Cooling Problems

  • Thread starter Thread starter Ignacy Sawicki
  • Start date Start date
I

Ignacy Sawicki

Hello,

Having a bit of a problem with my machine. I think that it is heat
related, but I do not want to prejudice any of you.

My hope is that changing the cooling for the CPU will do enough, but I am
not sure if the amount of problems I am seeing is not pointing to a much
more serious problem (a dying motherboard, for example).

Please let me know what you think.

Iggy
------------------------------------------------------------

Specs:
------
Athlon XP 2000+
AMD heatsink and fan
ASUS A7V8X motherboard (v14 BIOS)
PATA Drive: old Maxtor 53073U6 (30 Gb)
SATA Drive: new Barracude 7200.7 (120 Gb)
Sapphire ATI Radeon 9600 Pro video card
Win XP Pro

Hiistory:
---------
Been running my machine with the SATA as the main drive since January
2004. One day, it fails to boot: checking in another system confirms that
the drive (controller, most likely) is dead. Rebuilt system on old PATA
drive, while waiting for retailer to send a replacement. Up to this point
no problems. The ambient temperatures were hovering in the low 20s
Celsius (that's a bit above 70 F for the Americans...).

A few days later, I discover that I would have problems with rebooting
after using the computer for a few hours. It would reset, the video would
never turn on, but it would proceed to boot. If the computer was not very
hot, the video would appear on the blue WinXP welcome screen, and be OK
after that. Sometimes, even though the computer was reasonably cold, it
would boot OK, but I would not be able to view DVDs: the Windows desktop
would be OK, but the contents of the PowerDVD window would be completely
scrambled. At this point, I thought that my video card was dying.

Situation today:
----------------
Ambient temperatures: 30 Celsius (90 F).

Computer only works OK with case open (i.e. mobo temp is reported as 32
C/89 F) and CPU speed set at 1250 MHz (nominal is 1667 MHz). CPU temp
right now is 49C/120F.

If I try to close the case, the temperature begins rising quite rapidly
until about 57C for the CPU. However, pretty soon after it is closed, the
computer begins to seriously slow down, until it completely locks up.
Removing the case restores some functionality, but the machine needs to
be rebooted to restore all of it.

The system works nearly normally, apart from the HDD access (no video
problems whatsoever). The PATA drive speed is reported as about 1Mb/s by
DiskBench, which seems ridiculously slow. I get the feeling that the
longer the machine is on, the slower it gets, but it might be just an
issue with using more virtual RAM when the computer has been on for
longer. Then again, the boot time is reasonably normal. Just as an
example, Firefox can take about 40-50 seconds to open. Sometimes less.

If I try to install the replacement SATA drive, the system slows down
TERRIBLY (seems to get hung every time the drive is accessed for 30
seconds or so). It was OK for the first few boots (allowed me to clone my
system partition), but booting off the SATA drive is basically impossible
now. Sometimes I get through to WinXP, in which case it is extrememly
slow; most of the time, it gets hung early on in the WinXP boot screen,
sometimes it just resets itself during the boot process.

If I try to run the computer at the full speed, it does not boot anymore,
so I can't let you know what the CPU temp is then.
 
What is the condition of your fans? also have you
cleaned any lint built up on you fans and heatsink?
Also check you video card to make sure its fan is
working and clean. And are your case fans working
and clean? Also if possible remove the front of your
case to check for any lint build up. check any place air
moves in or out of you case.

Jim M
 
It sounds like you may be using an underpowered/unstable power supply unit.
I would try replacing the PSU with a HIGH quality adequate wattage model and
see if that doesn't fix the problems.
 
DaveW said:
It sounds like you may be using an underpowered/unstable power supply
unit. I would try replacing the PSU with a HIGH quality adequate
wattage model and see if that doesn't fix the problems.
Click to expand...

Would such a thing show up during hot weather? I changed to PSU (to a 350 W
one) a year ago and have had no problems since.

ASUS Probe reports the following voltages:
12 12.192
5 4.95
3.3 3.072
VCore 1.854

I guess that the 3.3 looks a bit low. Also this is running the system
underclocked, so who knows what will happen otherwise.

Iggy
 
Ignacy said:
Would such a thing show up during hot weather? I changed to PSU (to a 350 W
one) a year ago and have had no problems since.

ASUS Probe reports the following voltages:
12 12.192
5 4.95
3.3 3.072
VCore 1.854

I guess that the 3.3 looks a bit low. Also this is running the system
underclocked, so who knows what will happen otherwise.

Iggy
Click to expand...

There are two things wrong there. The 3.3V is low by almost 10%,
while the spec is 5%. Your Vcore is also high. Depending on the
model, the CPU takes 1.60V or 1.65V.

http://www.qdi.nl/support/CPUQDISocketA.htm

Static tolerance is 0.05V on Vcore. Max Vcore is 0.5V higher,
as in 1.60 + 0.05 + 0.5 = Absolute max. Your 1.85V value is
safely under the 2.15V allowed, but 1.85V is also higher than
the static tolerance of 1.60 + 0.05.

Take a look at the OVER_VOLT1 jumper, near the top of the board.
Position 2-3 (rightmost) is supposed to disable OVER_VOLT1 and
the manual says it is the default position for the jumper when
the motherboard leaves the factory.

If you were overclocking, then I could understand you wanting
to use OVER_VOLT1, but it is really for people trying extreme
overclocks.

Now, the second problem is, how to interpret the 3.072V from
the power supply. First of all, video cards do not tolerate
low 3.3V that well. In fact, there are some video cards that
like the voltage to be a hair higher than 3.3V (like AIW type
cards).

Either the video card is pulling the power supply down, or
something else is. The 3.3V isn't typically used for
motherboard power conversion, but it might be used for some
PCI peripheral chips. +12V or +5V can be used for switching
conversion to lower voltages. As this is an Athlon board,
+5V is likely to be used to make Vcore, leaving +12V lightly
loaded. It is possible to use a linear regulator off 3.3V,
to make the 2.5V for the DIMMs, but that is a fairly light
load. So, I would start pulling any PCI/AGP cards, one at a
time, while measuring the output voltage of the power
supply with a voltmeter.

To measure voltages on the PSU, some supplies have a six pin
AUX connector, and that has a couple of +3.3V pins on it.
The +5V and +12V can be measured on an unused disk drive
power cable.

So, try stripping components until the measured value of
+3.3V pops back up. The board will probably beep like
crazy when the video card is missing, but that isn't going
to hurt anything while you measure the power supply. Be
careful while using the voltmeter, not to short the leads
together, as big sparks and failures can occur. I recommend
connecting the black (ground) lead, to a lug on a connector
on the I/O panel on the back of the computer, keeping
one lead of the voltmeter far away from what the other
measurement lead is doing.

If the power supply doesn't respond with a higher +3.3V
as you pull hardware, then the power supply could need
replacing. There is no practical way to measure the
power consumption of the video card, so you won't know
if the video card is actually at fault, or the power supply
has become "a weakling". The best advice I can give, is
if the power supply is always on the low side, no matter
what load is on it, then swap for another supply.

In terms of your temperature symptoms, leave your computer
in its normal configuration, but with the side off the
case. Using your hand, touch the various power consuming
parts, like the surface of the video card, the heatsink,
MOSFETs around the Vcore circuit, etc., to see if anything
is kicking out more heat than you would expect it should.
If the CPU heatsink is cool to the touch, and yet the temp
reads 57C, that means the CPU die isn't touching the heatsink.
You should check that the clamping tension is there, to hold
the heatsink to the CPU. Perhaps the CPU die needs a new
coat of thermal grease, something you should be able to
buy at a computer store. A proper working CPU heatsink
will be warm to the touch, as it carries heat from the CPU.

HTH,
Paul
 
Back
Top