Computer Dead -- Need to Perform an Autopsy

CycloneRanger

My machine seems to have died. Everything was humming along nicely when it suddenly powered off; I was playing a game at the time. It immediately tried to power-on again, but nothing ever made it to the monitor. I turned it off and looked for anything obviously hot inside, but found no clues. I tried starting up again and got nowhere--some noises, fans running, lights on the motherboard, etc., but nothing on the screen.

I pulled out the RAM and tried again--no change. On feeling around I did notice that the CPU heatsink was *really* hot even after only a few seconds of power (it had been quite cool before). There's obviously a lot of power getting dissipated there and I can't imagine that's a good thing.

So, what is my next step here? I'm able to post from an old laptop (with a broken power cable and a logic board fault--this thing won't last much longer) but I really need to get my desktop working again. This is all fairly new gear (around a year; purchased all components aside from the optical drives and peripherals from Newegg) and relatively high-powered. How do I diagnose this further?

One last thing--I have suspected for a while that I have faulty RAM, but that does not seem a likely explanation here. I'd love to be wrong on this.

cmsamo

Does your motherboard have an LED display on it? Some have a place where they show an error code. If stuff is at least getting power, that's a good sign. If you can see a couple of digits on the Mobo, then google for them on the Mobo manufacturer's website.

Other than that, it's hard to diagnose. If the mobo is powering up, and the hard drive is spinning up, then it could be your graphics card has died? I assume from your post that you are using a graphics card and not onboard graphics. Does the mobo have onboard graphics you could try, to replace the 3D card?

Sometimes if you have faulty ram it just causes bluescreens, rather than stopping a PC from posting up at all.

CycloneRanger

The LEDs power on and all the fans turn on--plus, there's the CPU heating up. It seems unlikely that there's a power supply problem here. My PSU is also rated for a higher power output than I need for this system, so its sudden failure is less likely.

I'll try powering the machine on without the graphics card, but I'd be surprised if that were the issue. I suspect at this point it is either the processor or the motherboard and I'm not sure how to determine which is failing.

I've also had trouble starting up for many months--powering my system on after it has been shut down for any length of time would cause the system to start up but hang before reaching the login screen; repeated attempts were necessary to get it to start. I attributed this to bad RAM (there's a similar known issue in some units of this model of RAM), but it might be related to this new problem.

Raern

My computer was having weird start-up issues for a while. Finally solved it when it failed to boot entirely and I was forced to diagnose a dead power supply... on a system that didn't come particularly close to using the full output of that supply. Don't be too quick to assume it's not the PSU. All it takes is one part of the supply to die and you will lack the power to run something vital like the CPU or GPU regardless of the overall rating.

Second most likely cause is the motherboard itself, possibly a capacitor died or something. You can sometimes spot bulging or corroded capacitors.

CPUs generate a ton of heat, so I wouldn't worry about the CPU heatsink being really hot just yet. It's hard to imagine anything physically changing that'd cause the CPU to not work and generate more heat... well except maybe heatsink/thermal paste failure, but then the heatsink would be cooler.


As an aside, I'm fairly sure computers won't boot with no RAM in the motherboard at all. Usual instructions are to leave one stick in when troubleshooting or testing a new build. Having said that if the system was running fine and you tried to boot with no RAM it should give a BIOS error beep code instead of just being dead. Your RAM is probably fine.

Eat it You Nasty Pig.

The CPU heatsink is supposed to get pretty hot, that's it's job.

When you boot now (with the RAM in), do you get to a POST, or is it just some lights and spinning fans? If you don't suspect a PSU issue, the next step is determining whether the motherboard is still working.

Your previous trouble with startup to me suggests a motherboard issue; a connection or capacitor somewhere is failing.

If you overheated badly enough to fry your processor it could easily have taken the motherboard with it, but the only way to really tell is to swap in another processor and see if it boots.

Ego

CycloneRanger wrote:

The LEDs power on and all the fans turn on--plus, there's the CPU heating up. It seems unlikely that there's a power supply problem here. My PSU is also rated for a higher power output than I need for this system, so its sudden failure is less likely.

I'll try powering the machine on without the graphics card, but I'd be surprised if that were the issue. I suspect at this point it is either the processor or the motherboard and I'm not sure how to determine which is failing.

I've also had trouble starting up for many months--powering my system on after it has been shut down for any length of time would cause the system to start up but hang before reaching the login screen; repeated attempts were necessary to get it to start. I attributed this to bad RAM (there's a similar known issue in some units of this model of RAM), but it might be related to this new problem.

If you had an issue where powering on required you to turn on the system, then re-turn it on (if I'm understanding your last paragraph correctly,) that certainly sounds like a PSU problem. PSU failures are relatively common even if you have more wattage than you need.

Might want to invest in a multimeter and test the PSU directly. It's actually not that tough to do (http://pcsupport.about.com/od/toolsofthetrade/ht/power-supply-test-multimeter.htm) --the only part that's remotely uncomfortable is doing a pin short to let the PSU power up. And even that is really not a big deal. Multimeters are also cheap and handy tools in general.

CPUs tend to be sturdy, so if your PSU checks out fine and you've tried running with different memory chips, you've narrowed it down to the motherboard or GPU, for say a $20 investment.

CycloneRanger

Well, the PSU is the one component I can directly test, because I have two working spares. They're both currently hooked in to different parts of an old Dell in a (successful) attempt to get something working (I did have to short out a few pins to do that), but I should be able to at least verify that the PSU isn't an issue (or is) after work today.

I'll also try powering on without the video card; if that is successful it might also imply a PSU problem or video card problem.

If the video card and PSU check out, how can I distinguish between a failed motherboard and a failed CPU? Worse, how can I be sure that, if the CPU failed, it didn't fail because the motherboard is defective and damaged it? I do not have a spare CPU to try this with.

Raern

As Ego said CPUs are pretty reliable. I'm sure it's happened somewhere, but aside from melting one by overclocking I've never heard of someone having a CPU fail. Motherboards on the other hand are one of the most likely components to be at fault when a system fails after months or years of use.

You didn't mention system specs at all, but if your computer isn't too new then replacing the MB/CPU/RAM may be the best plan anyway after being sure the PSU works. Worth checking prices on new gear before buying replacement parts.

CycloneRanger

Well, I figured it out. I employed the old engineering strategy of "fumble around until you trip over a clue" and eventually noticed that there were a lot of memory slots just sitting there and I had not given much thought to which one I ought to use. It turns out that on a Gigabyte board, at least, you're not supposed to use the slot closest to the CPU if you're only using one RAM stick--you use slot #2. After figuring that out I retraced my earlier diagnostic steps and determined that it was...

The GPU.

Didn't see that one coming.

I still suspect my RAM of treachery, but after replacing my GTX 460 with an old 8800 everything was working at least minimally. Now I just need to figure out whether this thing is still under warranty (it damn well better be) and whether I should RMA it to Newegg or the manufacturer. Any ideas there?