ASUS P6T7 WS-Supercomputer Motherboard Boot Issue fron Power Off/Sleep

[pcpilot]

Hi All,

This issue is bugging me very much now as the only workaround is, once
running, to leave powered up 24/7....very costly and very energy
inefficient.

To explain more the issue developed suddenly after 3 - 4 months of use
fron 18th Jan 2010. Subsequently worked with Asus via email carrying out
all suggestions and eventually RMA'd late 2011.

Received replacement board (later version with 4 pin Molex Power Socket
on Motherboard) thought perhaps later revision might have solved issue.

Within days same problem F.F. GP diagnosis Error with/without memory,
expansion cards etc. unusable again for many hours (at best) & ONLY THEN
with mains power disconnected from PSU. Subsequent reconnection hit &
miss whether normal boot or the F.F. fail.

I have tried removing BIOS battery resetting CMOS neither have any
effect ONLY removing mains power for many hours seems to work. Hence
once running (as now) I WILL NOT switch off!!

Reveiwing the issue on the web came across this thread on the ASUS
Forums with virtually identical issue:
http://vip.asus.com/forum/view.aspx?id=20100113153305484&board_id=1&model=P6T6+WS+Revolution&page=1&SLanguage=en-us

The poster's setup was:

Intel Core i7 920 D0
ASUS P6T6 WS Revolution
OCZ 6GB (3x2GB) DDR3 1866MHz/PC3-15000 Reaper HPC Triple Channel Memory
Kit CL9 1.65V - x2 to make 12GB total.
Be Quiet 750W Dark Power Pro
Antec P182 case
Vista Home Premium 64bit
2 x Western Digital VelociRaptor 150GB
2 x WD 1TB Hard Drive (RAID)
Noctua NH-U12P fan
Sapphire HD 4870 512MB GDDR5 graphics card

My "Broadly Similar" setup is:

Intel Core i7 920 D0
ASUS P6T7 WS Supercomputer - BIOS 1001
Corsair Dominator (TR3X6G1866C9DF) (3 X 2G) PC3 15000 DDR3 1866 MHz
Ultra-Low Latency 9-9-9-24
Triple Channel Memory Kit to make 12GB total.
Be Quiet 750W Dark Power Pro
Antec P193 case
Windows 7 Professional 64bit
ASUS U3S6 USB3/SATA III Controller Card
2 X Western Digital Caviar Green 2 TB SATA 6 Gb/s 5400 rpm 64MB Cache -
WD20EARX-00PASB0 Firmware 51.0
2 X Western Digital Caviar Black 750GB SATA 3 Gb/s 7200 rpm 32MB Cache -
WD7501AALS-00J7B1
*Neither Pair Presently Configured RAID*
Coolermaster Fenrir Titan modded with Akasa AK-FN057 Apache 120 mm Fan
(CPU Cooling)
XFX Geforce GTX285 XXX - 670MB/1GB DDR3/Dual DVI TV (PCI Express) -
GX-285N-ZDD - Graphics Card
1 X LG GH22LS50 22 X Super Multi DVD Re-Writer SATA Drive - DVD Writer
1 X LG BH08LS20 8 X (Blue Ray) Super Multi Blue Ray Re-Writer SATA Drive
- BlueRay Writer
Keyboard/Mouse Logitech diNovo Media Desktop Laser Bluetooth
(USB/PS2/Bluetooth (MX-1000 Bluetooth Laser Mouse))

I have covered ALL of the suggestions noted in the above linked post and
as per the poster no joy either!!

Can any of the expert users offer any suggestions, especially in view of
the close similarity of the selected components.

Any thoughts much appreciated.

....a VERY hopeful

PC Pilot

[Paul]

Motherboards contain power sequencing logic. This includes functions
such as "backfeed cut", to prevent sneak paths for power to flow,
which can upset multi-rail chips.

It could be, that some assumption about voltage level on some node
is wrong, and the motherboard doesn't think it's turned off! By
waiting for it to drain, some analog power/control circuit
finally gets into the correct state, and then everything can work.

Years ago, in one of these cases, the problem was traced down to
leakage current flowing from the monitor, back into the motherboard.
A user noticed his computer would not start. If he unplugged
the monitor for a moment, pushed the button, then plugged in
the monitor again, the computer would then start.

I doubt that is the source of your problem. I mention that to illustrate
just how sensitive some of those analog control circuits can be.
It's annoying when one of them does that.

When an RMAed motherboard does this, that means it is likely a
design error. Or, your particular peripherals, are somehow
pumping current back into the motherboard. I choose to believe
this is a design error, as I expect after the "monitor prevents
startup issue", that was fixed in subsequent designs.

*******

To test the "peripherals theory", strip down to a minimum config, and
do some startup tests. If it starts properly, then at least
you have some idea what direction to look in. If it still
fouls up, then you'd suspect the motherboard design itself.

The minimum config in this case would be motherboard + CPU + CPU_fan.
No need for RAM, video card, storage cards, or even keyboard or
mouse if you want. Since you have a Port 80 diagnostic display
to use, you're looking for any code other than "F.F." to prove
the CPU is running. The CPU will barf, when it finds no RAM or
video, which is fine, and some other diagnostic code will show
on the display. But if any code other than "F.F." shows, then
you know "it started".

Motherboards don't need any plug-in components to work. If you
need to test the front panel "power" push button on your computer,
you can do that with no CPU installed. You should be able to
turn the ATX power supply on and off that way. But in your case,
you want confirmation the CPU VCore started, and all chipset
regulators powered up properly (i.e. testing sequence logic
if present). By having the CPU running, and having the
CPU attempt to load BIOS ROM code, then that's proof the
motherboard is (mostly) working. Motherboard + CPU and a
"non FF" code, means CPU, Northbridge, and Southbridge are
working. And for those to work, most of the onboard regulators
would be working as well. (Just a few "private" ones, such
as the clean regulator for the sound chip, multi rail regulator
for Ethernet chip, those might not be proved to be working by
this test case.)

Have fun,
Paul

[pcpilot]

....Hi again Paul...

..doubt this is significant but just in case will mention anyhow....

At around the same time (but not sure exactly) as the boot F.F. problem
first surfaced I lost use of an old basic PS2 MS Intellimouse mouse
which had been hitherto working without a hitch as a reliable non
conflicting second (always connected) mouse to the Logitech MX1000
Bluetooth Wireless Laser Mouse I always use.

Whilst keyboards function fine off this PS2 port, USB mice (I only have
the Intellimouse as a genuine PS2 mouse) connected using a USB/PS2
adapter are also not recognised (just like the PS2 Inellimouse) when
using this port.

By the way, the 'old reliable' Intellimouse works (currently on a
borrowed XP system) happily on ANY other machine, maybe just
coincidence?

Secondly, and again somewhat bizarely Windows 7 reports some of my
devices (connected to the 'onboard' USB Ports) as being connected to a
USB 1.1 Port/Hub and may work faster if connected to a 2.0
port.......the board is of course a USB 2.0 ported board.......any
thoughts?

Thanks,

PC Pilot

[pcpilot]

Hi Paul,

Many, many thanks for your very informative reply very much
appreciated!! Sorry for not replying sooner but I have been tinkering
with the setup and been unable to pass the F.F. fail state for several
days as the restarts have been automatically introducing a momentory
power off state (for some reason) thus re setting me to F.F. and a
further 12 hour wait. Finally bit the bullet and pulled the BIOS battery
and reset CMOS to get an uninterupted power up but did then have to
chance adjusting the BIOS settings from the default...fortunately to a
further uninterupted restart and bingo back in business!!

Well, where to start...........

I wonder if I might seek your thoughts in respect of the 'particular
peripherals' aspect of this issue. You will I am sure have noticed that
the guy with the same issue on the P6T6 board was also using both an
Antec Case (P182) ~ mine is the P193 and the identical
make/model/wattage of PSU (Be Quiet Dark Power Pro 750W). Would these
count as 'peripherals' in this regard and so be potential 'leakage'
sources? or by their very nature should I discount these looking only
for connected devices, cards, monitors, usb devices et al?

By way of brief additional background I had previously upgraded the PSU
from 600W to the present 750W to eliminate that as being the issue
(suggested by ASUS I think) though unfortunately the upgrade was also
from a Dark Power Pro (600W model).

The F.F. fail state does report right back to Motherboard/CPU almost
barebones only state. However whilst all cards, SATA's, Memory etc. are
removed the board is still powered by the PSU (ATX and Aux Power plugs)
and still connected to the on board USB/Firewire headers plus system/cpu
fans as well as the audio and front panel (for Power On/Off switch)
headers. In this state only removal/refitting of the CPU (which I do not
like to do too frequently) resets the BIOS Boot state, hence why ASUS
called the RMA.

I will (at next fail cycle.....just glad to be up and running again
right now and catching up on my work!!) take the board to ATX plus Aux
Power, CPU Fan and CPU only state and see what surprises that brings!!

Might you offer any suggestions of how I can isolate the leakage in
question, am I looking at a particular power rail or CPU voltage level??
or is it just an unscientific unplug all devices (as noted above)
approach? Given that seemingly the majority of these P6T series boards
(and the P6T7 in particular) power up/down normally is it simply the
combination of peripherals here that is pointing to a 'design issue'?

Assuming that to be the case is there any one 'likely' candidate you
might look to change first (e.g. Case, PSU etc.) here??

Sorry for the twenty questions but I am desperate to solve this as the
board is no issue once running and given its expansion possibilities
should (as intended) future proof me for a good few years yet!

Your help/advice is really appreciated,

........a very grateful

PC Pilot

[Paul]

I notice here, that a few of the reviews aren't kind to this board.
(You can also check the vip.asus.com forums, which will likely have
more comments.)

http://www.newegg.com/Product/Product.aspx?Item=N82E16813131390

I see you've already been over here. Make sure you browse all the
threads, to see if there are matching symptoms somewhere.

http://vip.asus.com/forum/topic.aspx?board_id=1&model=P6T7+WS+SuperComputer&SLanguage=en-us

*******

You're assuming this can be fixed by magic :-)

You may not have noticed it, but motherboards are actually pretty complicated.
It's a miracle they ever start. That's the strange part, that so many of them
work. Not that some of them don't.

1) Your symptoms are, a time dependent behavior where the CPU is
not able to execute any code (thus "F.F." on Port 80 display).
2) Your system "works" eventually, after enough manipulation.
That means there isn't a guaranteed static fault in the thing,
like a broken wire, a bad connection in the CPU socket. I'm not
comprehending here, why removing the CPU helps. LGA1366 do occasionally
have problems with sockets, like a memory not being recognized, because
all the CPU contacts aren't being made. LGA1366 also apparently
can have problems, which are voltage setting sensitive. (Memory not
completely recognized until something like Vnb is adjusted ??? Don't remember.).
I can't keep track of all this stuff - someone who owns an LGA1366
should be researching the weirdness for themselves. That's as much
weirdness as I've heard of.

The thing is, even if all memory is not present, or if memory is not
detected, the Port 80 display won't stay at "F.F.". It will advance
to another code, telling us firmware is being read from the BIOS
EEPROM. Completely missing memory (like pulling out all the DIMMs),
should not be able to prevent the display from advancing past "F.F.".

Well, what can hold the display at "F.F." ?

If the reset signal to the CPU socket remains asserted ("held in reset"),
then the CPU cannot fetch any instructions, and there is no chance
of the Port 80 display going off the "F.F." value.

Reset, is a "critical node" in the schematic. In that, it checks
for a number of conditions, before "releasing" and allowing the motherboard
to start. It checks for "Power Good" from more than one source for example.
(One of the Power Good signals, comes from the ATX supply, over the main ATX
cable. Other Power Good signals, can come from things like the VCore regulator.)

There have been cases before, where the VCore overcurrent protection, is
triggered by the power supply being slow at delivering +12V. Then, the VCore
supply to the CPU is not available, and obviously, no instructions can run
if there's no power. To clear the VCore overcurrent, the power supply
must be switched off at the back. (That fault condition, is "latched" for
safety reasons.) The affected people in that case, were using certain
models of Antec power supplies, mixed with certain models of Asus
motherboards. That condition was relatively reproducible, in that
you might *never* get those combinations of components to start (so
"draining" doesn't help).

The reset circuit may even have inputs, which check for certain rails
"going away" at some point, but the effect might not have been intentional.
It would be a leakage of some sort, or a condition existing, where the
reset circuit doesn't know the motherboard has been power cycled, where
the reset circuit won't "release reset".

Your symptoms are not suggestive, that yet another power supply
will fix it. If you do decide to (temporarily) substitute another
brand of power supply, make sure it's an entirely different brand
of supply. For example, my current computer uses an Enermax, and
if I needed a temporary substitute, I have a "Sparkle" brand ATX I
could use (both Enermax and Sparkle make their own supplies, so
they're unlikely to have the same power-up timing).

My suggestion to remove components and simplify, was in the hope it
was some kind of leakage problem. But there are other problems
which can affect the critical "reset node" in the schematic.
And the way "reset" behaves, is there for a reason. It's to
coordinate things on the motherboard, so the startup of the
motherboard is more consistent (starts properly every time).
That's why so many gated things feed that circuit.

*******

The USB hub interface, uses resistors for some announcements of
speed. It's up to the peripheral to send a "JK chirp" to the hub,
for USB2 speeds to be recognized and used. During startup,
the USB peripheral and hub are in effect "negotiating" the
speed of operation.

http://www.beyondlogic.org/usbnutshell/usb2.shtml

If there's no USB2 EHCI ("Enhanced") driver, then that can stop
USB2 operation. So even if the peripheral sends JK chirp, if
the USB2 block has no driver, there's nothing it can do with the
info it got. With no driver, there is no transfer of state info
to the OS.

If some voltage is out of spec, or if the Southbridge is
having troubles with recognizing USB2 signals, that could
also account for it. But if that was the case, you might
also have trouble getting hard drives on the Southbridge
SATA interfaces to work. So if it was a Southbridge
powering issue, there might be more affected interfaces
than just a USB anomaly.

Paul