xeon 5160 3 GHz and gigabyte GA-7VCSV-RH board problems


D

derf109

Here's what I have:

Xeon 5160 3.0 GHZ CPU (just one)
Gigabyte GA-7VCSV-RH board
2 GB Kingston RAM
PC Power Cooling 'Silencer 610' power supply

Power into the board with CPU and memory (no PCI cards) via connecting
both
24-pin cable and aux power cable: nothing.

Unplug 24-pin cable and plug it into a power supply tester: all lights
come on fine.
Leave aux cable in, keep 24-pin cable in tester: system turns on (CPU
fan spins,
no video though).

I've replaced the board once. Can't be a bad board.
PC Power Cooling people tell me PS is OK, tester verifies this.

What else could the problem be?
Do I HAVE to put 2 Xeon 5160s in the board? Currently I'm using only
one (yes,
it's in the right socket and in the correct orientation).
Or is this 5160 not compatible with the board?
Gigabyte said it was.
Or is it just a bad CPU? How can I test THAT without getting the
whole system to work
to begin with?

Thanks in advance
 
Ad

Advertisements

A

Arno Wagner

In said:
Here's what I have:
Xeon 5160 3.0 GHZ CPU (just one)
Gigabyte GA-7VCSV-RH board
2 GB Kingston RAM
PC Power Cooling 'Silencer 610' power supply
Power into the board with CPU and memory (no PCI cards) via connecting
both
24-pin cable and aux power cable: nothing.
Unplug 24-pin cable and plug it into a power supply tester: all lights
come on fine.
Leave aux cable in, keep 24-pin cable in tester: system turns on (CPU
fan spins,
no video though).
I've replaced the board once. Can't be a bad board.
PC Power Cooling people tell me PS is OK, tester verifies this.
What else could the problem be?
Do I HAVE to put 2 Xeon 5160s in the board? Currently I'm using only
one (yes,
it's in the right socket and in the correct orientation).
Or is this 5160 not compatible with the board?
Gigabyte said it was.
Or is it just a bad CPU? How can I test THAT without getting the
whole system to work
to begin with?

An LED power tester does not tell you enough. For example +5VSB
cpule be too low. Had this recently with a PSU, were +5VSB
only had 4.1V, not enough to turn on the mainboard, but
probably enough fpr a cheap PSU tester, since it will likely
not test this voltage in a meaningful way.

Plug in both connectors to the maonboard and then measure +5VSB
(should be violet). Also, wothout swithcin the board on,
it will likely not do anything. Second step is to connect the
power switch and see whether you can turn it on that way.

Arno
 
P

Paul

Here's what I have:

Xeon 5160 3.0 GHZ CPU (just one)
Gigabyte GA-7VCSV-RH board
2 GB Kingston RAM
PC Power Cooling 'Silencer 610' power supply

Power into the board with CPU and memory (no PCI cards) via connecting
both
24-pin cable and aux power cable: nothing.

Unplug 24-pin cable and plug it into a power supply tester: all lights
come on fine.
Leave aux cable in, keep 24-pin cable in tester: system turns on (CPU
fan spins,
no video though).

I've replaced the board once. Can't be a bad board.
PC Power Cooling people tell me PS is OK, tester verifies this.

What else could the problem be?
Do I HAVE to put 2 Xeon 5160s in the board? Currently I'm using only
one (yes,
it's in the right socket and in the correct orientation).
Or is this 5160 not compatible with the board?
Gigabyte said it was.
Or is it just a bad CPU? How can I test THAT without getting the
whole system to work
to begin with?

Thanks in advance

I see three power connectors. The manual only mentions two power connectors.
There is a 2x2, a 2x4 (dual proc power), and the main 24 pin.
The 2x2 looks to be near the 24 pin main connector. The pinout on the
2x4 power connector, appears to match the user manual.

http://www.it.com.cn/f/server/067/17/20060717_srv_mb002a.jpg

The Northbridge is a dual interface chip, meaning two processors are
not needed for continuity. The BIOS may have some preference for which
socket is occupied, but the hardware doesn't care.

You should be able to connect the three power cables, connect a computer
case power switch to the panel header, and turn the motherboard on
even if there is no processor or RAM present. The power switching
circuitry should still work. If you do the experiment with the
processor in place, heatsink/fan plugged in, you should be able to
get beep tones.

The computer case power switch goes to pins 11 and 13 on the F_PANEL
header. Pin 3 is the "key" or missing pin, so that tells you which
end to count from: 1,(3),5,7,9,11,13...

There appears to be a piezo speaker, next to the CMOS battery. Again,
it is not mentioned in the manual. I don't see speaker pins on the
main F_PANEL header. So I assume there is a piezo, and the beep tone
for missing RAM would come from there. You won't get a beep, if
the processor is missing. You should get a beep pattern if the
processor is present and the RAM is missing. There is integrated
video, so missing a video card isn't going to do anything.

HTH,
Paul
 
D

derf109

I see three power connectors. The manual only mentions two power connectors.
There is a 2x2, a 2x4 (dual proc power), and the main 24 pin.
The 2x2 looks to be near the 24 pin main connector. The pinout on the
2x4 power connector, appears to match the user manual.

http://www.it.com.cn/f/server/067/17/20060717_srv_mb002a.jpg
You should be able to connect the three power cables, c


First of all, thanks for contributing to this discussion. I
appreciate the help
and input.

I get nothing when plugging in. No noise, nothing turns on.
With or without a CPU present, nothing happens. And this is the 2nd
identical board they've sent me so I'm reluctant to blame that.

And there are only 2 power connectors, despite what that JPG
showed. The one near the 24-pin connector is not present.
 
Ad

Advertisements

P

Paul

First of all, thanks for contributing to this discussion. I
appreciate the help
and input.

I get nothing when plugging in. No noise, nothing turns on.
With or without a CPU present, nothing happens. And this is the 2nd
identical board they've sent me so I'm reluctant to blame that.

And there are only 2 power connectors, despite what that JPG
showed. The one near the 24-pin connector is not present.

OK. You can do a bit more debugging, if you have a multimeter.

I guess my first question would be, when you pressed the computer
case power switch, did the fan blades "twitch" ? If that happens,
it means the power supply turned on, only for an instant. And that
tells you to beware of an overload coming from the motherboard or
other components.

My second question, would be about the CPU coolers. To do what you
are doing, I expect the motherboard is installed in a case, the
heatsink is bolted to the case through the holes in the motherboard.
If that is so, then removing the motherboard and sitting
it on a cardboard box, is not going to be an option. To do that,
you'd need a processor cooler that did not need the case as a source
of mechanical support. I prefer a piece of cardboard as a place to
sit the motherboard at this point, as a means of eliminating any
short circuits between motherboard standoffs and stuff on the
bottom of the motherboard.

With the power supply connected to the case (the 2x4 and the 24 pin),
switch on the power supply at the back. The power supply should
immediately deliver +5VSB. If you look in the manual, the pinout of
the 24 pin connector is defined. +5VSB is on pin 9. If you have a
multimeter, and an alligator clip that slips over one of the probes,
clip the black (-) lead to a screw on one of the I/O connectors at
the back of the computer case. That allows just handling the red (+)
lead to make some voltage measurements. The metal of each pin should
be accessible from the back of the connector, so you can probe the
24 pin while it is plugged into place. The metal of each pin should be
exposed from the holes in the nylon shell. Set the meter on 20VDC full
scale, suitable for checking all the rails.

So, with the power supply switched on, the first check is for +5VSB.
It should be 5V +/- 5% or so.

PS_ON# is on pin 16. At this point in time, PS_ON# should be 5V as well.
It should be at roughly the same voltage value as the 5VSB pin is. That
means the motherboard has not tried to pull it low.

Now, connect the computer case power button to pin 11 and 13 on the
F_PANEL connector. When you press the computer case power button, the
contacts of the switch momentarily make contact. That means pin 11 is
shorted to pin 13, but only for half a second or so. The motherboard
is supposed to latch that pulse, and make a steady level from it.

Now, what should happen, when you push the button, is PS_ON# should change
states from 5V, to a DC voltage level closer to 0 volts. Somewhere
between 0.4V and 0.8V might be a good level on the signal. When the
signal goes low, the motherboard is trying to turn on the supply.
I expect this is where things are fouling up.

When the power supply detects a low voltage level, it turns on. The fan
should start to spin. For the first 35 milliseconds or so, the power
supply will ignore a short circuit. After 35 milliseconds, the POK
(power OK) will likely change state. Now the power supply can start
examining the overload situation. If the supply was overloaded, the
power supply could switch off, and stay off. You would have to
flip the switch on the back of the supply, to try again (recycle it),
if that happens (overload detected).

From a debugging perspective, you've practically won if you get the
fans to spin. Because now you've got a good quality power supply,
connected to a server board, and that board was undoubtedly tested
better than your average $60 desktop motherboard.

If the supply does not start, looking at PS_ON# may help. If PS_ON#
did not change states, then something stopped the motherboard from
doing what it was supposed to do. Some motherboards have interlocks
(Asus used to examine the AGP card for example), and an interlock
can be used to stop the power sequence, if a dangerous situation
exists. On the old desktop boards, that would be the detected
presence of a 3.3V only video card, in a 1.5V only AGP slot. That
would stop PS_ON# from going active. You can ask Tech Support for
their recommendation as to why PS_ON# is not asserting, when
pin 11 is shorted to pin 13 of the F_PANEL header. (If you suspect
a bad computer case power switch, you can *carefully* wedge a
screwdriver blade between pin 11 and 13, simulating a momentary
contact. Make sure you have the correct two pins, before trying
that.)

Since the manual from Gigabyte is a real plonker, I don't think I
can spot much else to go on. With no high resolution picture, I
cannot even tell if there is a green LED on the motherboard
surface, that is connected to +5VSB. Many desktop boards have
that LED, and the LED is a handy indicator that +5VSB is present.
The LED is doubly useful, if you see it "wink out" for a second
when you press the computer power on button. +5VSB should stay
at a steady level, as long as the power supply is on. If the
voltage blinks or winks, again, something is overloading +5VSB.
The power supply should have a high enough +5VSB rating, to
meet whatever +5VSB load is mentioned in the manual.

While you're at it, touch your (+) meter probe to the top of the
CMOS coin cell battery. It should read 3.0V or slightly more,
if the battery is fresh. If it is less than 2.4V, change it.

Hope that helps,
Paul
 

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