Baffling computer problem

[kb2vxa]

Hi guys,

I drew a blank, my guru who built this beat drew a blank and a very knowledgeable friend down under I call the Wizard of Oz drew a blank so lets see what ideas you come up with.

It all started with a grey blank screen, nothing worked so the only way out was to switch off the power strip the computer and peripherals are plugged into and start from scratch. BTW I always switch off mains power when the machine is not in use. Switching it back on and pressing the power up/down button it refused to boot, fans came on but otherwise nothing. After fighting with it finally the BIOS boot screen came up and to continue I had to step it to exit, then down to exit and disregard changes, then on the next screen "start Windows normally" and from that point onward everything worked... until next time.

I ran MemTest86 and being an intermittent problem naturally everything is OK, same with other diagnostics all showing nothing wrong. There's no appreciable dust inside, all the electrolytics LOOK good with no oozy stuff or bulging and the trouble shows up most often after everything has been completely off all night (no mains power) so it's not heat related. What it amounts to is the fans start up but no boot, just a dark screen sitting there looking at me all kinds of stupid. After fighting with it for up to 20 minutes switching the power strip off and on and pressing the power up/down button it finally decides to cooperate, even starting normally each and every time... until next time. Oh, after running self diagnostics in Action Center it called home reporting a BSOD event (WTF?) and as always the Microstupid robot was clueless.

Now you Windblows hating Tux fans shuttuppa you face! There's nothing wrong with Win 7 Ultimate, it won't BOOT so it's an intermittent HARDWARE problem OK? (;->) So now that I'm clueless, my guru is clueless and the Wizard is clueless, let's see what putting some heads together can come up with... pweez?

[W3WN]

I've seen things like this happen before.

Possible that you may have an intermittant problem with the power supply.
Possible that you have a wonky memory stick.
Possible that the video card or subsystem is giving up the ghost.
Possible that there's a chip starting to fail on the motherboard.

I've found all 3 as causes on different machines over the years for this or similar issues. I have a Dell Optiplex out with the issue, but it's a motherf***gboard problem.

Try swapping the PS, memory, video (one at a time, of course) and hope that it's not the motherf***g board.

[ad4mg]

Try pulling the RAM off the MB, blowing any dust out of the slots, and re-seating the RAM strips. If you have no luck, and have 2 or more strips of RAM in the machine, only install 1 strip in the first slot and try to start the machine. If it starts OK on 1 strip, add the next one. You may have a bad strip of RAM in there, or the strips may simply need to be re-seated.

I can't count how many times I've pulled the RAM from machines that won't boot, Simply cleaning the slots and re-seating the RAM has done the trick for me many times. You may also want to pull any expansion cards and do the same there, especially if your video card is a PCIE card. Last issue I ran into with an expansion card was a sound card in a 1X PCIE slot. Damn thing wobbled around in the slot like a wounded fish out of water. Threw it in the trash, turned the on-board audio controller back on, and all was well.

I'm betting on something to do with those memory strips.

[KK4AMI]

I'm no genius at computer repair, but it sounds like its time to do a "remove and replace" starting with a spare power supply and working your way to the mother board memory. How old is your platorm? Have you replaced the MB BIOS battery recently?

[KC2UGV]

Graphics card. If it's onboard, get a new one, and disable the onboard one. If it's not onboard, replace it or try the onboard one. 90% of random issues like this is the GPU.

[kb2vxa]

The blank grey screen that diagnostics called a BSOD event only happened once, after that the computer won't boot. Since the graphics card isn't involved with the boot process we can eliminate it. I replaced the BIOS battery a few weeks ago, of course settings went to default so I had to change them accordingly, eliminate the battery.

"Possible that you may have an intermittant problem with the power supply."
Agreed, but I don't have another to go the one by one substitution route I took in my TV repair daze.

"Possible that you have a wonky memory stick."
A very good possibility, but since MemTest86 turned up nothing it would be intermittent and I have no replacements there either.

"Possible that the video card or subsystem is giving up the ghost."
Not involved in boot, eliminated.

"Possible that there's a chip starting to fail on the motherboard."
The chipset is another possibility, in that case I'm forked, can't afford a new mobo.

"Try pulling the RAM off the MB, blowing any dust out of the slots, and re-seating the RAM strips."
That's the next logical step, but lacking a grounded mat and wrist strap I'll do it VEEERRRY carefully. While I'm at it I'll do the same with whatever cards are in there, got nothing to lose but (what's left of) my mind.

"I'm betting on something to do with those memory strips."
Methinks you're right, since it's a rainy day with less chance of ESD I'll get on it momentarily. If you don't hear from me something went wrong.

You'll find me crying in the chapel
The tears I shed are tears of pain
I know the meaning of failure
Now I bang my head against the wall

[HUGH]

I got through a couple of PSUs in the last 18 months, presumably the fans run off 5V, this used to be the reference for the other voltages. Can you measure what's coming off it to eliminate the PSU?

[KC2UGV]

The blank grey screen that diagnostics called a BSOD event only happened once, after that the computer won't boot. Since the graphics card isn't involved with the boot process we can eliminate it....

Really? Where do POST results get sent to again?

[n6hcm]

Really? Where do POST results get sent to again?

if it's really a graphics *card* then POST results go to the built-in video.

[KC2UGV]

if it's really a graphics *card* then POST results go to the built-in video.

Not if there's an offboard GPU... If there's a card installed, the PC will insist on using it. But hey, whadda I know? I only did the PC tech thing professionally for about 4 years or so...

[kb2vxa]

OH STOP IT! If you were paying attention when I said IT WON'T BOOT you wouldn't be arguing with me or each other. Anyway thanks for the help, when I reseated the RAM it crapped out entirely and most likely because flexing the mobo broke the intermittent connection altogether. Now I'm thinking of when I re-soldered a cold joint on a board that crapped out after 20 years in hard service but I digress. That pretty much narrows it down to the mobo, it died and went to meet its maker, Scott W2SJW so now if I get REALLY lucky he can come up with an exact replacement which for obvious reasons will save a LOT of trouble. He gave me a loaner so I won't go entirely out of my mind, I need just a little sanity to bring you my insanity. So in my usual fashion stealing a line from The Worst Band In The World by 10cc "but we're working on it". Eh, life is a minestrone.

[NA4BH]

It sounds stupid and everyone of the real computer people here will disagree but check your CMOS battery. Some of the equipment I work on has similar problems, the computer brains with my company determined it was the battery. Problem solved.

[n2ize]

How old is the system ? Sounds like it's time to get a new computer. Particularly if its a mobo problem. And, in particular, If it's an old single core 32 bit system I would suggest that you take it to the nearest swamp and reef it and buy yourself a new 64 bit multicore system.

[K0RGR]

Not much help now if you can't boot it, but what was the BSOD code? The BSOD meant that Windows saw something it didn't like - a code should have been recorded if it had access to the hard disk. You can then Google the code for clues. If nothing was recorded, that means that Windows wasn't able to access the drive - certainly some kind of hardware or BIOS issue. The errors will be recorded in the Event Viewer Logs. If you see intermittent errors pertaining to things like disk drives, that may be a clue.

[KC2UGV]

It sounds stupid and everyone of the real computer people here will disagree but check your CMOS battery. Some of the equipment I work on has similar problems, the computer brains with my company determined it was the battery. Problem solved.

Another viable solution. Fail to boot at all means very basic BIOS functions cannot load. It's either:
* Cannot load initial instruction address (CMOS battery, or corrupted PROM)
* Cannot load the GPU jump instruction (GPU problem)
* No power (Obvious, see lights on the board)
* Physical defect in the mobo (If none of the above resolve it, this is it).

In my trouble-shooting experience, swap out GPU. Then, swap CMOS battery. Then, try a bare metal CMOS flash. Check power supply. None of these work? Chuck the machine, and buy a new one. It's nigh impossible to repair motherboards reliably.

[kb2vxa]

I guess you guys are just swapping ideas, Scott and I had the beast diagnosed two weeks ago and he started ordering parts, an exact replacement mobo and new CPU cooler since disturbing a liquid cooler usually damages it. He took it to his shop last Sunday and for some unknown reason the C drive connector came apart when he removed it so he ordered an exact replacement and luckily was able to clone it to a USB drive so nothing was lost. The new cooler is a new design (for me at least) with a radiator that looks like one for a liquid cooler but is an air cooler, all I have to do to clean it is remove the fan and brush it off. The beauty of it is since it's all OEM parts, the OM being Altec Communications (Scott's company) no reinstalling anything, just putting the data back on C drive and away she goes. Last but not least an upgrade, disabling that POS Micro$hit Security Essentials and installing Symantec Endpoint Protection. It won't be long before I have my baby back, I can return this loaner with... ugh... Windblows 8.1 <gag> <choke> <RALF> on it and air the stink out of the room.

[n2ize]

In my trouble-shooting experience, swap out GPU. Then, swap CMOS battery. Then, try a bare metal CMOS flash. Check power supply. None of these work? Chuck the machine, and buy a new one. It's nigh impossible to repair motherboards reliably.

Actually it is possible to repair MOBO's. I know people that have done it. Power supplies can also be easily repaired. Besides, nothing is impossible for us AM'ers.

[KC2UGV]

Actually it is possible to repair MOBO's. I know people that have done it. Power supplies can also be easily repaired. Besides, nothing is impossible for us AM'ers.

Sure, anything is possible... I've hardly ever seen it aside from power connectors.

[W3WN]

Actually it is possible to repair MOBO's. I know people that have done it. Power supplies can also be easily repaired. Besides, nothing is impossible for us AM'ers.
Want one?

No, seriously. Something on the MFB blew out on my Dell 320 machine a few weeks back. It's being replaced by a Dell 760 I got from work (we just replaced 130 Dell 760's with Lenovo machines running < blecch > Windows 8.1 < /blecch > ) this weekend, and I hate on general principles to just trash it. You're welcome to it, as is. (No drives, they're going into the new machines, but other than that... )