When Good Computers Go Bad

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
Past results are no guarantee of future performance, or words to that  
effect as used in those investment ads.

I have this homemade computer running the first version of Windows Home  
Server. The hardware was basic: a dual-core Celeron, 4gB of memory, a  
gigabit Ethernet card, a drive port board to gain two extra SATA ports,  
some sort of cheap basic video card, and six drives (5 X 2tB and 1 X junk  
from the closet). This computer was dead reliable and I generally forgot  
about except on update Tuesdays so it just ticked along in its closet in  
the basement. Then one day it didn't tick...

Since the box was totally dead and unresponsive I figured it almost had to  
be the PS so I ordered a replacement. When the PS came I opened the case  
for the first time and noticed that the existing unit was a high-end  
Seasonic and those things just don't fail like that. I replaced it anyway  
and found the computer still unresponsive. THEN I actually started looking  
around and found: 1) all four pins on the CPU heatsink were popped and the  
HS was loose, 2) the 120mm case fan had completely lost its bearing and the  
hub/blade assembly was just lying there basically unattached, 3 the cooling  
fan on the video card was jammed up against the bottom of the Ethernet card  
and appears to have had both bearing failure and (odd as it seems) curled  
plastic blades.

At this point I realized that heroic measures were called for so I went  
online and found an Asus MB with six ports (doing away with the need for  
the extra port card) and onboard video (doing away with the video card) and  
onboard GigaBit (doing away with the interface card), 3.4 gHz i3 processor,  
and 8gB of memory sticks. After replacing all of the guts I found that the  
ASUS driver CD didn't recognize the OS at all because of its Server 2003  
roots and age so I scrounged around till I found most of the needed drivers  
(no HD sound, but who cares on a server?). Then it was a matter of  
convincing the OS that everything was kosher, no mean feat with WHS  
software which is extremely touchy about every little change. Eventually I  
convinced it that things were as they should be and error free. Thankfully,  
the OS demanded to be re-blessed because of the massive hardware changes  
but that went without a hitch..

Guess I can't afford to totally ignore the three computers in the basement  
any more. Sigh...

Re: When Good Computers Go Bad

"John McGaw"  wrote

Quoted text here. Click to load it

Because it was a server, you were unaware of the multitude of hardware  
failures until it stopped serving. I run a five computer farm in my living  
room and I own a multimeter. I would never buy a new PS when not needed.  
I've had a multitude of problems, that I have fixed. My oldest machine was  
new in 2003. On another I built a 3 drive RAID 0 array. The drives are long  
past the 5 year warrantee, but maybe they will last 15 years.  

Re: When Good Computers Go Bad

On 7/31/2014 5:32 PM, Norm X wrote:
Quoted text here. Click to load it

I have 5 computers running 24X7 also but in this case, since the machine  
was basically a junk-box assemblage of parts and in a location that is not  
easily accessible, I jumped to the conclusion that the PS was the fault  
based on my experience over many years of computer troubleshooting. A  
multimeter really wouldn't have been necessary in my case since I have a PS  
tester -- the problem would have been actually getting enough access to  
remove the side cover to get to anything. Oh well, now I have a spare  
Seasonic ATX PS and an essentially new computer. When my new case fan comes  
in this afternoon I can button it back up and stick it back on the cart  
with the others and wire up the KVM and get back to normal.

Re: When Good Computers Go Bad

John McGaw wrote:
Quoted text here. Click to load it

Your CPU, GPU, and hard drives have temperature sensors.

Speedfan has an email capability.

Perhaps an overheat event can generate an email ?



On some video cards, if the fan stops spinning,
the overheated GPU melts the fan body.

The video card may not actually have a three pin fan on it
with RPM monitoring. And even if it did, there would be
no readout capability, coming from the video card. The video
card does offer a temperature readout though, which is almost
as good as fan fail.


Site Timeline