Occasional BSODs and pops

[flashback8]

Hey everyone. I'm having a problem with a computer I built a couple of years ago and figured it's time to do something about it.

About two years ago, I built a computer with the following specs (as best I can remember).

-Antec mid-tower case (forget the exact model) + Antec 650W (?) power supply (I forget the exact specs but the latter was definitely a high-watt supply that came recommended by several system builders.)
-Intel Q9450 processor + Arctic Cooling Freezer 7 Pro + Arctic Silver 5 paste
-8 GB RAM (Mushkin DDR2-800 - #996580)
-Two WD 160GB HDs and two 640 GB HDs
-DFI LT X38-T28 mobo
-Vista SP1
-NVidia 9600 GT graphics card
-DWA-556 WiFi card
-On-board sound (Upgraded to an Asus sound card a few months ago, but anyway....)
-Samsung DVD burner
-Three 120mm fans throughout the box + One 40mm fan for the NB chip's heat sink

This was my first time trying to build a computer, although I've swapped out bad parts several times over the years and had no problems. I didn't try to overclock anything or do anything particularly fancy.

After running Memtest86+ overnight (no glitches), I installed Vista. No problems. I did the two-week break-in that I'm told one should do (i.e., run the computer non-stop for two weeks to make sure nothing's bad). All's well.

For about six months, everything was great. One night, there was a random BSOD. I thought that maybe I was just doing too much and didn't push the computer quite as hard for awhile.

(FWIW, I don't recall the exact timeline but the first BSOD might have been shortly after a power outage due to faulty electrical wiring. A roommate said flames shot out of her outlet to boot! In any event, my computer was behind a UPS unit and (cheap) surge protector, and nothing seemed amiss at first, but it's certainly possible something got zapped.)

About a month later, I got another BSOD. Great. I looked up the error code (IRQL-something-or-another) and got nothing definitive.

As time went on, the crashes went from occurring about once a week to about every day or two. I started to suspect I had a bad video card because it seemed like running things like Folding@Home would eventually trigger the crash. I bought an ATi video card (HD5800, I believe) and eventually tossed that in there about six months ago.

No dice. A day later, I got another BSOD. Things slowly got worse. The crashes were becoming more common, the BSOD messages were varying (assuming there were any) and the computer sometimes just stopped cold. I could hear pops every now and again. They always happen during start-up, sometimes accompanied by a crash. Sometimes, the pops accompanied a BSOD. If I played music, it seemed like the computer would crash shortly thereafter too. In any event, it was a miracle if I was able to go longer than an hour without a crash, assuming the computer didn't freeze up during bootup.

Wanting to troubleshoot a little further, I upgraded to Windows 7 last night, but not before running Memtest86+ again for ~3 hours. (No crashes, and no errors reported.) That seemed to help but I just got a BSOD a couple of hours ago, after doing some high def A/V stuff. (Don't know if it's a coincidence but it seems like A/V eventually triggers a crash.) Uggh. Looks like I'm back to square one.

AFAIK, nothing seems overheated. No part of the computer is hot to the touch. A bit warm in places, sure, but nothing that freaks me out.

The inevitable question: what's happening? I've done some reading and have narrowed it down to three culprits, in the order that I'd prefer to try.

-Power supply (There is an occasional smell that isn't quite right, although it's definitely not the distinct smell of a blown cap or whatever. Combine that with the aforementioned surge and pops, and we have a prime suspect.)
-RAM
-MB
-CPU (Highly doubt it)

Thoughts? I'm not in a hurry but I'd like to have a good idea of my next step before I buy random parts.

Thanks for any help you all can provide! I really appreciate it.

 

[Slayn]

I would guess memory, especially based on the BSOD irql thing. What timings/voltages are you running the memory at?

 

[Psycogeec]

a failing power supply would not have to "heat up" to fail when power is needed from your high use things.

the first thing i thought of when reading your post was simple drivers things, as some of those drivers can cause issue like you described, BUT when you re-installed a system over , most non certified drivers are tossed out by microsoft.
(untill you added them back in, which mabey you did)

First thing i would do with your Top SUSPECT being a power supply and strage smells, is voltage test the power supply.
for $10-30 USD you can get the most simple of voltage meters, test the DC voltages on some of the open molexes easily and without much strain you can see reality.
then switch the meter reading to AC (of all things) and read the DC power again, which can show up if the power is "dirty" "pulsing dc" or ac in the line or general fluxuating noises will show on the AC side of the meter.

then i wouldnt make a lot of assumptions, at first, minor differences in voltages are not as critical as it would seem, as most of the motherboard and cards will regulate FROM those voltages, with thier own curcuitry. but if it was wacked off in voltages, or the voltage droops ON various hard tasks (high consumption) , then it probably isnt so good.

get that one item out of the way, have a real voltage meter tool when your finished and then go from there to the other items.

there still exists 2 types of power supplys in this world under many brand names , one that wont last time, and ones that are built to the hilt. i have seen 3 "switching" power supplys fail miserably and one even ketch fire. if your smelling really bad smells comming out of a power supply, then work with that first, without it you really got nothing else

as long as the system is fully operational with a few glitches here and there, i wouldnt fork over more money $$$, for that system, untill i knew more about why it was failing , a Shop might just because time is money. When you can tweak down things (lower cpu), alter timings on memory or busses, and have the time to test , most of the things you listed on a WORKING system can be tested. . . . then you still have drivers

also dont forget it might be 2 MINOR things also, some minor issues in more than one area of need.

good luck.

"surge protectors" do NOTHING for the power untill massive surges come across, some are so useless as to a surge being able to destroy your computer before they kick in thier protection. even a UPS of various cheaper sorts cannot provide awesome wall power when the wall power is going badly. only a "regulated" UPS is capable of totally fixing wall power , and then of course the quality of that still counts. if a High power consumtion Heater (~1500W) or AC (1300-5000W) is slamming your power , a surge protector and most unregulated UPSes will do nothing to completly fix that. although a good computer power supply could ride over it without a problem.

you can usually test for, or realise that is happening when your roommates Air Conditioner kicks on, and the lights dim, and you get a BSOD , the correlation is more likly to be assumed

If you have bad wall power, only a primadonna fully regulated UPS thing can provide you good power from that. and they come in many prices, regulated UPSes are much more reasonably priced and more avilable than 10 years ago.

When you go to test things like ram, for pushing its timings or its buss, or when you got to test a CPU say for its capability, you can always undervoltage and overclock, THEN set it back to a bit below those harsh timings for the hard hitting critical times.
Ex: when you go to test memory, lower the voltage a bit and raise the Speed at which it has to go, then do the memtest, after THAT passes, then revert back to your normal settings.

When you go to test the CPU for example, get the highest point which some TESTING stuff will see it fail, then drop it back a notch from there.

Not all testing junk , even prime or furmark, tests the reality of all possible happenings in the system, and most dont test it together. for example a passed furmark tested video card may completly fail when pushed with a ATItool (type) test.
a Reality of both processors AND the video may fail when both are slammed in some momentary reality , instead of a test.

It isnt stable, untill you can do everything all the time without any failures ever for real, anything else is cute benchmarks/stability tests, which can Vary greatly as to thier ability to simulate reality, awesome tools to test, but not definitive stability answers for OCing.

Find the MAX, then notch down from "excess" at least when you have big problems , is a way to survive more operational realities when pushing things.
the cool thing about Overclockers , you dont have to overclock to push things harder to test them, even if you run a system at normal speeds, cranking them up to find thier weak points , before turning it down can be so usefull.

 

[psionic98]

the smell may not really be the psu though unless you know for sure the smell is coming from the psu.. your mobo heats up under load and can also blow something that can smell.. although I am with psyco and say start with the psu and work your way into more expensive items.. if you have a spare psu laying around id start there to save some cash.. though for like 70 or so you can get a decent corsair unit that should last for quite some time.. antecs are ok, but they are hit or miss as well..

now im not sure if I missed the "pop" part, but does it pop every time it BSOD's? I saw you said you had the surge and you were behind a ups and cheap bar.. the bar+ups "should" have saved you... but are you still running the system through the same bar+ups? If so .. take them out the loop as well.. even if the power from the wall isnt totally clean, the 2 items could also have an issue.. though likely the ups should catch or trigger if the bar/socket has some issues... id just take them out in case the ups is screwed and perhaps its not catching surges anymore

from there.. if the psu is good and the other stuff is out of the way i'd go with tearing down the rig and looking over the mobo and ALL connection points to the chasis and power connectors/wires If the surge made it to the system but didnt harm anything specific you could have some issues with frays or if not popped caps, expanded ones that are on teh verge

 

[Psycogeec]

we also didnt notice a lot of temperatures or voltages in the OPs post.
many of the overclockers would have an Eye on the temps , the fans and the voltages ,(at least) WHEN some strange stuff was going on.

i wonder what flashback8 is using to see anything in the hardware ??

 

[wingman99]

Try removing all your memory except for one stick and see if you have BSD and also swap that stick and test again.

 

[flashback8]

Hey everyone. Thanks for the replies. I appreciate it. I'll try to reply to everything here.

The IRQL thing was the most common error I saw but I did see others. That said, who knows if they were related.

Another thing I forgot to mention was that, by the time Vista was really dying, programs would sometimes seem to corrupt files I was working on at the time. If, say, I was trying to demux the audio from a file, the demuxer would say the audio was corrupt and then say it was fine the next time I tried to demux. This never happened until near the end of Vista's life; the software was rock solid until near the end. One way or another, I'd imagine that would imply memory.

I ran hwmonitor and got the following voltage values. The values punched into the BIOS are in parentheses.

CPU Vcore - 1.17 V (1.23 V)
FSB VTT - 1.10 V (1.10 V)
NB Core - 1.34 V (1.39 V)
DDR - 1.87 V (1.90 V)

Here are the temps too. This is while the system's relatively idle. I doubt the temps are particularly accurate, but hey, while the program's open....

CPU cores - 50° C (Cores 1-3), 60° C (Core 0)
GPU - 43° C
Hard drives - 37° C

The first core seems a bit hot but I thought that was still well within the acceptable range. The rest seems fine.

The memory timings appear to be on auto in the BIOS. I believe they're supposed to be 5-5-5-18 stock. Oh, and in case I didn't state it the first time around, I have four 2GB sticks.

Do I need to report anything else? Just checking.

I don't have a meter but will look into getting one. I don't have much time right now to take measurements but will get to it sooner or later. What I might do first is do what one person said and just check one memory stick. It certainly could be the PSU but I'd think that, if it were, there would be some sort of consistent indicator, or smell, or whatever. (Believe me, I'm quite familiar with the smell of busted PSUs!)

Regarding the pops, crashes don't always accompany them. "Pop crashes" are kind of rare, to be honest. None yet under 7, knock on wood.

Thanks again!

EDIT: I checked Mushkin's web site. Looks like they recommend 4-4-4-12 timing and 2.0-2.1V. D'oh! Time to play with the BIOS.

 

[Slayn]

I would raise the voltage and timings. The voltage at 2.1v is good. The timings at 4-4-4-12 might be hard to pull off. I would leave them at the "stock" 5-5-5-18, that seems doable. Lowering the timings ( thus making it faster ) will not help.

2.1v 5-5-5-18 or 6-6-6-18 IMO.

 

[Psycogeec]

Another thing I forgot to mention was that, by the time Vista was really dying, programs would sometimes seem to corrupt files I was working on at the time. If, say, I was trying to demux the audio from a file, the demuxer would say the audio was corrupt and then say it was fine the next time I tried to demux. This never happened until near the end of Vista's life; the software was rock solid until near the end. One way or another, I'd imagine that would imply memory.

.

Yes you more implied memory or low possibility cpu , minor data inacuracies comming out .

 

[OnDborder]

Welcome to the Forum! http://www.ocforums.com/showthread.php?t=491697 flashback8
Ditto on raising the ram and nb voltages. cpu only if oc'n.
Seeing that you are using a dfi x38 board, have you ever replaced the thermal conductive material under the northbridge hs? If not, I would recommend you doing that with Artic Silver 5.

 

[RJARRRPCGP]

I looked up the error code (IRQL-something-or-another)

Your processor is unstable, AFAIK. I got that one when Linpack failed.

And I used to get that one with my T-Bird pretty much all the time, during 2002.
(required to have the core frequency lower than 1050 mhz on my T-Bird 900) (back when overclocking with multiplier only!)
(so I knew it was related to the chip)

 

[wingman99]

I would say cpu or memory or motherboard, not PSU because that would effect your system all the time.

I would run prime 95 and mem86.

What new parts have you purchased for your pc?