adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
Thursday, my wife and I went to her mothers for the evening AGAIN, (this is a problem, but not the one I wish to discuss). When I left, both my machines were turned on and happily churning out BOINC wu's.

When we returned, endless hours later, one was dead. The normal light on the router which shows a machine is connected to a port was out, so I turned on the screen and there was the "No signal" message bouncing about. I think there was a light on the keyboard. There were no fans running on the box.

I pressed the start button and nothing happened. I pulled out the mains left it for a while, put it back in then tried again to start it. Briefly, (less then a second), the disk activity LED and the LED on the floppy drive flashed, but that was it, the power light did not light at all. There is no light on the keyboard. It is in the same state today after several restart attempts.

I put the system together from some old, some new parts last autumn. The MoBo, (ECS P4S5A-DX), CPU, (Intel Northwood 2.533GHz), RAM, (1x128MB 1x32MB), and ATI graphics card, (G-Force II MX-400), were previously in another case together and working fine.

Currently, these parts are in an older case, with a Seagate Barracuda, SMC LAN card, and Antec 400W PSU, (all new in the autumn), a CD-ROM from yet another machine, and the original floppy disk drive. The case has one additional fan installed. In neither machine has the CPU been overclocked, indeed, because of a misunderstanding with FSB values, for half of it's 3 year life, it was clocked at 1.9GHz.

The system has NT4 on it, and basically has just been running BOINC, (Predictor, Rosetta, LHC, SIMAP and Einstein), and has a backup Apache server system on it, although it has never been necessary to have it active. I have a hardware monitor running, and it has never shown any excessive temperature, typically max's out at 45C, well less then my other machine. The box is opened and the CPU fins and fan blades brushed and vacuumed every couple of months.

So my question is, what is wrong?

I can rule out enemy action as our apartment was empty at the time.

The brief flashes from the bootable devices imply the PSU is not totally dead, and why should it be, it is new, and well overated for the load.

I believe the PSU starts when told to do so by the MoBo, (as opposed to the start button connecting directly to the PSU), certainly the front panel switch connects to the MoBo, so presumably at least something is working on the MoBo.

The very quick shutoff suggests something seriously bad has happened somewhere.

Any ideas? (Sadly, I think what has happened may have happened even if I was at home, so "stop visiting your wifes mother", whilst having enormous appeal, may not work).
 

Scout

Distinguished
Dec 31, 2007
1,068
0
19,280
When you put your nose in the case and sniff around, as well as smell the power supply, do you detect any acrid oder? If so, something may have burned out. See if you can detect anything like that. A power surge may have done some damage while you were away. Although if your other computer was OK then that theory may be a bit lame.

I would try clearing the BIOS. If that doesn't work, you're really in a situation where you need to swap out some parts to figure out what's gone bad. The power supply would be my first suspect. It's an Antec which is a very good brand, but I've lost a brand new Antec before in a puff of smoke!
 

Grimmy

Splendid
Feb 20, 2006
4,431
0
22,780
I remember a Compaq Server that I had for lab testing blew it's PS on me. Scared the living crap out of me too.

Then the smell came around.. Man, I was thinking at the time, I hope that is only the PS I smell .. hehe

Had to call it in for RMA.. That was 6 years ago.. I can still remember that certain poppin noise it made.. :lol:
 

sylvez

Distinguished
Dec 27, 2005
141
0
18,680
oh yeah speaking of which...

i blew a PSU for no reason whatsoever

the thing about it is its on the same line as my two other PCs

so its not that strange really, to see one blowing and the others in perfect working condition
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
Today, I had some time to play with the machine.

First, there was no smell, either originally or now. I am familiar with the smell of burning elctronics I'm afraid! If there had have been an event while I was out, it is well possible it would have cleared before I got home, there are other machines in the area and considerable air movement.

The first thing I did was take the cover off and try again. Both the fan on the PSU and the CPU fan started spinning but stopped again almost at once. Otherwise, the machine was the same as before.

I took the Antec out and put an old 300W Sparkle in. I knew this PSU to be working when I took it out of the same machine in October - the reason I'd put the new Antec in was because I thought the higher spec board, processor and hard disk may be a bit much for the old, and quite noisy Sparkle.

Anyway, turning it on, with just the MoBo powered, the same thing, the fans twitched but nothing ran.

I have now removed the data cables from all the disks, (Seagate 120GB, Hitatchi x56 CD-ROM and the "floppy" which is actually an LS-120), pulled out the video card, the network card and the RAM stick.

With just the MoBo, CPU/heatsink and front panel switches connected it is still doing nothing, fans twitch but thats it. The behaviour is the same with both the Antec and the Sparkle.
 

cubber

Distinguished
Feb 3, 2006
87
0
18,630
Looks like a fried mobo, it if doesn't start beeping madly due to the lack of RAM and vidcard. Sometime's electronics just go poof for no good reason...

Bummer. At least old mobos like that are really cheap on eBay. Might be able to pick up a whole system with a similar mobo for dirt cheap for extra parts.
 
I hope I don't sound crazy but this recently happened to me:

Similar issues, I would try to turn the comptuer on, and it would shut off right away. What was wrong with the Dell Optiplex 620?

BAD POWER SWITCH!

Never heard of that before, but the bad power switch was shutting the computer off every time I turned it on. I told Dell I thought it was the power switch after replacing the power supply, but Dell insisted that I replace the motherboard, RAM, CPU, and ANOTHER power supply before they sent me the $2 front assembly. The computer was only a couple weeks old. (This isn't a rip on Dell, I like 'em!)
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
That's a bit of a weird one isn't it?

I'd have thought that a switch going bad would first manifest when trying to start or stop a machine, not whilst it was running? Same with the reset switch, (although I'm not sure I actually connected it in that box as I don't use it).
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
Little bit more info, don't know if it helps. I put the Antec PSU back into the machine, and happened to leave the keyboard connected and have seen a little more then I have reported before.

If the PSU has been unplugged, OR if the PSU's own switch has been switched off, when switching it on or plugging it in BEFORE touching the front panel, 3 lights on the keyboard turn on then off. Once pressing the button on the front, the fans twitch then nothing. Having pressed the button once, pressing it again any number of times does nothing, no lights, no twitchy fans.

Unplug, or cycle the PSU's little switch and then again, the keyboard lights go up and down, then the first press of the button causes the fans to twitch.

The start button is not a latching button, it is a 2 wire momentary switch. To me, the fact that anything happens at all, (i.e. fans twitch), indicates it must be working.

I have also tried clearing the CMOS without any improvement.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
I guess I can rule out the PSU now. I put the Antec into this machine earlier today and it worked fine.

I guess also that it is now the MoBo or CPU.
 

cowboytech

Distinguished
Feb 26, 2006
170
0
18,680
About the best you could hope for is that the on switch is sticking, to test that possibility just unplug the on switch and just quickly jump the two contacts on the board, if it runs its a sticky switch. By the way, with all those problems, I would start with a base post test with the major components, PSU, MB, CPU, RAM, Video, and nothing else. Now we would know its one of these items and youve checked the psu, so, its one of the other items, and never forget the power cord.
 

goolong

Distinguished
Mar 1, 2006
3
0
18,510
My suggestion is do a visual inspection of the motherboard, you might have some blown capacitors. Also a bad powersupply could still give power, just not enough voltage. Best way is to first look at the capacitors and if all seems well there, swap out the powersupply and test again.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
The PSU from the dead machine was installed in this machine this evening as a test. This machine normally has a 450W Antec in it, it's a much more thirsty rig then the older setup, a hot Prescott, on a bigger Asus, with a lot more in the box. If the 400W can drive this, I think it could drive the older set up as it had been doing since October.

I intend to whip the MoBo out in the morning and have a look at it.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
I am already down to PSU, MoBo, CPU and heatsink. Nothing else in the box is connected, either power or data. I did try with the keyboard, the result was a little different, it is described further up somewhere.
 

goolong

Distinguished
Mar 1, 2006
3
0
18,510
Just check all the capacitors on the motherboard, if its bulging up that means its blown and a new mb is needed. Also, try reseating the CPU, put some new thermal grease on because some motherboards will detect if the temperature is rising too fast, it will not let it boot. Hope that helps.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
Okay, so I have taken the board out of the box and had a really good look at it, nothing strikes me as odd. All the electrolytics look normal etc. Similarly, once cleaned of the old Arctic Silver, the processor looks fine.

Reseated, it behaves as before.

Questions.

1. Would trying to power on the MoBo without the CPU in place reveal anything? I'm thinking if the short is in the CPU, the PSU may stay running, but also the absence of the CPU may create a different fault which presents the same way without reealing anything.

2. Would trying all the +ve pins on the main power connector on the MoBo against a support screw with an AVO on resistance setting reveal a short on the MoBo, or is there enough stuff across all the rails to render such an operation pointless? Would doing something like that endanger the board if it was good?

3. If the CPU is damaged, can I risk damaging another machine/MoBo by swapping the potentially faulty CPU with the known good chip in the new host machine?

4. Opposite of 3 really, can I risk damaging a known good chip by placing it into the potentially faulty machine?

I do have a working machine, (this one), which has a 478 chip, but I really don't want to screw this machine up as well.
 

DragonDoc

Distinguished
Jan 26, 2006
149
0
18,680
Okay, so I have taken the board out of the box and had a really good look at it, nothing strikes me as odd. All the electrolytics look normal etc. Similarly, once cleaned of the old Arctic Silver, the processor looks fine.

Reseated, it behaves as before.

Questions.

1. Would trying to power on the MoBo without the CPU in place reveal anything? I'm thinking if the short is in the CPU, the PSU may stay running, but also the absence of the CPU may create a different fault which presents the same way without reealing anything.

2. Would trying all the +ve pins on the main power connector on the MoBo against a support screw with an AVO on resistance setting reveal a short on the MoBo, or is there enough stuff across all the rails to render such an operation pointless? Would doing something like that endanger the board if it was good?

3. If the CPU is damaged, can I risk damaging another machine/MoBo by swapping the potentially faulty CPU with the known good chip in the new host machine?

4. Opposite of 3 really, can I risk damaging a known good chip by placing it into the potentially faulty machine?

I do have a working machine, (this one), which has a 478 chip, but I really don't want to screw this machine up as well.

I'm sure your Power Supply failed (what exactly caused it I cant say - spike maybe). :oops:
When the Power Supply failed it may have taken the board, CPU and memory with it. :cry:
I have seen Power Supplies spark - flames and nothing else damaged, and also seen nothing from a Power Supply fry everything. :roll:

You may want to take it into a shop and have them test the board and CPU. Most shops have that capability. :wink:
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
As I said yesterday...
The PSU from the dead machine was installed in this machine this evening as a test. This machine normally has a 450W Antec in it, it's a much more thirsty rig then the older setup, a hot Prescott, on a bigger Asus, with a lot more in the box. If the 400W can drive this, I think it could drive the older set up as it had been doing since October.
... the original Antec is fine, and a second known good Sparkle PSU did not help.

There are no shops around here, I have to get everything mail order unless I drive into Copenhagen, which is best avoided.
 

Spin

Distinguished
Apr 12, 2004
17
0
18,510
The same thing happened to me. It just died suddenly for no reason while my other comp was fine.

The first indication you gave - fans would come on for a bit and then turn off would suggest a PSU problem. You said the 300W does work.... its been known sometimes the PSU will still power up a rig when there's not enough juice in it still but eventually it'll stop working. And besides, PSUs do die after a while also, like w/ any electronics. Eliminate the easiest possibility first. Put in that 450W and see if it works. If it still fails, then its not the PSU.

2nd indication - NO beeping sounds (considering you have the case speaker hooked up) , this is either the MB or CPU. Mine turned out to be the CPU (AMD XP 1500+ overclocked) that I had for 4 years that burned out. Swapping it out w/ my xp 2500+ fixed the problem. But this was only after I went out and bought a new MB already and when it still didnt work, I was stumped til I found out the curlprit was the cpu. From my experience, cpu's rarely ever die and mobos are much more fragile so I assumed it had to the mobo..........

My lesson? Always test out the easiest parts first before you do anything. Test out the memory 3rd before you do anything to the mobo.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
The 300W Sparkle does work, or at least it did when I took it out of the same box in October, it just doesn't alter the situation. The 400W Antec was installed in this machine, (a much bigger rig), yesterday and ran fine. I do not believe the problem to be with the PSU anymore.

As there is only the MoBo and CPU left, I guess it is one of these.

I don't think the "potentially faulty" MoBo will take a Prescott chip, so I can't put the chip from this machine in there. The only thing I have left is to try the "potentially faulty" Northwood in this machine, but I was wary of that in case doing so risked damage to this rig which I can't afford to loose. That is what prompted my questions above.

The system speaker is still attached - I was trying to get the minimum set up and have the "no memory" beep code. Nada.
 

Sonic_Reducer

Distinguished
Feb 20, 2006
167
0
18,680
well it seems it as a fried mobo just trie and run it without anything connected to it and without any pci or agp just with the cpu to see if beeps, but those power surges can fried it up.
 

adrianxw

Distinguished
Mar 20, 2003
103
0
18,680
Err, as stated above, several places now actually, there is only the PSU, the MoBo, the CPU and the front panel switches/LEDS left in the box.

The PSU is NOT dead, there are no spikes!