Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Posted on
- Abit LG-81 random shutdown/heat issue?
January 25, 2006, 3:41 am
rate this thread
I'm trying to build a new machine, and I've run into some issues
here. I grabbed an Abit LG-81 mainboard from an online retailer. I
attached an Intel 640 3.2 GHz chip, with a Thermaltake Golden Orb II on
it. I equipped this unit with 1GB of PQI Dual DDR Kit (5400) RAM, a
Western Digital 320GB hard drive, and recycled a Ben-Q DVD-RAM drive.
I've installed a MGE 400W Switching ATX power supply, with 2 fans on
Problem is, after it is all assembled, and setup, OS is installed,
and software (Office, Creative Suite, Various players, burning
programs, etc.) is all setup, the machine is exhibiting an issue. Mind
you, this didn't start until this machine had been running flawlessly
for more than 24 hours. After being booted, and running (mostly at
idle) for between 3 minutes and 2 hours, the machine will click off. It
does not shut down, it simply clicks off. At first I thought it was a
glitch, phenomenon. Then I began to try to reproduce it. Upon first try
with Prime95, I was able to make it click off. I initially ran the
torture test to see if it was a heat issue. I wanted a program to be
able to stress that. I loaded Everest Home Edition, and the FanEQ
program that came with the board. I was successful, in a way, to get
the machine to click off after about 15 minutes. Now when I say click
off, you can literally hear the power supply click, as it's going from
the on position to the off position. The fans all spin down, and leds
dim and fade away. Windows does not shutdown. Nothing ever gets logged
in the event logs. Seems to be hardware related.
I spent hours repeating this with mixed results. At one point I
enabled the fan alert noise, in the PC Health page in the bios setup. I
was actually able to make the machine sing (with this alert beep)
several times. During each of these tests, the machine appeared to be
normal. The fans would all be spinning, very well I might add. The CPU
temp would not exceed 38 degrees Celsius! I continued this testing, and
seemed to isolate this as a heat issue. Even while testing, when
periodically the temps would read 117 Degrees for a second or two, the
chip was never that hot. I could touch the bottom of the heatsink
without it being that hot. Not burning. NOT 117C. This seemed to
isolate a heat sensor issue, potentially. Even though these high temps
were observed, It was only twice, and they NEVER showed over 38 degrees
Celsius. I've sat and watched several times, observing temps realtime,
and they don't spike, or show above this range at all. The temps slowly
climb, then stop @ 38 when the fan's cooling reaches a temperature
I tried to isolate this some more. I went back to the main
symptom. The switching power supply is being told to shut off. I wanted
to test this, just in case this was software related. It appeared that
even if this machine was at idle, it would turn off in an hour or two.
I proceeded to download memtest86+ from www.memtest.org. I booted the
machine from this disc, and allowed this to run for 26 hours. It made
almost 200 passes on this 1GB of ram (in two sticks) and found 0
Now, that's not the power supply. But it does show that the
machine can run for 26 hours outside of WindowsXP. I then decided to
try another PSU. I grabbed a 500W Antec Neo HE PSU, and installed this
in the case. I booted the machine, loaded Windows & Prime95, started
testing, and within 10 minutes the machine clicked off. I confirmed
this a few times. It's not an issue of the machine not getting enough
power. It's stable, and the voltages on both PSU's are reading
identical. Both in Windows, and on the Multimeter's screen!
Just for kicks, and out of frusteration, I rebooted, and allowed Xp to
load. I walked away. 24 hours later, still @ idle, the OS remains, and
the machine is still powered up. I've tried applications testing, and
mock "working" on the computer, and the error does not occur. In fact,
this error occured first @ idle.
So, now I'm left pulling my hair out. I'm left with no other
option as to why this is happening. I've disabled all warning/shutting
down for fan failure/sensor temperature. I've experimented with this,
and I've been unable to isolate this further.
BTW, latest BIOS (01-16-2006) and all of the WindowsXP updates are
installed. Please help me Abit community. :)
Re: Abit LG-81 random shutdown/heat issue?
Well the really obvious and easy thing to do is test the processor in
another MB but thats a problem if you dont have access to another MB.
Thats why Im so glad to have 3 systems now which I thought was insane
a few years ago. Who needs more than one? But after having so many
problems every few months if you dont have access to another system to
swap parts ----- even if its a cheap problem to fix until you find out
what it is you left with two really bad choices --- pay an arm and a
leg at a repair shop or pay an arm and leg constantly buying more and
more new replacement parts swapping them out often buying the wrong
Of course you can avoid all that if someone happens to know the exact
solution to your problem cause theyve experienced themselves but the
odds are probably slim for that.
Whats that 117 C bit about? You kept posting about the CPU staying at
38C but then you mentioned something 117. Obviously if it went up to
117 then thats the problem though you say it was cool to the touch.
Re: Abit LG-81 random shutdown/heat issue?
The description does sound like THERMTRIP.
Get a copy of Throttlewatch here. Intel processors should start to
throttle back CPU activity if they think they are starting to
overheat. THERMTRIP is set to 20C higher than the throttling
temperature. The sensor used is different than the diode that the
Winbond monitor chip uses, but the diode used by the internal temperature
monitor and the diode, should track quite well. In a sense,
observing throttling is an independent way to measure silicon
die temperature, as the processor is determining for itself,
that it is too hot.
I'd pull the CPU HSF and have a look at the thermal paste pattern,
and see if there is a thin uniform layer and no air voids. I would
also examine the install and see if there are signs of excessive
forces in the socket area, or see if the heatsink is not sitting
flat for some reason.
Occasionally, you run into cases of people who have processors
that just seem to run too hot internally for the level of cooling
provided. Just as the user puts thermal paste between the CPU and
the heatsink, Intel also puts their goop between the integrated
heat spreader (IHS) and the silicon die. While there are adventurous
people who have stripped the IHS from dual core Athlon64, to
discover a less than adequate thermal contact inside, I'm not
aware of anyone doing similar research on Intel products.
The two things I'd check for, as you've already done, is signs
that there is an overheat problem, that is hot enough to trigger
THERMTRIP. The second thing I'd look for, is a brass standoff
underneath the motherboard, that shouldn't be there. Brass standoffs
should line up with plated holes, and not touch any copper power
tracks underneath the motherboard.
That is a nice looking design, BTW.
I'm really surprised that memtest86 didn't trip it as well.