GA-P35-DS3P problems

A friend purchased a P35-DS3P mobo with two sticks of 1gb DDR2 RAM, a 3-pin
Coolermaster HSF, and a PCI-E video card to upgrade an older P4 478 system
in an ATX case. The PSU is a 480W Antec Truepower, and when everything is
connected the only response is a quick flicker of the CPU fan--no POST, no
PSU or case fans are spinning up, nothing. OTOH, when the CPU fan connector
(3-pin on the HSF, 4-pin on the mobo) is moved to a case-fan connector on
the mobo, it boots with no problem.

While this was connected, I connected the original Intel HSF (4-pin) to the
mobo's CPU-fan connector and although it still started fine the HSF
connected to the CPU-fan mobo connector (not attached to the CPU) would only
blip on every few seconds, never fully cranking up. I tried the same test
with an old 478 P4 HSF with a 3-pin connector and it would blip once, then
nothing. The 3-pin connector fits only one way with the guide on the mobo,
so I assumed the 3-pin-connector HSF's would work.

At this time it appears to function fine as long as the HSF is connected to
a case-fan connector on the mobo, but these seems make-shift to me. What am
I missing on this 4-pin CPU-fan connenctor issue?

Bob Davis wrote:
> A friend purchased a P35-DS3P mobo with two sticks of 1gb DDR2 RAM, a
> 3-pin Coolermaster HSF, and a PCI-E video card to upgrade an older P4
> 478 system in an ATX case. The PSU is a 480W Antec Truepower, and when
> everything is connected the only response is a quick flicker of the CPU
> fan--no POST, no PSU or case fans are spinning up, nothing. OTOH, when
> the CPU fan connector (3-pin on the HSF, 4-pin on the mobo) is moved to
> a case-fan connector on the mobo, it boots with no problem.
>
> While this was connected, I connected the original Intel HSF (4-pin) to
> the mobo's CPU-fan connector and although it still started fine the HSF
> connected to the CPU-fan mobo connector (not attached to the CPU) would
> only blip on every few seconds, never fully cranking up. I tried the
> same test with an old 478 P4 HSF with a 3-pin connector and it would
> blip once, then nothing. The 3-pin connector fits only one way with the
> guide on the mobo, so I assumed the 3-pin-connector HSF's would work.
>
> At this time it appears to function fine as long as the HSF is connected
> to a case-fan connector on the mobo, but these seems make-shift to me.
> What am I missing on this 4-pin CPU-fan connenctor issue?
>

I don't pretend to understand all the symptoms. It almost suggests
the motherboard has the ability to detect whether a four pin fan is
connected to the four pin header. Like they were sensing whether something
was connected to the PWM pin. Or maybe the 12V pin on the fan header,
has a traditional transistor connected to it for control (two
control methods, one digital, one analog) ? Maybe an analog
transistor for controlling the fan, is not getting fully
switched on (board defect) ?

Since you connected the original Intel four pin fan to the header, and
got some blips, that header may not be completely dead.

A test you can try, is go into the BIOS, while you're using the Intel
4 pin (and the real CPU cooler is connected to the case header), and check
the setting for "Smart Fan Control Method". The motherboard supports
a number of options, with an unclear explanation of what each one
does. I'd [Disable] that setting, since the description says "Forces
fan to run at full speed". That might tell you whether the header had
some basic function. See whether the fans behave properly after that.
(Always shutting down system power, before moving the fan connector - if the
fan connector is offset by one pin by accident, damage could result.)

The motherboard supports QST. The Southbridge has fan interface pins on it,
for driving PWM pins (pulse width modulation - a digital method for analog
control). Using PECI, the chipset can also access the CPU temp. And
implement automatic fan speed control. I believe people have
noticed the "blipping" behavior, most likely having something to do
with the microcontroller inside the chipset not being ready to control
the fan when the board starts up. Perhaps with the [Disable] setting,
you'll get a more traditional and reassuring response ?

The following is quoted from the ICH9 (Southbridge) datasheet:

"Intel Quiet System Technology (Intel QST)

The ICH9 integrates four fan speed sensors (four TACH signals) and 3 fan speed
controllers (three Pulse Width Modulator signals), which enables monitoring and
controlling up to four fans on the system. With the new implementation of the
singlewire Simple Serial Transport (SST) 1.0 bus and Platform Environmental
Control Interface (PECI), the ICH9 provides an easy way to connect to SST-based
thermal sensors and access the processor thermal data. In addition, coupled
with the new sophisticated fan speed control algorithms, Intel QST provides
effective thermal and acoustic management for the platform."

QST is part of a larger subsystem, fully implemented on some Intel "Executive"
motherboards. The P35 only has a fraction of the functions enabled. The Q35 enables
more than fan control. With a Q35 Northbridge, the motherboard can be
controlled remotely, without the main CPU running. The P35 doesn't allow that,
and the only function I'm aware of, is that fan control thing. The full set of
features in Q35 is called AMT, but just try and find documents that fully explain it...

(Example of an Intel motherboard with the stealth control features...)
http://www.intel.com/products/motherboard/DQ35MP/index.htm

This is the best I've been able to find on AMT. QST is just a small part
of it, and isn't even mentioned here.

http://download.intel.com/idf/us/docs/PS_ADTS003.pdf

Now, assuming the above tweak makes a difference, would I be happy about this ?
Consider if the BIOS ever gets reset (battery change, or another helpful person
resets the CMOS), then the user may be exposed to the same bad behavior when the
BIOS defaults are restored. So I don't know if I'd want to leave it in that state
(buggered, if the CMOS is reset). Your decision, whether it is worth trying an
RMA, as I'm not sure whether the symptoms are consistent with damage, or are
"normal" for this board.

HTH,
Paul