Power supply or hard drive?

[Nite Owl]

I've been having a transient problem with my desktop snapping off. I don't know how else to describe it, it just suddenly shuts down. It started a few months ago but it was rare but in June it started to happen more frequently. It started freezing too. I've called Dell a number of times and it always just started up and I was on my way until it happened again. My son thinks it might be the power supply, Dell says the hard drive. I was on the phone between the hardware and software techs today for about 5hrs and they are sending someone to install a new hard drive on Monday. I have things backed up but don't really want to replace the hard drive unless I really have to.

Today the fan was sort of noisy and I decided to shut down and restart, it didn't shut off by itself this time. When I restarted it would only go as far as the black screen with the Windows flag and the little blue squares would stop and that was it. Did system restore and various diagnostics for the hardware which were normal.

The tech tried a new installation of Windows this afternoon and it wouldn't install. When it got to the 'do you agree' it wouldn't accept and go on.

It's a Dell 8250, P4, 2.4,windows xp pro sp1, 2 1/2 yrs old. When it works it worked well, not slow at all.

Any ideas of which it could be?

[Lugnut]

1. I just heard today

Of another Dell user who had to replace her power supply. She claims that it's a common problem with Dell computers that are about 2 or so years old. Hopefully somebody who actually knows something can help you here.

[Nite Owl]

2. That's about when

it started too at about 2yrs old.

[bemildred]

3. Dry/hot weather where you live?

Could be heat issues, power supply, power adequacy (amps), electrical grounding issues, in some combination. This stuff will drive you nuts.
I'm getting similar issues on my wife's very old eMachines right not. The PS is new. Right now I think it's grounding between the Monitor and desk box, maybe a bad power bar or bad (reverse polarity) wall socket. I put a new CPU fan on it, and it went away for a while, but I put the case back on and it got hotter here and it came back. The CPU and hard drive are not hot when it pukes, hence I'm thinking it's power related.

I am skeptical that your problem is the hard drive.

[Nite Owl]

4. No, the house has

central air and the vent is right by the cpu. I have a ups and looked at the log just yesterday and nothing has been recorded for outage or power surges.

I do have a dog that sheds though and that could mess it up inside. I can't open the case but back gets the canned air. The case doesn't(didn't) feel hot.

[bemildred]

5. I don't have any real firm ideas.

If it happens regular, you can do fault isolation, replace/change things one at a time and see if it goes away.

I don't think a failing hard drive shuts the machine down, it hangs it.

I doubt it's dog hair, etc. but it never hurts to vacuum inside (carefully) and suck the crap from the fan gratings, make sure the CPU fan runs, etc.

The AC puts out dry air, which makes me think grounding again, but if both the Case and Monitor are plugged to the UPS that would seem unlikely to be it. I do like it better when the monitor plugs into a power socket on the back of the case power supply, seems to work better.

[Nite Owl]

6. It's still under the extended

warranty so Dell is paying for whatever it is until it run right but I don't want to do all that reinstalling if it is the power supply that needs to be replaced.

Good luck with your problem too.
Such a love/hate relationsip with these machines.

[bemildred]

7. Yeah, make them pay, but copy your stuff first.

Love/hate: yeah, you got that right.

[jayfish]

8. Something That Might Help You Diagnose The Problem.

Before I start though, I would say you are looking at a bad RAM stick somewhere. Ok here goes.

>right click on "My Computer"
>Select "Properties"
>Select "Advanced"
>Find the "Startup and Recovery" section on the advanced page and select "Settings"
>Un-check the "Automatically restart" check box
>Click "Ok" 2x
>Reboot

This will not fix anything in and of itself. What it should do however, is give you a BSOD rather than just shutting down. Using the info displayed in the BSOD can help you more accurately diagnose the problem. Post your results.

Jay

[bemildred]

9. A word with you Sir.

That seems a good idea to look into (the bad RAM).
Can you get transient failures that way?
How would that work?
Any input you have on the subject will be appreciated.

[jayfish]

11. RAM Problems Are Notoriously Transient.

But the symptoms are usually the same: hard locks or reboots. I'd try to give a technical example of what a general problem could be but it's difficult to find an exact cause in any one case. I'll give you a less technical example from a PC I worked on just last week.

The symptoms were that this machine would generate Windows errors, BSOD's, hard locks and reboots when accessing a certain program and only that one program. Since the main use of the PC was this one program, getting rid of it was not an option. Since I'm not one to ever look at hardware first and all the information I gathered pointed to a driver issue, I started with the OS. It was a mess; spyware, corrupted system files, incorrect and duplicate drivers, you name it. I got the OS back to where it was running correctly and updated all drivers yet the problem still existed. I had a couple of good sticks of memory laying around so just for grins I swapped them out. ...problem solved.

Thing is I really cant say why the memory was bad and what it was about this program that triggered the problem.

Jay

[Nite Owl]

14. I haven't been getting any error messages

but the system has been freezing suddenly. That's why I decided to shut down when the fan got noisy yesterday so I would get to it before it crashed and shut it down properly. It wasn't even that bad I was just being caucious. Then doom. Did all the scans over the weekend too and they were clean.

[bemildred]

15. Thank you.

My previous experience (where I understood it) was failure in POST, (maybe with a beep code), or reboot/lock, and soon. I had never heard of memory as a source of transients, but my education is very spotty at that level. This problem pops up at longish intervals (hours/days/weeks). I'll play with the memory and see what transpires.
:thumbsup:

[jayfish]

17. Something I Forgot To Add.

I tested the memory from that PC with several SW based memory testers and it passed every time. If you have two or more sticks of memory, the easiest way for you to test is to take them all out and rotate them one at a time until you run into problems. Then you can isolate the bad stick.

Jay

[bemildred]

24. FWIW

Replaced 2x32M with a single 64M memory module. Computer runs
fine, most of the time, but if left on, The problem persists, i.e.
it performs a hard shutdown.

Because of the way it manifests, I am thinking it's a problem with
the PS and/or a grounding problem. Replaced the PS some time back,
it's an old machine, and it was fine for a while. Now it is hot & dry
here and I have noticed that electrical issues tend to occur in those
conditions.

Later.

[jayfish]

25. What Processor Is You PC Using?

Jay

[bemildred]

26. eMachines 400i, Celeron 400 I believe.

From 1998 or thereabouts. I'm amazed the hard drive is hanging
in there, too.

[jayfish]

27. Hmmm. I Was Going To Go...

with a heat related diagnosis next, but a C400 should be fine heat wise. Actually you might have a pretty good overclocker there even with stock cooling. It still might be heat though. Here is what I would try and only because it's fast, easy and fun. ... well maybe not fun. Run your PC without its cover on and point a fan at the CPU area. Let it run for a while and see what happens. If it shuts down then you can eliminate heat as a culprit. If it doesn't then you will need to find out what component in your system is overheating(CPU,memory,videocard,chipset).

Try it and let me know.

Jay

[bemildred]

28. The cover is off now. That "seems" to help, but can't be sure.

Checked it right after a shutdown (by touch) neither the CPU
or hard drive or power supply seemed hot. No smells. The CPU
fan seems to run fine too, and is newish. The video is an old
ATI, on-board. I haven't really checked it.

The other thing is this, the "fix" after a shutdown is to
disconnect the power (unplug the PS) for a short period, then
it will turn on again with the front button. But it doesn't like
it if you do it too soon, and it will shut down again, which does
sound like a heat issue.

It's an ATX PS. No on/off switch on the PS. The monitor is powered
separately (power bar), no extra connector on back of the PS.

I'm kind if stumped, thinking I'll have to do fault isolation
like you say, but don't know if I'm that motivated, if you know
what I mean. I have enough junk parts to put together a 300Mhz
Celeron in a tower case with an old AT PS, which did not show
this problem last time I had it together.

Do you have any thoughts on the CPU fan and bad goop between the
fan and CPU, or anything like that? It just doesn't seem right with
and old Celeron like this.

Edit: the period for which it will run before puking seems to vary
widely too, somewhat random.

[jayfish]

30. It's Possible The The Thermal Compound Between...

the CPU and heat-sink has gone bad. (it is an aging setup) It's not hard to replace, just make sure you use as little as possible. It's meant to fill the natural cracks and crevasses between the two surfaces, not to actually transmit all the heat by itself.

http://www.1coolpc.com/install/arcticsilver.htm

Instead of waiting for a while before you turn the pc back on try this. Unplug the power from the box then depress the power switch. This will clear any residual power from the PC. Plug the power back in and hit the power button. It might come right back on. BTW that could be a different problem all together or a part of your main issue. Hard to tell without being able to put my hands on the box. Keep experimenting and posting the results.

Power

[bemildred]

32. Tried the power switch last night, no response.

I'm going to see about the thermal grease today, that makes
a certain amount of sense.
Later.

[bemildred]

34. Completely refused to come on this morning.

Eviscerated it.
Went and bought an ATX PS tester ($10) and some Arctic Silver goop ($10) and a new fan ($15) at CompUSA.
Thought about new PS, but it's a MiniATX ($70 for 150W) so I decided to wait.
Checked the CMOS battery (OK).
Tested the PS. First acted funny, then came on and all connectors were reported OK.
Cleaned the CPU and old fan and applied a bit of goop and put the fan back,
made real sure the fan power connector was oriented right.
Put it back together, checking all connections carefully.
On power up says CMOS is wrong, went all through BIOS setup.
Turned off APM stuff, front switch is now just a switch.
Now seems OK, but time will tell.
I consider that it not coming on after sitting all night (even after various tricks) means it's not a grounding issue.
It was funny the way the PS was wrong, then reset and seems OK. Thinking maybe APM related.
Hmmm. We'll see if it lasts.

[bemildred]

35. This machine seems to be fine now.

It seems to have been a BIOS APM related issue, and seems pretty clear it was not heat or the PS. I make no claim to understand it, but based on what I did to it when it started being happy again, it was the BIOS APM stuff. It is possible that the CPU fan connection had something to do with it too, either wrong orientation or two wires instead of three.

In any case, I was very careful about the CPU fan connection, I used the PS tester to reset the ATX PS, and I turned OFF all the BIOS APM stuff, and it was happy as a clam.

YMMV, and thanks for the help.

[Nite Owl]

10. I can only get on

in safe mode. Did that yesterday to make some cd's to save stuff. If I try to boot normally it gets stuck on the first windows page, the one that usually flys by, black with the microsoft flag. The little blue squares freeze and it won't load. So should I still try this?

[jayfish]

12. Does The System Shut Down While Running In Safe Mode? -NT-

Jay

[Nite Owl]

13. Trying to remember

I think the tech just told me to restart and then it just went to the same page that wouldn't load windows.

[jayfish]

16. This Rarely Works But,...

at the menu screen where you select safe mode, there is a selection to return to the "last known good" configuration. It is entirely possible that the "tech" had you do something that caused you boot problem. This is supposed to set Windows back to the state it was in prior to the last time you were able to boot normally. If this doesn't work, I would say your next step is an "in-place upgrade" This will probably get you back into Windows but if your problem is indeed HW related, it will return. At least you will be able to get into the OS and get some info. A little tip to remember when talking to corporate tech support is to immediately ask to have your problem escalated. This will put you in contact with a tier II tech or higher. Your experience will be much better.

Jay

[Nite Owl]

18. That was done

before I even called them. Didn't work. Tried it again with Dell. We did try to reinstall windows but when we reached the 'I agree' it wouldn't go any further. He said something to me about that too, that we could get this to work but it it would just happen again down the road so bite the bullet and get the new hard drive. Are their tests that can be done on the power supply that I should ask the tech to do on Monday when he comes? I think I have most of my stuff backed up on cd's. Sure wish I had gotten that external drive I was looking at last week though, it sure would make this a lot easier.

[jayfish]

20. Well, If It Wasn't A Dell.

(not saying Dell is bad)you could go into the system bios and examine the various voltages yourself in real time. If the PSU was under-volting the CPU or 12v rail (Hard drives CD-ROMS, etc;) you would see it there. Other than that the tech could use a voltmeter to test the PSU. It's probably less expensive for them just to replace it though. Go ahead and let them replace all of the hardware they think needs to be replaced. Then when they finally stumble onto the problem you will have all new HW.

Jay

[Nite Owl]

19. Did you mean

as in when it shut down unvolintarily or shut down and in go to shut down and let windows do it? If the later it is sporatic, can go for days with everything ok. Several times it happened when no one was even near the computer. Just snap, off. I have a ups too and the log is ok, shows no power surges or use of the battery. It has a voltage regulator on it too. The freezing is new in the past week or two. The cursor just freezes and control/alt/del doesn't do anything.

[jayfish]

21. Involuntarily.

It really is starting to sound more and more like memory to me. Possibly a flaky CPU too, but more like RAM.

Jay

[Nite Owl]

22. Thanks jayfish

I'm going to ask them about that too. If it happens after the new drive then they have the ps and the RAM to consider. The warranty is up in December but if the problem is on record they are obliged to keep fixing it 'til it's right. So glad I got my little laptop so I'm not offline in the meantime!

I don't think I would get another Dell being as they are so red and I know that now but I gotta say they have been totally ameniable so far. Very short wait times and spoke with techs in Canada. Every tech said don't worry we will work at this until it is fixed and you are satisfied. I hope they meant that.

[jayfish]

23. I Don't Mind Dell Servers...

and/or desktops and the support dept's that go with them. Their laptops and support for them are atrocious though.

Jay

[Nite Owl]

29. Well the new hard drive

is in. He came on Saturday night and I finally got everything installed. It is much quieter now. Maybe the noise was something I just got used to but now it doesn't even sound like it is on it's so quiet. SP2 is downloaded ok and I didn't reinstall Norton but used AVG. Got all the Word files back and most of the pics. Lost some links that I had but most were bookmarked on Yahoo anyway.

Maybe they were right and it was just the hard drive, guess time will tell but so far so good. It's kind of nice having a real clean machine.

Oh, he was also sent another motherboard by Dell. Refurbished. He put it in and it wouldn't work so he put the old one back and it was fine.

[jayfish]

31. Excellent.

Hopefully your problem is solved. Thnx for the update.

Jay

[HamstersFromHell]

33. Just a few thoughts about any computer with these symptoms

I've seen similar behavior in a few machines in my years in the business.

If it's an "out of the box" machine (i.e., brand name), you have to remember that they compete with each other on one common ground - price.

Therefore in order to still make a profit and be price competitive they cut corners *everywhere*. Nothing usually sub-standard, but often marginal if you add more things to your box that draws additional power or creates more heat.

I can remember Packard-Bell 386SX computers that came with such a marginal power supply (36w!!) rating adding 8 Megs. of RAM or a second hard drive would cause it to fail to boot entirely.

Intel factory heatsinks are notorious for being a bit on the marginal side, especially the Celerons. I've had client computers that would BSOD or shutdown for no apparent reason and opening the case would revel "normal" levels of dust, but removing the CPU fan from the heatsink would show the very small fins underneath completely clogged with dust. Canned air cleaned em up fine and resulted in a 30 degree or more reduction in CPU heat.

Another thing to look at is if the BIOS/chipset contains code for overheating shutdown. Some are set to merely throttle back at a certain temp., while others only offer a point at which to shut down entirely. Poking around in the CMOS might reveal that the shut down point is set much lower than what the CPU can actually stand. Intel usually says their CPUs will operate fine up to around 170 degrees F., while a lot of "heat monitors" in system BIOS' will default to shutdown at around 150 degrees...and it takes just a little heatsink dust to reduce efficiency to the point it'll trigger.

One last thing is if the CPU fan has 3 leads...the third is for an RPM monitor for the fan in the chipset, and is usually set to immediately shut down if it's not receiving the pulses telling it the fan's running. I've only seen it once, but a faulty fan intermittently not sending pulses can cause random shutdowns at any time. (Usually there is a setting in the CMOS to ignore fan rpm....you can set this to off temporarily to check if it's the culprit, but I don't recommend running the computer this way...it's there to prevent you cooking your CPU in the event of fan failure.)

Don't go replacing things until you've throughly cleaned the CPU heatsink...most of the dust will be hidden under the fan in the fins and not obviously visible without removing the fan from the heatsink.

[WLKjr]

36. My 2 cents

Here are some more trouble shooting steps:

Check your fans, sounds like it could be a cooling issue.

you may have a chip creep too, pull out the ram/cpu/cables/cards and reseat them into where they are supposed to go. This little trick sometimes helps BSOD.(blue screen-o-death)



let me know if this helps.