System troubleshooting question

[Opus4]

I've finally run into a PC issue where I'm not 100% sure what the problem is and was wondering if someone might know something to suggest that I've missed:

I have an XP system that simply freezes, sometimes while Windows is starting, sometimes after it has been running for a while. This started happening after a reboot & Windows said it found new hardware, too much had changed, had to be reactivated, etc. Memory and hard drives test out OK, so I was suspecting the motherboard or cpu & figured I would be building a new system. I haven't tried replacing the power supply, but I'm not sure if that would cause system freezes.

I had last booted into Safe mode & left it sitting -- it has run for a couple days that way w/o freezing. Immediately after restarting Windows normally, though, it froze again. (I haven't had time to do much more troubleshooting or research on new parts, so I just left the system sitting in Safe mode a couple days ago out of curiosity.)

I still think there is hardware that failed & perhaps isn't getting used while in Safe mode, but I wanted to see if anyone who knows more about troubleshooting systems had a better idea.

Thanks.

- Andy

[MeInMaui]

I would try doing some more rigorous testing of the memory first. It could also be that your power supply is marginal. You may not be drawing quite as much in safe mode so that the problem doesn't present itself.

One other thought is that you could maybe try booting from a linux CD and try running some benchmarking tools or a stress test to exercise your system. I should think that would reveal any hardware problems and decouple the testing from your installed windows OS.

Aloha,
Mike

[stuckless]

I had a similar issue a couple weeks ago with my laptop (running ubuntu linux). I booted with the Ubuntu install cd, and did a memory test... it immediately came back with bad memory areas. I replaced the bad memory chip and i've been good so far (cross my fingers ) btw... if you are going to do a memory test, you should probably just do it one stick at a time... if not, then it may be hard to tell which stick is bad :0

[Opus4]

Originally Posted by MeInMaui

I would try doing some more rigorous testing of the memory first. It could also be that your power supply is marginal. You may not be drawing quite as much in safe mode so that the problem doesn't present itself.

I was really leaning towards the motherboard or power supply because of the strange thing I mentioned where Windows thought I had a bunch of new hardware after a reboot. But, the fact that safe mode was running made me rethink & wonder if it was Windows after all or whether safe mode just wasn't hitting the problem hardware.

I don't know how to test a power supply, so I was thinking of ordering that first & see what happens if I change it.

One other thought is that you could maybe try booting from a linux CD and try running some benchmarking tools or a stress test to exercise your system. I should think that would reveal any hardware problems and decouple the testing from your installed windows OS.

I've got some testing tools that boot from a floppy, but that sounds better, so I should do some google searches for that.

Originally Posted by stuckless

I had a similar issue a couple weeks ago with my laptop (running ubuntu linux). I booted with the Ubuntu install cd, and did a memory test... it immediately came back with bad memory areas. I replaced the bad memory chip and i've been good so far (cross my fingers ) btw... if you are going to do a memory test, you should probably just do it one stick at a time... if not, then it may be hard to tell which stick is bad :0

Either all my memory is bad or it isn't the memory -- Windows freezes on any of it, but a bootable memory test utility says it is all fine. That's why I'm guessing it isn't the memory. Although, with all the things breaking around my house over the last month, I won't say I can't have 4 sticks of memory go bad at once! Every time I have someone come to the house to repair something, I hear "I've never seen that happen before" or "That doesn't usually happen".

Thanks for the comments so far; I'll look for a bootable test CD next.

- Andy

[techhome]

I'm in a similar boat with am Athlon X2 system. You may want to start with Memtest 86+ which will rigorously test the memory subsystem. Download memtest as a CD ISO and burn the DOS boot disk. Reboot the PC and it will boot into DOS and automatically start running the entire test suite in a continuous loop.

In my case, the PC which is experiencing pretty much exactly the same freezing symptoms, ran memtest for 18 hours error free. Hence, I believe the CPU and memory is fine and that the motherboard has failed. I'll get to test my diagnosis out next week when the replacement motherboard arrives.

If you wish to troubleshoot further, maybe try Sandra 2010.

Doug

[KarylFStein]

Late, but I echo the above: run memtest86+ first then swap out the PSU. I've had two similar issues.

In one case, the fix was to increase the voltage to the RAM to 1.9. It was plainly put on the sticker as 1.9V, but I thought--based incorrectly on the NewEgg reviews--that you only needed to increase voltage if you overclocked the RAM to the lower timings and not if you kept with the defaults in the SPD. I don't usually touch the "auto" voltage/clock setting in the BIOS for CPU or RAM as I'll take stability over perceived performance, but the failure with memtest86+ made me focus more on the RAM. Those sticks are now gone in favor of some "vendor approved" ones. They're running at lower timings, but I can't tell any difference except I haven't had any weird failures in about a year now.

The other issue turned out to be a failing capacitor on my MB. (I only realized that after having to fix my HD100 and realizing what to look for there ). So, not PSU related, but I've read about PSU issues causing unexplainable issues like you describe. At worst, you've got a good-quality PSU as a spare. At best it saves you the higher expense of swapping out other parts.

[Striker:WG]

memtest is an excellent program for memory testing, i've diagnosed numerous issues with it.

you can also do a torture test with Prime95 to test your system stability when the CPU and memory is being heavily taxed.

check your motherboard for any blown or bulging caps, that will be the most likely thing to go on the motherboard.

as for the power supply, your best bet would be to monitor the voltage rails with a multimeter and check and see if there are any major swings in the power lines when you are doing tasks on the PC.

-Striker-

[KarylFStein]

Fast...slipped that you tried the memory test already.

[Opus4]

Yeah, memtest was the memory test utility I used. I'll be looking into other utilities tomorrow. I finally have time to look into parts more this weekend, so I might as well run some more tests too.

- Andy

[Fuzzy]

Of all the component failures I've had over the years, Power Supplies have been number one. After that, it's been motherboards. I've never had a CPU fail, or memory, and only ever had one HDD die (i think they've gotten too small before the reached end of life). I've had a single video card die, but that was because the fan died, and it cooked.

Basically, try replacing the power supply first... If that doens't fix it, you aren't out much having a spare one around. Trust me, it'll get used.

[robogeek]

Any bulging or leaking capacitors on the motherboard? That's usually the culprit if the motherboard itself is the cause of random freezes, especially if it's an older motherboard or been heavily used 24/7. Otherwise, as others have suggested, swap out the power supply with a known working one to see if the problem goes away. Does the BIOS have a health monitor that gives realtime voltage, fan speed, and temp readings? If so, look for any voltage values that are way out of spec or any that are fluctuating significantly. Any values way out of spec or fluctuating could indicate bad power supply or bad capacitors on the motherboard.

But, since XP boots and runs without freezing in safe-mode, it could be a bad device driver or possibly malware or rootkit loading with the device drivers. Since safe-mode uses some generic drivers, I would look at any audio or video cards as possible causes, and any other addon cards and usb/firewire devices plugged in to the motherboard. If the audio or video hardware (or other addon cards) are failing, then usually bulging/leaking capacitors will be found somewhere on cards or the motherboard if the audio or video are integrated on the motherboard...unless the damage was done by a power surge or static discharge.

[Opus4]

A couple of the other test utilities mentioned above run in Windows, so they are out since Windows won't run. I downloaded the Ubuntu install CD. It had an option to test the memory, but that turned out to be memtest 86, which I already ran several times. I tried booting into Linux to see what other test tools it had, but that failed to run too.

There are 4 capacitors next to the cpu that are not as flat-topped as all the other capacitors, so either they are just different types or they are the problem. One had a glop of brown stuff on top of it, so that one didn't seem to just be different.

So... thanks for the all suggestions. It appears to me that it really is the hardware, not just Windows, which is what I was trying to determine.

- Andy

[Opus4]

As I mentioned in my last post, all the comments about bulging capacitors seem to have been on the mark. And, when I look at my office PC, I see 2 capacitors next to the cpu that are not as flat as the ones around them, so that might be starting to go too. It hasn't started acting up yet, but I was already planning to replace it later this year. I suppose I could look into replacing the capacitors on the boards, but I'm not sure if they've affected anything else on the motherboard.

I ordered overkill to replace my failed HTPC, with an Intel i7 920, but I use it for more than SageTV, try to keep my PCs for years, and am seriously considering building the same system for a new office PC too and getting a backup motherboard in case one of those dies somewhere down the road. Even if I skip the backup mb, one system could replace the other temporarily.

- Andy

[null_pointer_us]

Glad you found the issue, but I'd like to add something to the thread anyway for posterity.

Do not test multiple sticks of RAM simultaneously. Test them one at a time.

While troubleshooting a similar problem, memtest86+ ran fine overnight with all four sticks. But the issues persisted. When I tested the sticks individually, the third stick would not even allow the machine to POST.

[Opus4]

I've always tested them in pairs, changing the slots they are installed into and sometimes changing the paired combinations if there are more than 2 sticks being tested.

I might be wrong, but I thought there was some portion of a stick that a memory test program couldn't include in the test (the app has to run somewhere). I've had pairs that passed when installed one way but failed when reversed. Maybe that was coincidence & someone else can say swapping them around like that is a waste of time.

- Andy

[davephan]

If you had an image some time ago when the system was stable, then you could recover to that image and find out if that fixes the problem (before you roll it back take a current image, so you can roll it back after the test). Since you mentioned it is stable in safe mode, then it could be a software problem.

Sometimes we have unstable systems work that are tough to diagnose. I've applied the unstable system's image to a known good identical system and/or a known good image to the unstable system do help determine if the problem is a software or hardware problem. Sometimes it is a software problem and sometimes it is a hardware problem.

I've seen both power supplies and system boards cause instability, rarely its a RAM problem, never a CPU problem (not yet, after many years). Between the power supplies and system boards, usually it is the system boards in servers, probably because the server power supplies are much higher quality than workstation power supplies. I have less experience with workstations than servers, though I've seen more trouble with power supplies than system boards in workstations.

Dave

[Opus4]

Originally Posted by davephan

If you had an image some time ago when the system was stable, then you could recover to that image and find out if that fixes the problem (before you roll it back take a current image, so you can roll it back after the test). Since you mentioned it is stable in safe mode, then it could be a software problem.

Thanks, but I've given up on it. I understated (just my style of writing, I suppose) the capacitor problem I noticed after several people suggested checking them -- one had stuff leaking out & a few more were bulging (the capacitors, not the people). I do create disk images periodically and have near daily backups, btw, but it fails outside of Windows now too, so nothing's getting re-imaged or restored on it now.

I've seen both power supplies and system boards cause instability, rarely its a RAM problem, never a CPU problem (not yet, after many years). Between the power supplies and system boards, usually it is the system boards in servers, probably because the server power supplies are much higher quality than workstation power supplies.

So far I've had just about everything fail at some point -- power supply, memory, many hard drives, video cards, monitor, fans, half my CD drives don't want to open (perhaps too much dust build up after several years?). But, this is my first motherboard failure.

So... the new parts have been shipped, I've created an XP Pro SP3 slipstream install CD, and am just waiting for the parts to arrive. Newegg often ships from practically around the corner, but this time they chose to ship from CA. It is still scheduled to get here before next weekend, though.

- Andy

[Opus4]

It turns out that the power supply and motherboard are both bad. While ordering parts, I tossed in a ps tester -- it showed that some of the voltages on the old ps were fairly variable, going a little above & below where they should have been. The new ps showed stable readings. After swapping the power supplies, though, the system still freezes (not that I really expected anything different based on the capacitors, but I had to check).

- Andy

[davephan]

Originally Posted by Opus4

It turns out that the power supply and motherboard are both bad. While ordering parts, I tossed in a ps tester -- it showed that some of the voltages on the old ps were fairly variable, going a little above & below where they should have been. The new ps showed stable readings. After swapping the power supplies, though, the system still freezes (not that I really expected anything different based on the capacitors, but I had to check).

- Andy

Is the power supply tester hardware or free software? If it is free software, do you have a link or name for it? I just RMA'ed a bad power supply.

I bought my first home computer, a Commodore VIC 20 in about 1980, so it's been about 30 years owning different home computers. Since then, I've owned a lot of computers over that time period. Thinking back, I don't remember very many where the system boards and/or power supplies lasted over five years. Although when they get to be five years old, they are generally obsolete anyway. Computers generally just don't last very many years, not like some cars that will run for 20 years, computers just have to be replaced every few years.

Dave

[Opus4]

Originally Posted by davephan

Is the power supply tester hardware or free software? If it is free software, do you have a link or name for it? I just RMA'ed a bad power supply.

It is hardware -- when buying everything, I searched newegg for power supply testers & picked one. I think it was around $25.

I bought my first home computer, a Commodore VIC 20 in about 1980, so it's been about 30 years owning different home computers. Since then, I've owned a lot of computers over that time period. Thinking back, I don't remember very many where the system boards and/or power supplies lasted over five years. Although when they get to be five years old, they are generally obsolete anyway. Computers generally just don't last very many years, not like some cars that will run for 20 years, computers just have to be replaced every few years.

I try to hold onto things: I still use my Carver amp & preamp I bought in 1986. (Obviously I haven't cared about surround sound so far...) My last TV was ~16 years old & still worked when I finally switched to HDTV. And, my car was 18 when it got replaced because it had no rear shoulder straps for baby seats. (I guess the car was obsolete!)

But, yes, I agree the PC is fairly old at 5.5 years, so I'm not really upset about replacing it; it is just that the thing was still plenty fast enough for SageTV. Before that, I was using a P2-400 for SageTV; it was 'retired' due to incredible slowness at 6 years old, though it still ran fine. (Wow, I just realized that the thing might still need to be taken to a recycling place. Have I mentioned yet that I'm a procrastinator??)

- Andy