Re: Power-on problem with K8SE board

Frank Tarczynski wrote:
> I replaced a Tyan S2882D motherboard with a S2892 (K8SE) motherboard.
> The 8GB of RAM, power supply (FSP Group FX700-GLN), case (LIan-Li
> PC100), cards (LSI SAS adapter, Adaptec 29160, Sondblaster and GeForce
> 8600GT video) and matched Opteron 265s all worked fine with the old
> motherboard and were moved to the other.
>
> Nothing happens when I first press the case's power button to boot the
> system. I have to wait about 1 minute for the system to boot-up.
> Nothing happens, at all. After about 1 minute, the system starts or
> sometimes I see the fans spin for a few seconds and then the system will
> fire-up with the next power button push.
>
> The motherboard has a M3291 SMDC card and it's LED comes on as soon as
> the power supply is turned-on the motherboard appears to have power.
>
> I've tried jumpering the power-button pins on the mother board but get
> the same behavior.
>
> I've reseated the power cables on the motherboard and reseated all the
> card a couple of times.
>
> The board will shut-off OK when I press and hold the power button.
>
> The system runs fine once it boots, the BIOS reports all the right RAM
> and CPUs.
>
> Anyone got any ideas?

There are reviews for your power supply, here.

http://www.newegg.com/Product/Produc...82E16817104015

I'd probably get out a multimeter, and monitor the level on PS_ON#.
That is the control signal the motherboard uses, to turn on the
supply. First, see if the motherboard is actually attempting to
turn on the supply.

The sequence is

1) Power supply delivers +5VSB. Some motherboards have a LED which
is powered by that rail. That is the "standby" supply, and is
used to keep RAM powered, for computers in S3 Suspend To RAM state.

2) The logic to control PS_ON#, runs from +5VSB. It has to, because that
is the only source of power available at the time.

3) The front switch is conditioned by motherboard logic. That is how
the "delayed" reaction to shutdown via the front power switch
is handled. The four second filter is implemented in motherboard
logic. The front panel switch is a momentary contact - the motherboard
logic latches the momentary signal, and makes a steady level from it.

4) The motherboard drives PS_ON#, a signal on the main cable. The signal
is pulled up to the same voltage as +5VSB. A high level (more than
2.0V), is a logic 1 and means "keep the supply off". When the
motherboard grounds PS_ON# (drag the signal to below 0.8V), the power
supply is supposed to interpret that as a turn on request.

5) The power supply also has some kind of "power good" signal, it
sends back to the motherboard. Power good is used, along with
locally evaluated power good conditions, to determine when
the motherboard should be brought out of reset. Power good is meant
to help the hardware know, when all the voltages have reached their
full level.

6) But the thing is, if the fans start to spin, you know the
supply is on, and your symptoms right now are telling you
that the main portion of the supply is not on yet. That is
why a look at PS_ON# with a multimeter, should tell you a bit.
You can probe the backside of the cable connector, when it is
seated on the motherboard connector. I pick up ground from
the chassis for my black lead on the multimeter, and then only
need one hand to probe the main connector for a level on PS_ON#.
Press the front panel switch, and watch what happens to the voltage
on PS_ON#. It should respond instantly for a turn on request.

This is a simplified diagram of a power supply.

AC --- rear_switch --- rectifiers --- DC_high_voltage -+
|
+------- +5VSB regulator
|
+------- +3.3/+5/+12/-12 main regulator
^
|
PS_ON#

So +5VSB is needed to power the motherboard logic, that drives PS_ON#.
PS_ON#, pulled below 0.8V, causes the main part of the supply to
turn on. Then the PSU fan and the motherboard fans start to spin.
If you don't hear a beep, a possible reason at that point, could
be a lack of a Power_Good signal But there are many other reasons
a motherboard may not have executed any BIOS code, after the power
comes on. It is hardly ever a Power_Good problem.

If you need wiring harness color codes, for the various signals,
the Intel formfactors.org site has specs. The first document, is
the most modern one. The last one in the list, is from a time
when power supplies used to have a -5V output. Motherboards
have not relied on -5V for some time, but occasionally I see
people buy replacements supplies, and they cannot get an older
motherboard to run. Comparing these three documents, will give
you some idea of the evolution of the ATX spec.

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

http://www.formfactors.org/developer...X12V_1_3dg.pdf

http://web.archive.org/web/200304240...12V_PS_1_1.pdf

The ATX spec just handles desktop motherboards. The server stuff
has its own variations, so there are other specs that define
some of those. One of the differences there, is a 2x4 connector
for processor power, and there are at least a couple different
ways of wiring those (just to make life difficult).

Paul