motherboard or cpu--which one is damaged

[dynamicfluids]

As originally reported in thread602-934008, my custom-built dual processor computer worked relatively flawlessly for two years, before it developed repeated problems starting the computer. The computer would go thru a series of lock-ups and reboots before it reached a functioning state.
It seemed after each restart the computer would progress a little farther before it locked up again. After 5-10 restarts/reboots, the computer would run fine. Once I shut the computer down, I would have to go through the whole process again. I tested the RAM, no problems there; I replaced the lithium battery on the motherboard, didn't seem to help; I reinstalled the OS, which seemed to help for a week before the problems returned. I also blasted the inside of the computer with compressed air as well as cleaning the dust out of each heatsink. This seemed to buy me a couple days before the problem returned.

During recent boots, I noticed only one CPU is registering. The computer worked fine, I just only had use of one processor. Tonight, I restarted the computer, and now both processors are registering at boot time.

Is it safe to assume the motherboard is damaged? Or could it be a processor? If it is a processor, how do I identify which one is functioning. I would gladly pay the $215 for a new motherboard, or $160 for a new processor if I could guarantee I was replacing the correct item.

I would appreciate any feedback. Thanks.

Scott

Pertinent Specs:
Windows XP SP2
Giga-byte GA-7DPXDW motherboard
Dual Athlon 2000 MP processors
2 x 512mb registered ECC RAM
Antec TrueBlue 480w power supply
ATI All-in-wonder AGP card
Adaptec SCSI card
Asante network card
Turtle Beach Santa Cruz sound card
3 SCSI hard drives
EIDE CD-RW drive
EIDE tape back-up drive
Connected USB scanner

 

[G0AOZ]

Have you tried reseating both processors and applying new heatsink compound?

Antec PSU's have a good name, but try temporarily changing to a known good one just to rule that out of the equation.

Do you have any bulging capacitors on the motherboard? This board might be old enough to have been fitted with some of the dodgy caps that were around at that time.

Have you tried running this board with absolute minimum attachments, i.e. just an FDD and video, and no other drives or cards?

Have you checked CPU and system board temperatures, if the BIOS provides these figures?

Are you running with the latest BIOS update?


ROGER - G0AOZ.

 

[tf1]

What a beast of a fault! Could this be a temperature induced problem with the MoBo or the PSU? Once warm and running it seems OK, but from cold it would seem to be a right sod.



Regards: tf1

 

[dynamicfluids]

Roger, I've tested the power supply, the voltages are in the correct range specified by AMD.

I haven't looked for bulging capacitors on the motherboard, but I'll take a look. If I find one, is it possible to replace?

I haven't run the computer without all of the boards and drives removed, I don't think that would cause a cpu to disappear, but I'll try that after examining the mobo.

The cpu and system temperature are ok. The bios has been updated to the latest version.

Thanks.
Scott

 

[G0AOZ]

To be sure, what you need to do is monitor ALL the PSU voltages as the machine is booted up and going through its various lock-ups, freezes, reboots, etc. Whilst what you're doing is ok, it's only testing it at a given moment in time, but it doesn't I'm afraid eliminate any changing states which might just exist as the board goes through different stages. Having said all that, I would still try a different known good PSU.

I have also found bulging capacitors within PSU's, although I would not recommend replacement unless you are entirely au fait with dangerous voltages which can be found inside even with the power off and disconnected.

Motherboard capacitors can certainly be changed - I've done this several times, although unless the board is a very expensive one it is hardly worth the effort from a commercial standpoint. See

http://www.badcaps.com

Definitely worth disconnecting all cards and drives and just running say a DOS application on floppy disk. You are then clearly eliminating a whole bunch of peripherals from the equation!


ROGER - G0AOZ.

 

[Martin999]

Just a thought, can you tell which processor is not registaring, then swap them round ???

Martin

 

[dynamicfluids]

Roger,
I don't have any "know good" PSUs laying around. I'll try to borrow one or I guess I'll have to purchase one.

DOS applications? Floppy disk? I don't think I have these laying around either.

Martin,
One of my original questions was how to determine which processor is registering. I'm still waiting for a reponse from anyone who might know.

Scott

 

[tf1]

Have you tried removing a processor (say P1) and firing up? If P0 fires up in socket 0, move P0 to socket 1 and test again? Then do the same test again with P1 i s0 and then s1.

That should determine if there's a bad processor or a bad socket or a firmware or OS problem.



Regards: tf1

 

[G0AOZ]

Well if you don't have any DOS applications on floppy disk, how about a Windows 98 boot disk with SCANDISK on it. Run a surface scan on the floppy - that'll give it a bit of work to do without you having to introduce a hard disk or a CD...

TF1, I'm not sure if the CPU will work in the "second" socket on its own. Most of the twin CPU boards I've seen usually demand the CPU be put in the "first" socket. But still worth the swap around to see if it makes any difference...

ROGER - G0AOZ.

 

[PRPhx]

Does your event log have any error entries when the reboots occur? Also, try setting the BIOS setting (if you have it) to HALT ON ALL ERRORS. This SHOULD give you a BSOD reporting what's happening to what program.

 

[dynamicfluids]

I checked for bad capacitors on the mobo-- everything looks ok.

I checked the voltage readings in the BIOS and it revealed that not only are some of the voltages out of spec for the ATX standard, but they are out of spec for Antec's own product claims. The +5V varied up to 7%. The max variation for the ATX standard is 5% and Antec's stated max variation is 3%. The other voltage out of spec is the +12V. It registered at 5.3% max variation, which isn't far off of the 5% ATX standard; however, it is well over the 3% Antec standard. All the other voltages were at or below the 5% max variation.

Now, the question is, who makes a quality, reliable and quiet power supply around 500W? I've noticed that all the new power supplies are ATX v2.01. Are these PSUs backward compatible (out of the box) with older systems such as mine?

Scott

 

[tf1]

Silent! This is what last month's CS Magazine Lab tests found.

Antec Neo HE 500 8.28dB
Enermax Liberty 500 (modular cabling on this one) 8.07dB
Enermax Noisetaker 600W 6.73dB
Nesteq Semi-Fanless 620W 0-5.93dB

These are in order of noise and price. The last three all feature modular cabling. The Noisetaker features a huge fan that only runs occasionally when needed (and it was hardly ever running in their tests).

The Liberty is probably best compromise and value for money.

BTW, they are all 'backward compatible'.

Regards: tf1

 

[UUVGuy]

PSUs are one of the weak links in any computer system. A PSU can fail for many reasons. These failures are not manufacturer realated per se. In other words, you can have a PSU failure no matter whose PSU is installed. Have you contacted Antec to see if there is anything they can/will do about yours?

I have had very good luck with Antec though any of the other PSU mentioned by tf1 would be great. Just remember though that even if you get the latest and greatest, due to many factors, it too could fail sooner than planned.

Glad you were able to track it down though. PSU problems are the bane of techs.

Jim W

 

[dynamicfluids]

I e-mailed Antec. They stated that my PSU appeared to be defective; however, it's not covered under the three year warranty anymore, so I bought an Antec Neo HE 550 yesterday. I guess I should have read the box concerning cable quantity and type before leaving the store, because when I got home I realized the box contained only two cables with three 4-pin molex connectors each--I need a total of 10 for all the fans and drives in my box. The PSU did come with lots of cables I didn't need: PCI-X and SATA. You would think that, for backwards compatibility, all of the cables (except the motherboard connections) would be standard 4-pin molex and that adapters for the PCI-X and SATA power connections would be provided similar to the floppy drive adapter cable that is included, especially since all of the custom system configurations that builders of their own boxes create with the extra case fans and drives, etc. etc. Go figure. Back to the store for me today.

 

[G0AOZ]

That's a drag having to go back to the store for the extra connectors... Let's hope that new PSU does the trick for you!

ROGER - G0AOZ.