new system bootup woes


P

Paul

VanguardLH said:
I've seen where an underpowered or weak PSU (they get limp over time)
won't power up a system (to the POST screen) that is minimally
configured until the HDD is disconnected. Too much power draw on a weak
PSU means no boot or unreliable boot despite the PSU's fans will spin.

This is true. The very first PC I got (late to the PC game),
the power supply on that failed in the limp state. The power
supply still operates, and can give enough power to run a
12V @ 0.1amp fan. Any more load than that, and the voltage
falls off pretty rapidly. But that supply also had a lot of
years of usage on it, so the failure mode isn't a complete
surprise.

Which reminds me, I took a trip to the recycler last week,
and I forgot to load that thing into the car :)

Paul
 
Ad

Advertisements

A

Adam

VanguardLH said:
If shorting the PWR pins on the onboard front-panel pins doesn't work,
and to test the PSU is okay, bypass the onboard logic for soft power up
of the ATX PSU. Short the PS_ON line (green wire: pin 16 on a 24 power
connector, pin 14 on a 20-pin connector) to a ground line (black wire).
See pinout at http://www.smpspowersupply.com/connector_atx_pinout.GIF.
The signal floats high but the onboard logic pulls that line low to tell
the PSU to power up. Shorting it to ground effects the same pull to low
state. If done while the PSU is disconnected from the mobo, some PSUs
won't power up until they sense a load, so attach an HDD.

Front panel power switches can go bad or be defective so the PWR pin
short is a good test; however, if that doesn't work, make sure the PSU
will come up if its PS_ON line is pulled low.

Thanks, the "unconnected supply" test (shorting the green wire to black/gnd
wire) passed since
the PS fan spins.

On a side note, the Antec case's front panel power switches seem really
flimsy so
I may need to go buy a switch button for PWR-GND pins. Any recommendations?
 
D

Dustin

"Adam" <[email protected]_thanks.com> Wed,
17 Dec 2014 06:00:51 GMT in alt.comp.hardware.pc-homebuilt, wrote the
following message:
Host OS: Ubuntu Desktop 12.04 LTS
Motherboard: ASUS Sabertooth 990FX AM3+ AMD SATA 6Gb/s USB 3.0
ATX Computer Case: Antec Three Hundred ATX Mid Tower
Power Supply: OCZ ModXStream Pro 600W Modular


I am having trouble with power to brand new system.

After talking with OCZ tech support and doing a simple test,
we were able to get the PS fan to spin-up. OCZ Tech support
concludes that this is a strong indication that the PS is
functional.

Next, I suspect that the Antec case's I/O panel wiring to mobo pins
may be the culprit. The wiring seems fine to me but still no
power.

Take the wiring off the pins. Locate the power pins, touch them both at
the same time with a small screwdriver; See if your box powers up. if it
does, you've determined the problem is most likely in the panel wiring
and you can track it down from there.
 
D

Dustin

"Adam" <[email protected]_thanks.com> Wed,
17 Dec 2014 07:45:38 GMT in alt.comp.hardware.pc-homebuilt, wrote the
following message:
Oh boy, that's new to me. I'll see if I can find any "how to"
links on troubleshooting with a voltmeter.

huh? Oi... :)

You really should do that when you have the chance. it's a handy tool
and skill to have.
 
P

Paul

Adam said:
Okay, sounds like what the mobo manual calls the standby power (SB_PWR) LED,
which lights up no problem whenever mobo is connected to PS so far.

Make sure any USB loads are disconnected. No iPod or iPad charging.
Disconnect the memory card reader. The purpose of doing that,
is to try to get as much +5VSB power available as possible.

You would check the voltage level on the +5VSB wire of the PSU.
With a multimeter set to the 20V full scale range, probe pin 9 (purple)
and measure with respect to ground. I use an alligator clip on the
multimeter black lead, and clip to a ground screw on the back of the PC
to establish the black wire potential. Then, I have one hand free to work
on probing with the red multimeter wire.

When the 24 pin PSU connector is installed into the motherboard, there
is a little clearance where the PSU wire goes into the back of the connector.
It's possible to put the tip of the red lead, between the wire and the
plastic connector shell. And make electrical contact. That's how you
measure the voltage on the purple wire, while the system is in use.

The PS_ON# wire, when the computer isn't running, should be pulled
high (to the same voltage level as the purple wire you just checked).
So now you've measured green. If the voltage is low (and you haven't
attempted to turn on the PC yet), then the open collector motherboard
driver isn't working right.

But you've seen the green LED, which is a first order indicator
that it's working a bit on +5VSB. If the LED was dim, you'd suspect
PSU trouble.

Now comes the fun test. Connect PWR to GND on the motherboard
panel header (a momentary touch is enough). The fans should
all start to spin. If they don't, and you really think you were
successful in grounding the PWR pin, now you go back and check
the PS_ON# green wire. It should drop to 0.4V to 0.7V or so.
That is the open collector motherboard driver, pulling the
line low and asking the power supply to turn on the main rails.

If the PS_ON# signal is still high ( greater than +2.0V ), then
the motherboard needs to be replaced (under warranty). If the
PS_ON# wire is 0.4V like it is supposed to be, and the PSU
doesn't react, replace the PSU. You've already done the
"unconnected supply" test, so somehow I don't expect the
problem to be an un-reactive PSU. It's more likely at this
point, to be a motherboard open collector driver that
can't pull PS_ON# low. (Or as Vanguard mentioned, a defective
front panel switch.)

What's strange about this whole thing, is the failure frequency
of this interface. I just don't get it. The motherboard has
a couple thousand high tech electrical signals, and that damn
low tech signal is the one that fails. They must cover that
pin with the two minute functional test at the factory (or their
test bench wouldn't turn on), so they must be able to verify
PS_ON# drives low, by virtue that the woman doing the testing
is able to complete all her tests (in the two minutes allocated).
(There was a video available, of that testing process at Asus.
That's how I know it's a woman :) )

Asus used to produce anywhere from 3 million to 5 million motherboards
per month, so that 2 minute functional test is one expensive test to
carry out. But it's essential. They can do boundary scan structural
testing until the cows come home, but a functional test covers things
like that PS_ON# signal. Just to prove the motherboard works. It's
really a final acceptance test, to make sure as many duff motherboards
don't get put into cardboard boxes, as possible.

So we're basically checking a couple voltages here. To fault isolate
and tell whether the PSU or the motherboard should be replaced.
The purple wire runs the Asus green motherboard LED. We want to
check that it reads 5.0V +/- 5% or so. The PS_ON# signal, when the
system is off, should also be 5V (same voltage as purple). When the
momentary switch on the front is pressed, the PS_ON# voltage
drops. And using your multimeter on the green wire, verifies it
dropped to around 0.4V or so. It won't drop to zero, because the
driver doesn't go all the way to exactly zero. It just has to pull
low enough, to be well below the 1.5V to 2.0V threshold on the
thing receiving the signal.

+5VSB (0.0V level +5VSB
| means "run please") |
Pullup \_ Pullup
Resistor \ Resistor
| PS_ON# |
PWR X----+---- Motherboard ---- Open -------------------+- ... control
/ logic Collector (to of main
| GND X----+ Driver ATX + PSU
| | supply) | section
(Front GND GND
Panel
Switch - normally open, momentarily close to operate)

On some supplies, if the PS_ON# signal rides around 1.5V to 2.0V
(in other words, right on the threshold), the power supply actually
comes on half-way. This leads to weak rails, and an inability to supply
full load current. Apparently, the on-off control isn't as "digital"
as it should be. So if the supply appears limp, sometimes it's not
actually the supply. And this is an important reason to be
measuring the green wire, and that it's driven to 0.4V or less.
Otherwise, the wrong component could get the blame. And this
whole procedure is for "fault isolation", identifying the defective
part.

Paul
 
A

Adam

Jonathan N. Little said:
Also most motherboards have a speaker pins on the front panel header, and
they include a little speaker dongle that if you connect it you can get
post beep codes that can help.

The Antec case did not come with a 4-pin mobo speaker for
post beep codes so I bought one. But, the orientation is not clear.
There are only two actual wires (one red and one black separated by two
blanks/GND?).
Connecting the black to "Speaker" pin and the red to "+5V" pin makes
orientation consistent with the other connectors. Hope my guess is right.
:)

As someone else mentioned did the CPU fan start? LED on motherboard light
up? HD spin up? Monitor get a signal or NO SIGNAL message?

No, I did not get CPU fan to spin yet. Still cannot PWR ON the system.
Antec case's POWER button seems flimsy. Will try a switch button instead.
Yes, standby power (SB_PWR) LED lights up no problem whenever mobo
connected.
Can't see VGA_LED once video card is connected. Poor LED placement on mobo.
Not sure about the other tiny LEDs yet. Haven't been able to power ON the
system.
No, no drives (HDD, DVDs, etc.) are connected yet. Just the bare minimum.
No, I haven't gotten to video signal yet. Can't power ON yet.

I shall never take fans spinning for granted again. :)

As I said you should try with minimal connections...remove everything
except video and the connection to the "pwr sw" pins on the front header.
This narrows your search.

OCZ tech support made sure that I disconnect everything except for
one HDD yesterday. I've reconnected the video card.
 
Ad

Advertisements

A

Adam

Jonathan N. Little said:
If the system powers up to where the power supply fan runs then start
looking a the motherboard. Any post codes? Indicator lights on
motherboard? Reseat RAM. Check CPU, remove all cards except video (if not
using onboard) and see if it posts then.

No, not the system, literally the only device connected to
the power supply was just a HDD (no mobo or anything else).
OCZ tech support just wanted to see if the PS fan will spin-up, which it
did.

Next, I connected the mobo with CPU and RAM (but no video card).
OCZ tech support didn't want the video card connected.
Then, the PS fan no longer spins.
 
A

Adam

Dustin said:
"Adam" <[email protected]_thanks.com> Wed,
17 Dec 2014 07:45:38 GMT in alt.comp.hardware.pc-homebuilt, wrote the
following message:


huh? Oi... :)

You really should do that when you have the chance. it's a handy tool
and skill to have.

Yes, sir.
 
M

mike

Thanks, added it to my wish list just in time for Santa. :)
If you intend to build computers, you simply must have a multimeter.
Depending on where you live and what coupon flyers you get, Harbor
Freight will give you a voltmeter sufficient for your task for free.
But I wouldn't recommend any of their offerings if you have to pay for it.
Sears has had what looks like a decent multimeter on sale for $10
recently.

This link should be in everybody's toolbox.

www.pinouts.ru

One caution about testing unloaded power supplies...
depending on the quality of the design, a power supply with
no load on the 5V may have insufficient magnetic flux
to guarantee that all the voltages are in spec.
You can chase your tail forever trying to fix something that
ain't broke.

Are you absolutely sure you have compatible ram?
I've been in situations where ram that worked fine in
one board caused another board to act completely dead.
 
A

Adam

Dustin said:
"Adam" <[email protected]_thanks.com> Wed,
17 Dec 2014 06:00:51 GMT in alt.comp.hardware.pc-homebuilt, wrote the
following message:


Take the wiring off the pins. Locate the power pins, touch them both at
the same time with a small screwdriver; See if your box powers up. if it
does, you've determined the problem is most likely in the panel wiring
and you can track it down from there.

No, shorting PWR-GND with a small screwdriver didn't work either.
 
Ad

Advertisements

A

Adam

mike said:
If you intend to build computers, you simply must have a multimeter.
Depending on where you live and what coupon flyers you get, Harbor
Freight will give you a voltmeter sufficient for your task for free.
But I wouldn't recommend any of their offerings if you have to pay for it.
Sears has had what looks like a decent multimeter on sale for $10
recently.

This link should be in everybody's toolbox.

www.pinouts.ru

One caution about testing unloaded power supplies...
depending on the quality of the design, a power supply with
no load on the 5V may have insufficient magnetic flux
to guarantee that all the voltages are in spec.
You can chase your tail forever trying to fix something that
ain't broke.

Are you absolutely sure you have compatible ram?
I've been in situations where ram that worked fine in
one board caused another board to act completely dead.

RAM:
Crucial Ballistix Tactical 240-Pin DDR3 SDRAM 8 Dual Channel Kit 1866 (PC3
15000) BLT2CP4G3D1869DT1TX0

Motherboard:
ASUS Sabertooth 990FX (pre-R2.0 w/ 1394-FireWire) AM3+ AMD 990FX SATA 6Gb/s
USB 3.0 ATX AMD Motherboard
http://www.amazon.com/dp/B00539LU3E/ref=pe_175190_21431760_M3C_ST1_dp_2

I think this RAM/mobo pairing is compatible, isn't it?
I just re-seated the RAM. Heard the click but still no luck.
 
P

Paul

Adam said:
Thanks, the "unconnected supply" test (shorting the green wire to black/gnd
wire) passed since
the PS fan spins.

On a side note, the Antec case's front panel power switches seem really
flimsy so
I may need to go buy a switch button for PWR-GND pins. Any recommendations?

http://www.amazon.com/HDE-Power-Motherboard-Switch-Button/dp/B00345WH4A

http://www.newegg.com/Product/Product.aspx?Item=9SIA4SR1WK8427

http://www.amazon.com/Switch-Power-Style-Push-Button/dp/B000NWFEZQ

I don't know if a local computer store would have something
like that or not. They wouldn't make a big profit from it.

I just made my own, using a switch from Radio Shack.
I made mine, back when you could buy stuff...

See if you have a computer recycler in town, and they'll
probably have some switches harvested from old computer
cases.

My two switches (one for Power, one for Reset), don't
have the right 1x2 female on the end. I used Molex socket
pins, and put shrink wrap around the pins to insulate them.
It's a horrible solution, but all I could find without
spending a fortune. frontx.com has 1x2 housings and the
right pins, but then that would be a separate order.

1x2 plastic body

http://www.frontx.com/cpx075_8.html

Female pins (strip of ten, enough for five plastic bodies)
Also provides spares if you make mistakes.

http://www.frontx.com/cpx076.html

Example of a switch.

http://www.active123.com/456721-Push-Button-Switch-SPST-3A-OnOff-Black-Square-Cap-Prodview.html

( http://www.circuittest.com/456-721-spst-3a-on-off-black-square-cap.html )

My least-favorite switch :) A pack of ten of
them, to drive you nuts. Works out to $0.85 each.
The problem with these in the past, was high spring
force. It hurts your finger to actuate one of these.
I don't know if the fact they're made by a different
supplier, changed that part of it or not. I'd much
rather take my chances on some other switch. I think
Radio Shack used to stock these. If I was forced to use
these, I'd mount them on a prototyping box, to provide
support when pressing the button. As the switch tends
to twist around in your hand, if pressing it stand-alone.

http://www.active123.com/45610210-P...tch-SPST-1A-OnOff-Red-Cap-Pkg10-Prodview.html

You could use one of these extension cords, as long as
you were careful not to drop it back into the computer
chassis. Either chop the male end of, and fit a switch.
Or momentarily put your screwdriver tip across the male
pins, to turn the computer on.

http://www.svc.com/ele-187.html

Paul
 
P

Paul

Adam said:
RAM:
Crucial Ballistix Tactical 240-Pin DDR3 SDRAM 8 Dual Channel Kit 1866 (PC3
15000) BLT2CP4G3D1869DT1TX0

Motherboard:
ASUS Sabertooth 990FX (pre-R2.0 w/ 1394-FireWire) AM3+ AMD 990FX SATA 6Gb/s
USB 3.0 ATX AMD Motherboard
http://www.amazon.com/dp/B00539LU3E/ref=pe_175190_21431760_M3C_ST1_dp_2

I think this RAM/mobo pairing is compatible, isn't it?
I just re-seated the RAM. Heard the click but still no luck.

I don't know if you know this, but you can connect just
the motherboard, with no CPU or RAM in place, to the
ATX power supply, use the PWR-GND interface, and test
that the PSU comes on. Very little is needed, to carry
out a logic test of the motherboard switch interface.

What happens in that case, is the VCore regulator isn't
enabled. When nothing drives the VID pins on the CPU socket,
the regulator goes to a safe state. No voltage in the
CPU socket is the basic idea. (There are many VCore design
standards, and you should check the VID table for the correct
standard, to know what the "open circuit" state is. The ones
I've checked, default to "zero volts".)

The power switch stuff continues to work, even if no CPU
or RAM are present. I've actually tested that stuff on
motherboards here, as my "first test" on the bench. Not
because it was particularly necessary, but I wanted to
prove a point.

Paul
 
M

mike

RAM:
Crucial Ballistix Tactical 240-Pin DDR3 SDRAM 8 Dual Channel Kit 1866 (PC3
15000) BLT2CP4G3D1869DT1TX0

Motherboard:
ASUS Sabertooth 990FX (pre-R2.0 w/ 1394-FireWire) AM3+ AMD 990FX SATA 6Gb/s
USB 3.0 ATX AMD Motherboard
http://www.amazon.com/dp/B00539LU3E/ref=pe_175190_21431760_M3C_ST1_dp_2

I think this RAM/mobo pairing is compatible, isn't it?

Ram has always been a crap-shoot for me.
Don't think I've ever bought any new ram, so what I have has
cryptic markings. I have one mobo where the vendor and Crucial
disagree on what RAM it takes.
I just re-seated the RAM. Heard the click but still no luck.
Assume you've tried whatever jumper changes to reset the CMOS?
If you had a voltmeter you'd be able to check the CMOS battery.
Or whether there was voltage on the power switch and whether
it went to zero when you pushed the button.

Assume you put the processor in the right way and didn't bend
the socket pins or get a speck of dust in there...or static-zap
the processor.

Had one system that wouldn't boot after I got it home from the
garage sale on a motorcycle. Removing the fan/heatsink and reseating
the processor fixed it.

That begs the question, are you absolutely sure you have a compatible
processor?
I have a Dell system that I tried to upgrade. Same processor family.
Speed compatible. All indications from Intel are that it should work
exactly like the one that's in there.
Problem is that Dell never updated the bios to put in the newer
identifier and the processor is rejected. No boot.

I seem to be long on identifying potential problems and short
on solutions.
 
A

Adam

Paul said:
http://www.amazon.com/HDE-Power-Motherboard-Switch-Button/dp/B00345WH4A

http://www.newegg.com/Product/Product.aspx?Item=9SIA4SR1WK8427

http://www.amazon.com/Switch-Power-Style-Push-Button/dp/B000NWFEZQ

I don't know if a local computer store would have something
like that or not. They wouldn't make a big profit from it.

I just made my own, using a switch from Radio Shack.
I made mine, back when you could buy stuff...

See if you have a computer recycler in town, and they'll
probably have some switches harvested from old computer
cases.

My two switches (one for Power, one for Reset), don't
have the right 1x2 female on the end. I used Molex socket
pins, and put shrink wrap around the pins to insulate them.
It's a horrible solution, but all I could find without
spending a fortune. frontx.com has 1x2 housings and the
right pins, but then that would be a separate order.

1x2 plastic body

http://www.frontx.com/cpx075_8.html

Female pins (strip of ten, enough for five plastic bodies)
Also provides spares if you make mistakes.

http://www.frontx.com/cpx076.html

Example of a switch.

http://www.active123.com/456721-Push-Button-Switch-SPST-3A-OnOff-Black-Square-Cap-Prodview.html

(
http://www.circuittest.com/456-721-spst-3a-on-off-black-square-cap.html )

My least-favorite switch :) A pack of ten of
them, to drive you nuts. Works out to $0.85 each.
The problem with these in the past, was high spring
force. It hurts your finger to actuate one of these.
I don't know if the fact they're made by a different
supplier, changed that part of it or not. I'd much
rather take my chances on some other switch. I think
Radio Shack used to stock these. If I was forced to use
these, I'd mount them on a prototyping box, to provide
support when pressing the button. As the switch tends
to twist around in your hand, if pressing it stand-alone.

http://www.active123.com/45610210-P...tch-SPST-1A-OnOff-Red-Cap-Pkg10-Prodview.html

You could use one of these extension cords, as long as
you were careful not to drop it back into the computer
chassis. Either chop the male end of, and fit a switch.
Or momentarily put your screwdriver tip across the male
pins, to turn the computer on.

http://www.svc.com/ele-187.html

Paul

Or, an extension wire from PWR to GND on a small breadboard? :)
Will that work?
 
Ad

Advertisements

P

Paul

Adam said:
Yes, sir.

This is the multimeter I've used for years. I don't like autoranging
meters all that much, and would rather rotate the selector and
stay set to one particular range.

http://www.repeater-builder.com/radio-shack/22-168/22-168a-face.jpg

I use the right-most two holes for most measurements. Most of
the time, I'm doing Volts or Ohms. I measure current a lot less
often. I would measure currents of say, up to half an amp,
with the multimeter. But nothing higher.

You can get an acceptable meter for $20. The one in that picture
was $100 when new, roughly. One feature that meter has, is an
opto-isolated RS232 interface, so the meter will upload
readings once a second over RS232, and the interface is
opto-isolated, so you can work on high voltage on the meter
end, without damage to the PC end.

*******

For PC current flows, I use one of these, a DC clamp-on ammeter.
It's a non-contact instrument, so no circuit wires need to be
cut to make a measurement. It uses a Hall probe. No, you don't
need one of these, as they're too expensive. I like mine, and have
worked on my car with it (checked to see that starter motor
was drawing 150A DC peak). I also used to check the
compressor current on my central air conditioner with it
(as it also does AC).

https://web.archive.org/web/2009040...om/instruments/product.asp?catid=26&prodid=87

They make those in AC-only ones (no good) or AC/DC ones.
And the AC/DC is what I've got. The AC-only ones tend to
be cheaper. There's no reason for a Hall probe to be
expensive, so it's just the usual market pricing
(know the buyers can afford it).

That clamp-on meter, is how I know (roughly) the power
draw of the CPU, when idle or busy.

HTH,
Paul
 
A

Adam

Adam said:
Or, an extension wire from PWR to GND on a small breadboard? :)
Will that work?

Just keep PWR at PWR on small breadboard until a power ON short to GND is
needed.
No dangling wire to cause trouble.
 
P

Paul

Adam said:
Or, an extension wire from PWR to GND on a small breadboard? :)
Will that work?

Those signals aren't particularly sensitive, so if you
wanted to run a couple wires a short distance, things should
still work. The tricky part, is insulating the connection
which gets made to the header, so the pins aren't always
touching one another. I used shrink-wrap to prevent the
pins from touching.

Paul
 
Ad

Advertisements

M

mike

Those signals aren't particularly sensitive, so if you
wanted to run a couple wires a short distance, things should
still work. The tricky part, is insulating the connection
which gets made to the header, so the pins aren't always
touching one another. I used shrink-wrap to prevent the
pins from touching.

Paul
I'd be very careful about attaching wires to the motherboard.
If the humidity is 90%, you're likely ok. If it's 30% you're
gonna zap some chip input eventually.

I had occasion to troubleshoot a non-booting machine.
The wire from the power button went directly to the input
of the big glue chip that runs everything. And the signal
never came out of the chip to start the power supply.
Game over.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top