We can't be sure
    which has failed.

If you owned a multimeter, you can probe the 24 pin connector, while it
remains connected to the motherboard. You set the multimeter to volts,
like 20V full scale, and clip the black lead to the chassis. The chassis
is grounded, same potential as any black wire on the harness. Clipping
onto the chassis, is to avoid the danger of shorting something. I use a
shiny screw, like on a rear I/O connector, for my electrical connection.

With the red lead, you can touch the exposed metal, which sits just
inside the plastic shell of the 24 pin connector.

Start from power off at the back of the computer. (Note - do *not* change
the 115V/230V slider on the supply!) Turn on the power at the back of
the computer. The green LED should come on. With the multimeter, verify
pin 9 (purple, +5VSB) reads 5.0 volts. Now, check pin 16 (green, PS_ON#)
as well, and it should be very close to the same 5.0 volt value. Next,
press the front panel power button. Go back to pin 16 (green, PS_ON#)
and verify the voltage is very close to 0.0 volts. Acceptable, might be
less than 0.8V or so. If it's riding around 2.0V, then the motherboard
is having a problem pulling the signal all the way to ground. Closer to
zero volts at that point, would be better.

Anyway, post back, report your pin 16 readings.

The power supply also has a status signal, PWR_OK on pin 8. That's only
important if the power supply fan is spinning. It's possible, for the
power supply to withhold PWR_OK, even if the power supply fan is spinning.
The computer can't boot, until the power supply says it is OK to boot.
But that's not even a consideration (don't need to measure it), unless
the power supply fan starts to spin. The status is only important,
if you get as far as the main rails come on, and the 3.3V, 5V, 12V
are present, and the motherboard still won't run. In such a case,
you could then check PWR_OK and see if the state indicated it was
ready to run. The motherboard has its own internal "I'm OK" logic as
well, but tracing that down is a lot tougher.

Without a multimeter, about all you can do, is pull the supply, and
connect PS_ON# to COM for a test, with the switch on at the back
of the supply. If the fan spins, the supply could be OK, and maybe
the motherboard has a problem. If the fan doesn't spin, then maybe
the main portion of the ATX supply is bad. That's about the best
we can do, in terms of diagnosis without a multimeter. Using
the multimeter, if the chassis isn't too crowded, might be
a bit easier than pulling the supply, if you haven't done it
before. The multimeter can't tell the whole story, except
in certain cases (like, you see a good level of zero volts
on PS_ON#, 5.0 volts on +5VSB, and the supply still won't start,
in which case the second half of the supply is bad).

Another small note - in the event the computer will not respond
to the front power button, you switch it off at the back for one
minute, then switch it on again. Certain faults detected inside
the computer, are "latching" type. To clear the latching faults,
you remove all power from the computer and let it drain. And then,
it should be "re-armed" to detect faults again. I think you've
already done that, but again, I have to mention it. Letting it
drain, is going to happen when you do your tests above anyway.


Re: No Power (DELL)

Quoted text here. Click to load it

1.  Unmentioned here is that you reset the APC UPS correctly.
The larger ones require a reset after a power outage.

2.  APC's advertising stresses their hardware's capacity to
protect you from damage.  APC has an interest in restarting
your Dell a.s.a.p.

Don Phillipson
Carlsbad Springs
(Ottawa, Canada)

Re: No Power (DELL)

Paul Wrote:

Quoted text here. Click to load it

Thanks I didn't realize that fact. I've had this same event occur in
the past with
older models and All I had to do was press a reset switch in the back.
When I originally posted I was hoping there was a simple reset-switch/
breaker/fuse or some sort of safety mechanism i was missing.

Big thanks on posting the diagnostic instructions. It will be at least
a few more days before I can dig that deep into it. Luckily the tower
is not crowded at all and most everything is easy to access. The power
supply should be fairly simple to remove. I can buy a multi meter
fairly cheep.
Paul Wrote:
Quoted text here. Click to load it

After I read this I pulled the BIOS battery and let it sit for another
36 hrs just in case it

Well it didn't :-(
Oh and there is no fan running at any time. I have the voltage switch
on back set on 115v. (I live in U.S.)

One more question before I take out the supply.
Is it possible that there may still be a simple safety mechanism
(reset/fuse/switch) I'm missing?


Don Wrote:

Quoted text here. Click to load it

Yes I did. :-)

Quoted text here. Click to load it

Thanks!! I hadn't even thought of that angle. I will pursue it.

Thanks again,

Re: No Power (DELL)

Doomed Soul wrote:

Quoted text here. Click to load it

You tell me you have a green LED glowing, when the power supply is
plugged into the wall AC, and the rear switch on the supply is
in the ON position.

That tells me the fuse inside the ATX supply is good. And, that
the +5VSB half of the supply is running.

We don't know whether it is making exactly 5.0 volts on the +5VSB rail.
It could probably tolerate some error in that department and still start.

If there is some +5VSB present, the supply main cable
(24 pin) is connected up, then it's a matter of the motherboard
sending that PS_ON# signal, to get the fan running on the power
supply and the main 3.3V/5V/12V rails along with it.

There are other faults that could happen, but the fans would
be running when that happens. If there is no "main" set of
rails running, it's pretty hard for other things to foul up.

The power supply could be "deaf" to PS_ON# (damaged). The motherboard
could be refusing to send PS_ON# (damaged). One or the other or
both could be at fault. In rare cases, you could replace the supply,
still no joy, replace the motherboard, and then hear fans. The power
supply can be tested stand-alone with the "paper clip test". And
some diagnosis of the motherboard, attempted with the multimeter
while things are wired up.


Re: No Power (DELL)

Quoted text here. Click to load it
 >the 115V/230V slider on the supply!) Turn on the power at the back
 >the computer. The green LED should come on.

My dad had me wiring up 3 way toggle switches when I was 6, but I've
never taken any electronics courses, All I know is what my Father
taught me growing up. I do know how to use an Ohm meter tho and your
instructions are easily understood so please be patient with me.

I have no idea what the position was originally on the 115/230. Right
now I have it set to 115v. I thought it was a reset button like on the
older models I've owned in the past. I see no other On/Off switch in
the back. Unless 115/230=On/Off?


Re: No Power (DELL)

Doomed Soul wrote:
Quoted text here. Click to load it

That switch is only fatal to the computer, if you live in a 230V
country, and switch it to 115V by accident. If you did that,
double the rated voltage would get applied to the main capacitor.

If you live in a 115V country, and switch it to 230V, the power supply
will behave "weakly". Once it is switched back to the 115V position
(while powered off), the supply should be its old self again.

It is best to only move the slider, with the power off. (Like, if
you were in a 230V country, moved the switch, then thought better
of it, and moved it back. Doing it with the power off, gives you
another chance to think about what you're doing :-) )

The slider can be seen in the schematic of this power supply. It
fiddles a connection between the bridge rectifier and the main caps.
The 470uF caps store a lot of energy, and help the supply "ride through"
short power outages of under a second. The 470uF caps are also why
we can use cheap UPS boxes, without the computer crashing when the
UPS switches over. The hold-over of the supply is what makes the


To use the multimeter:

1) Stick the red and black wire set, in the "volt/ohm" holes, not the
    "current" hole. I generally *never* measure current with my multimeter.
    My meter leads always stay in the same two holes. Most of my diagnostics
    can be completed with voltage. I have a separate meter for current
    measurements, with a tremendous operating range (up to 400 amps).

2) The meter should be rated for the job. If you're going to measure
    something which is 1000V, then the meter has to be rated for it.
    I remember seeing a multimeter ruined in the lab, when an idiot
    connected one to a super-high voltage supply. The guy was working
    on his Masters degree in electronics.

3) Set the multimeter to the volts range. If there is something that
    distinguishes between "AC" or "DC", set that as well. Inside the PC,
    the voltages are "DC". The only AC would be on the primary side of
    the power supply. And there is no reason to be working inside there!

4) Clip the black lead to the chassis, which is ground. The chassis is
    safety ground on the AC side, and is also the zero volt DC potential
    level of the secondary side. If you have probe tip test leads, you
    slip an alligator clip over the probe tip, then clip the alligator
    tip onto a shiny I/O screw in the I/O plate area. You do it this
    way, so there's no way to short the probe tips together. Big sparks
    could result if the tips got shorted, and you were measuring +12V.
    Now, you can make voltage measurements with the red probe, and only
    move the red probe around. You can stick the red probe, into the
    back of the 24 pin connector, and get it to touch metal in there.
    The plastic shell of the connector, prevents the probe tip from
    shorting two adjacent wires.

5) The multimeter can either be autoranging (which I find annoying), or
    manually set. Select a voltage range suitable for checking 3.3V/5V/12V
    like the 20V full scale range. If you set it to 200V full scale range,
    then the readings will have one less digit to offer you. If you set it
    to 2V full scale, then all the readings will be "overrange" and the
    display will flash its displeasure.

The meter is only accurate to perhaps +/- 1.5% or so. That's how much
the meter can be off by. There are various voltage references, and
in a "real" lab, there would be a requirement to calibrate the
instruments at regular intervals. I used to get a laugh at work,
when these dudes with "extra white" dress shirts would show up,
while the rest of us were wearing jeans and casual attire, and
they'd spend a week calibrating stuff. I'd never seen any of
our test gear, that far out of calibration, but doing the whole
lab like that was considered to be some part of ISO9000 compliance.
The "extra white" dress shirts were calibrated I guess. Whereas
our dingy clothing was in the uncalibrated state. At least it
was easy to keep track of where they were located.


Re: No Power (DELL)

I finally found time to open this thing up and look around. Before I
start attempting to perform the readings you posted I first wanted to
see if I could find a switch, fuse, reset, or anything. I even pulled
the power supply and removed the casing and I could not find anything
at all. As far as I can see there is no visible on/off switch or fuse.
Only switch I can find is the main switch at the front of the tower.
I'm not saying it's not there, I'm just saying I can't find it.
In case I'm missing something,...
 I just wanted to post this before I put everything back together and
start testing.


Site Timeline