Random BSODs

[DanF]

I have a Toshiba notebook, running Windows 7. All of a sudden it started to get random BSODs (random error codes).

I've re-installed Windows 7 from the recovery partition. It was installing the drivers and Toshiba apps when the first BSOD after a fresh installation occurred. I thought, it's time to test some hardware.

Ran Memtest for like 24 hours with no fails at all.
Did an extended test on HDD and no errors. I've used Parted Magic totally on RAM and ran smoothly as ever.

Also:

Tried to install Windows from an OEM disc. BSOD occurs during installation
Tried to install Windows with each RAM installed at a time on different slots.
BIOS settings reset to default just in case something was messed up.

Also noticed: sometimes screen turns grey with artifacts. Not sure if it's related to the screen, the cable or if it's the VGA (which might also be related to the BSOD then?).

What are your thoughts?
Thanks

 

[Larry Sabo]

What does Bluescreen View (offline) say?
What happens if you try with a different HDD and known good RAM?
How dirty are the CPU fan intakes/exhaust rad?
Have you tried it bare MB? If so, does the GPU get much hotter than the exhaust rad?
Tried a live Linux CD, monitoring temps?

 

[DanF]

Can't run Bluescreen View, as it won't even let me finish the OS installation.
I haven't tried with a different HDD yet, but have tried 4 different memory sticks.
No excessive dirt is present.
Bare mobo... do you mean without WLAN card for instance?
Linux CD works fine, temperatures are all between 45 - 55 C.

 

[lcoughey]

Did you get a full sector-by-sector backup of the drive? Even though the extended test past, I still suspect the hard drive. And seeing that the data on it is the only part of the system you cannot replace...it is best to be sure that it is 100% backed up.

 

[DanF]

Data was manually selected and backed up beforehand.

EDIT: Trying a different HDD as we speak.

 

[Boston Pro Comp Serv LLC]

What are you using to test the drive? Is the smart data available? Booting to Linux live disc is a good suggestion as well. I'm curious as to why you didn't test the drive or ram, or investigate the BSOD prior to the attempted reinstall.

 

[DanF]

I didn't bother with the BSODs, as they were too random. Sometimes they used to hint an AVAST file, sometimes a Windows system file (and almost all the time a different file). During installation these would almost always have a different error code.

I used Parted Magic to test the drive and SMART data looked fine to me.
RAM tested (both using Memtest as well as using different physical modules).

 

[DanF]

Different HDD encountered a BSOD as well. Will take notesof the BSODs.

1: 0x00000124 (during installation)
2: 0x0000003B (win32k.sys during Installation)

 

[Larry Sabo]

Can't run Bluescreen View, as it won't even let me finish the OS installation.
I haven't tried with a different HDD yet, but have tried 4 different memory sticks.
No excessive dirt is present.
Bare mobo... do you mean without WLAN card for instance?
Linux CD works fine, temperatures are all between 45 - 55 C.

I suggested doing it offline, i.e., with the drive slaved to another PC. By bare MB, I meant out of case and with no ODD or wireless installed, just to eliminate as many variables as possible. I was going to suggest touch-testing the CPU temp versus the exhaust rad temp in case it's a defective heat pipe, but temps are cool so that's irrelevant. Also, sometimes by carefully inspecting the MB, you can find hot spots where some component has been overheating, errant case screws shorting, etc. What's the model? If it's not recent, it may have those flaky NEC caps under the CPU. Given your test results, I would call it a bad MB, myself. Can't think of anything else to do/check, except possibly swapping the CPU, which virtually never fails.

 

[DanF]

It's a Toshiba satellite L350-277

 

[Larry Sabo]

In that case,99% sure it's flaky NEC TOKIN capacitors.

 

[Markverhyden]

I agree, probably caps. I'm going to a customer site now but when I come back I think I may have the service manual and Toshiba Test for that series.

 

[Markverhyden]

test and diag iso - http://verhyden.org/owncloud/public.php?service=files&t=e9cfbbb07ce2058e182bbff169ed074f

service manual - http://verhyden.org/owncloud/public.php?service=files&t=e9cfbbb07ce2058e182bbff169ed074f

 

[DanF]

Thanks. Using the test and diagnostic tool right now.

 

[DanF]

All tests passed. Do you still think it's the NEK TOKIN?

Since it's quite an old system, client might decide to simply get another laptop, but if you still think it's the NEK TOKIN capacitor I might ask the client if I can give it a go.

 

[Larry Sabo]

As far as I know, the only way to tell is to remove the MB and look under the CPU for a large rectangular cap, like in the link I provided earlier. It should say NEC TOKIN right on it, IIRC. If you want to repair it, you need to heat the cap and MB really well before it will lift. Not a problem with a BGA rework station but with a hot air wand alone, it will take quite a while. Most techs will replace it with 4 x 330uFd caps, which you can pull off old MBs or order from Digi-kKey, Mouser, etc. I wouldn't install another NEC TOKIN, because of their unreliability.

Edit: If you look under the CPU and it's not a NEC TOKIN cap, then the cause is something else entirely.

 

[Markverhyden]

As we all know these D&T programs have limited capabilities. Such as they have no way to test individual electronic components. As I mentioned earlier I'm with Larry and it might be caps or something else. Given the age of the computer I'd point them to a new box. You have too many strange things happening to guarantee a component level repair.

 

[FremontPC]

You have too many strange things happening to guarantee a component level repair.

Agree. The results are just too whacky. If this thing had a Nvidia GPU, it would have been in the right time slot for that whole debacle... I'm surprised it happened with Intel graphics. Maybe just the odd lemon or like Mark and Larry said, bad caps that don't trip over the factory diags but can't handle Windows Setup.

Well, 5 or so years out of a cheap Pentium laptop, not so bad as a poke in the eye with a burnt stick.

 

[DanF]

I kept this system and gave it a try in any case. But... when I removed the Nec Tokin, this is what I got. Does it look right? Are the connections on the edges the positive and the middle section the negative? Or should there be 2 x Negative connections? Maybe the middle part is still a piece from the Nec Tokin cap? Can't seem to be able to remove it.

 

[Larry Sabo]

The middle part is the ground plane--don't try to remove it! You connect the replacement caps with their negative ends contacting ground plane and the positive ends attached to the outer rails. The two outer rails are actually connected, so 4 caps @ 330uFd will provide capacitance equal to that of the NEC TOKIN. I don't have a picture of how they look installed, but will update the post if I find one. Here is how the one I replaced looked before the replacement: http://prntscr.com/95nrg1.