Serious problem!

[francescopadormo]

I need some advice about a major problem with my new(ish) computer.
It is 6 months old( i bought from a 'we'll make it for you' website which has since gone bust...) , and the processor is an AMD Athlon 64 2800. I'm not sure of the details about the motherboard or hard drive, but could probably find out.

I have been told by a computer repair person that it "sounds like" a hard-drive problem, which is why i'm posting in here.

So basically, every time I turn on my computer, within 5-10 minutes it crashes to a Blue Screen Of Death. These are often different, sometimes 'Irq_Not_Less_or_equal' sometimes 'Memory Management' and other such wonderful messages.

If I then turn my computer back on again after a crash, it might get halfway through starting up then crash again. Then if i try again it will usually just make whirring noises but not actually try to start up. It usually needs a few hours(at least) before it will then work again.

I have checked to see if it is overheating, but it doesnt look like it. The air coming out of the fan at the back is not THAT hot, and all the fans inside seem to be working OK.

I have tried diabling the L1 and L2 caches, as it sometimes advises me to, but that did nothing. it has recently started advising me that my BIOS needs re-installing, but I have NO IDEA how to do this, or what this means. (in case my fears are true, i have no floppy-drive, which i suspect is what BIOS is installed on).

Some-one also suggested that I format the hard-drive, but I don't think I will be able to, as it takes longer than the amount of time the computer ever works for.

It also sometimes recommends that I re-install drivers, which I have tried doing using the CD which came with it. This either did nothing, or crashed before the installation was completed.



If anyone has any suggesdtions about what I should try, or advice on what I could do, please post telling me. Seriously, i'm desperate!

Thanks

 

[dvdpiddy]

Holy crap dude your harddrive is f'ed up okay either buy new one or rma it.

 

[ww1065]

Your power supply could be going south. Have you disconnected the power cord and reseated all cards, memory, and connectors?

Overheating could still be a problem. If the heatsink is not properly seated over the CPU, or substandard heatsink compound was used, the heat will not be drawn away from the CPU properly. In that case, air temperatures are meaningless.

My company laptop started doing something similar to that & it was simply the BIOS battery dying.

 

[francescopadormo]

Thanks for your advice.

I don't think the BIOS battery is dead, as the date is still right when I turn it on.

I'll check the fan is positioned correctly, and try re-seating everything though.

I don't know what " rma"-ing a hard drive means - a little help!?

Any other suggestions???

 

[ww1065]

I don't know what " rma"-ing a hard drive means - a little help!?

Return Merchandise Authorization. It might be a bit difficult in your case since the source has gone belly up. You might get some good from the HD manufacturer.

It really does not sound HD related, more like a power supply, bad memory, or simply a loose connection.

I hope this gets to you before you pull the heatsink off of the CPU: Do NOT pull the heatsink off before you buy some new heatsink compound!!! I apologize for yelling, but it is critical to use new compound in such circumstances. You need to clean all the old compound off the heatsink and CPU, then put a pea-sized amount on the CPU, then carefully work the heatsink down & lock it down to the CPU. Arctic Silver is the compound of choice.

EDIT: Alcohol is good to use, be careful to not scratch anything. Also, the above comment is right on the money!!

 

[francescopadormo]

Thanks again for the tips.

Unfortunately i don't think i can run Memtest86 as i have been informed this takes some time, which I never have on this computer. The longest it has worked since it started playing up was under 15 minutes.

Don't worry - i haven't taken the heatsink off yet! But do you mean that i should pull off the heatsink, then clean off the old compound with alcohol, then re-attach it with new compound?

Also, if it was a loose connection, would it really crash at such regular intervals?

 

[ww1065]

I recommend leaving things like pulling the heatsink off for last. There is a possibility that the vendor you bought from had quality issues that lead to their demise, however. They may have used poor quality compound (which has failed over time), or improperly installed the heatsink onto the CPU, leaving voids in the compound to let the CPU overheat at those locations.

Start conservative first!

Start with unplugging the computer from the outlet, and checking all connections. Always keep a hand, forearm, etc. on the metal part of the case (better yet, buy an anti-static strap from Radio Shack). Carefully remove and reseat the memory stick(s), power supply connectors, drive connectors, and even the cards.

Loose connections have been known to do just what you are experiencing. Bad connections can heat up, then fail.

At this point, you've got to play doctor to a patient that can only moan, and not give you more specifics.

Check the connections, and then try the Memtest86, you have nothing to lose. Hopefully it is a connection issue & Memtest can run long enough to provide meaningful information.

You may already have trashed drivers when it crashed while trying to update/reinstall them, so you can't do more harm at that point running Memtest.

Do not try to upgrade your BIOS at this stage!!!!!! I know you mentioned an error stating that it needed attention; a problem crashing the computer while reflashing the BIOS will send the motherboard back to its maker (unless it has a backup BIOS)!

Power supplies that have noisy or out-of-spec outputs will also cause problems. Wattage ratings are useless if the power supply is a piece of junk, or simply failing.

As to the BIOS battery, the laptop clock stayed accurate. . .

 

[francescopadormo]

Thanks again for the tips. I'll give the connections a go later, and try running Memtest too.

Re: power supply - how can I tell if its out-of-spec or failing? Are there any obvious things to look for?

 

[ww1065]

Re: power supply - how can I tell if its out-of-spec or failing? Are there any obvious things to look for?

Besides an accurate voltmeter and an oscilloscope, substitution is the only way. You are experiencing one of the common symptoms.

You can bypass the trashed driver issue by putting in a bootable Linux Live CD. This will boot and run directly from the CD ROM drive. It would give you web surfing, simple word processing, and a few other items without using Windows or its drivers. You can give the computer a workout this way without the worries of trashed Windows drivers creating problems.

Live CD's are made to operate as standalone ways to use or try Linux.

It will run slooooow, as it operates off of the slow CD ROM drive, but it will give you an idea on hardware issues. If it runs Linux from the CD drive for a couple hours, you may have solved your hardware issues & then can go make software repairs.

'Simply Mepis' and 'Kanotix' are two good ones to try.

Note that you will have to burn an iso image onto the CD, simple file copying will not work!!

I admit to having the Linux bug, but use Windows more than half the time. The Live CD's are a great way to isolate any software/driver issues that may reside on the hard drive. I use this to diagnose Linux hiccups, as well.

 

[pscowboy]

francesco:

Stop trying to be too, too careful about making a move.

Before you do anything else:

1. REPLACE THE POWER SUPPLY! ($30-$40)
2. Undo the heat sink. Meticulously clean off the two mating surfaces. Apply a very thin coat of Arctic Silver to each, and replace.

These two things are simple operations for someone just a little handy to perform. If you're not comfortable with this, corral a dude who can.

We must get those two hardware issues out of the way before proceeding with any other troubleshooting. You most likely have software corruption because of one of those two.

 

[francescopadormo]

OK, so I took your advice and have run Memtest.

Unfortunately, before it crashed(a surprisingly long 15 minutes) it DID come up with an error.

I guess this means my RAM is f**ked and i'll need to replace it.

BUT does this mean that the other things that mght have been wrong with it(hard-drive, power supply, etc) are OK?

I only really want to replace the things which are broken, as it's practically new!

I'd appreciate any advice on what still needs to be done.

 

[pscowboy]

I've been doing this stuff (especially hardware) for over 20 years; so trust me.

Attend to the ram. You already know you were gypped on that. Buy from Crucial, Kingston, or Corsair. Make sure you get the exact memory for your mobo. Read the imprint off the board yourself if you don't know. Be interesting to see if the mobo they told you, you were getting, is really in there. Put the new memory in, and see if you're stable. But read on.

You were probably given poor quality goods for your purchase. Your symptoms are screaming power supply. Spring for the $40 and change it. Do not buy somebody's used one.

While you have the box open, work on the heat sink - cpu relationship. That'll cost you next to nothing. Most computer dudes have Arctic Silver laying around.
You may come across a pad between them. If it's in perfect shape - NO PEELING, NO CRACKS, NO SCRATCHES, leave it alone and resecure the sink. If it's paste, use alcohol (or paint thinner) to clean it off. Polish and dry the surfaces right away with a VERY soft cloth.

After all the repairs are done, you still may have a corrupt OS because of the hardware failings. See if you boot up and stay stable. Otherwise, you're looking at a re-install.

 

[francescopadormo]

Coool, thanks for the advice.

I'll get started on the power supply etc tomorrow.

BTW, is it OK to just pull off the heatsink from the processor? Or does it need to be loosened up first?

 

[ches111]

Dude,

Unless your Bios is telling you that you are having heat issues with the CPU "LEAVE IT ALONE"....

Everyone here seems to think it is a NO Brainer for people to replace their obviously inferior compound with Arctic Silver. I would be cautious and take one step at a time until stable.

If memtest showed an error you could have one of two issues. Motherboard or memory. Your Motherboard may be having memory issues due to a couple reasons too. Is it a Power Supply issue? Improperly seated memory? Dust?

First, If your PCs interior is not "clean" clean it with a can o air.
Test
Second, Re-seat all of your components.
Test
These next steps are subjective:

Third, Run checkdisk. "note any errors" if errors look at your HDD and find the vendor Name. Download and run THEIR HDD diagnostic tool. If it throws an error that says your drive is failing then replace your drive.

If the drive is OK and problem persists. Replace memory.
Test

I think you get the jist. Start with the easy NON-Intrusive things and move up from there. Replacing the CPU cooler/CPU being last. Motherboard failures and Power supply failures can be the toughest things to detect as both will have similar symptoms.

The one step approach is 1) cheaper, 2) gives the best results, 3) will help you to define what the problem is instead of having a pile of used equipment that you no longer know if it is good or bad.

Good luck,

 

[Codesmith]

Don't jump to hasty conclusions and start randomly buying new parts.

Running a memory test on an unstable system is unfortunately inconclusive.

All it tells you is that there was a problem when performing a memory operation, that in and of it's self it does not indicate differentiate between a problem cause by a bad powersupply, motherboard, overheating CPU, overheating chipset ect...

What you really need is to start swaping parts with a stable system.

PS

If you cannot swap parts with a stable system and are forced to guess and replace parts, I would try PSU then RAM. After that I would suspect the motherboard.

First I would try booting to CMOS setup (BIOS) and watch the temperature readings. If you see the reading rise steadily until the sytem crashes then you know you system is overheating.

Don't bother testing any other component until memtest86 runs without error.

Also you should be testing your system will all unnecessary compontents removed. That includes all USB devices, Sound cards, PCI cards ect.

All you need at this point (beyond the obvious PSU, CPU) is keyboard, mouse, video card, and whatever you are using to run memtest (FFD or CD-ROM).