Daniel Prince wrote:
> My brother just bought a Gigabyte GA-K8NSC-939 motherboard and a
> retail AMD 3700+ socket 939 CPU. He installed the motherboard into
> his old case using his old Antec 400 watt power supply, his old
> Crucial one gig DDR ram module (double sided) and a Matrox AGP video
> card (550?).
> The computer will not start. He says that nothing happens when he
> pushes the power button. The power supply fan does not even turn.
> The power supply does not have an on/off switch. Nothing happens
> when the power supply is plugged in. We do not have any spare power
> supplies. My brother said that after he installed the new
> motherboard, he connected the old motherboard to the power supply
> and the power supply did work.
> I did some testing of the power supply and I found that the 5 VOLT
> STANDBY has 4.49 volts on it when the power supply is plugged in.
> When I shorted between the PS_ON and a ground line, the CPU fan
> started spinning but when I removed the jumper, the fan stopped.
> (This is with the main connector disconnected from the motherboard
> but with the 12 volt connector connected.) Should the CPU fan have
> stopped after the jumper was removed?
> How should I go about testing the POWER GOOD line?
> The manual is not very clear on how the case connectors are
> connected. I think that the case power switch should go between the
> two pins with red colored bases. Is that correct?
> I was asleep when my brother connected the case power switch
> connector. If he had connected it incorrectly to one of the red
> connectors and to one of the green connectors, could that have
> damaged the motherboard? (The green connectors are for the reset
> If I understand the manual correctly, if just one DDR module is
> used, it should be connected to slot one or slot three but not to
> slot two or four. Is this correct? How can I tell which slots are
> which, the manual does not say?
> What should I do next to find out why this motherboard is not
> powering up? Thank you in advance for all replies.
From the manual:
X No Connect
Power _/ PW- X X RES+ \___ Reset switch
Switch \ PW+ X X RES- /
Power _/ MSG- X X HD- \___ IDE LED
LED \ MSG+ X X HD+ /
A typical LED drive, consists of the (+) side connecting to a resistor
limited supply. The (-) goes to an open collector pulling to GND. The
power switch would have a pullup to +5V on the (+), and GND on the (-).
Offsetting the connectors should not harm anything.
The 4.49V measurement for +5VSB is low. Probably not low enough to
prevent the thing from starting. (Stuff on that rail would be CMOS.)
The question is, does that rail drop even lower, when connected
to the motherboard.
The PS_ON# to COM connection at the power supply, only works for as
long as the jumper is in place between those two pins. One of the
jobs of the motherboard logic, is to condition the "momentary contact"
of the front panel power switch, to make the steady logic low on the
PS_ON# pin. Typically the motherboard will use open collector drive
to the PS_ON# pin. The power supply should have a pullup resistor to
+5VSB on its end. The power supply is off, when the voltage on PS_ON#
is close to +5V (logic one). The power supply comes on, when the
voltage on PS_ON# is pulled to 0 volts. The threshold between those
two states, could be in the 1.3V range or so, plus the thing may
have a bit of hysteresis.
You shouldn't really leave "half" the PSU connected, while fiddling
with PS_ON#. If you want to provide some load, remove the PSU entirely
from the case. Take a hard drive or two, and connect the hard
drive to the PSU. The hard drive I/O cabling should be disconnected.
So only the +5V and +12V are connected to the hard drive.
The hard drive will draw 5V@1A and 12V@0.5A and the spindle should
spin while the power supply is energised via PS_ON#.
The PWR_GOOD signal will be a logic level as well, but I don't know
what to expect for that one. You'd be best advised to go to
formfactors.org and download an ATX PSU specification, if you want
With revision E (E3, E4, E6) Athlon64 family processors, there is
no slot dependence for RAM. You can place one stick, in any of the
four slots. Processors previous to revision E (C0, CG etc), do
have a slot dependence. If using one stick, there are only two
slots that will work in. Previous to revision E, there were
two 64 bit data busses. The "Primary" bus works all the time.
The other bus only works in dual channel mode, and is the 128bit
extension of the primary bus. Thus, the other bus is not functional
by itself, but can only function as a "helper" for the primary bus.
If the processor is previous to revision E, then you'd be advised
to try all the slots, one at a time, for a response. Chances
are, your processor is a revision E, so the RAM will not be an
Some motherboard makers have crazy BIOS designs, where the BIOS
must be updated for each new Athlon64 processor that is introduced.
In some cases, using a single stick of RAM, will be enough to
enable booting into DOS and flashing the BIOS. But there is no
guarantee of that. If the processor you have, is some recent
revision (E6 say), it could be that the CPU support chart will
list a very recent BIOS to support it. http://www.gigabyte.com.tw/Support/M...ProductID=1881
But the thing is, you have be able to get past the "power
switch" step, to get the power to come on. Worrying about
RAM, processor revision, BIOS revision, is moot, if you cannot
get power to the board. Either the power supply, or the
motherboard is at fault. Starting the supply, with the
ATX12V 2x2 connected to the board, with a processor in
place, is not the smartest thing to do. I cannot predict
what would happen, if that is what you've tried. It would
have been safer, to connect just the bare motherboard
(no CPU), connect the 2x2 and the main power connector
to the motherboard, and test whether you can get the
PSU fan spinning that way. Don't get too creative with
your testing ideas, or you may fry something. Half-biasing
stuff on the board, is not necessarily very healthy, and
motherboards don't place infinite emphasis on protecting
the board against all possible power conditions. Users
fiddling with PS_ON# is not considered a normal operating