Boot Mystery

[whovous]

I posted a longer version of this before - let me try to be more focused this time.

System includes:
Gigabyte GA770TA-U3 mobo
AMD 3 core CPU
An enormous CPU cooler that has no visible name
MSI 5750 GPU
OCZ Modstream 500 PSU
OCZ 100G SSD
A standard HD and DVD drive
4 Gig of RAM from GSkill

Built system for vacation cabin. After a large storm, the system would not light up at all. One slight flicker when I hit the power switch, then total darkness and silence.

Took system home to play with it. Hooked it up, and immediately everything worked fine.

Took system back to cabin, where it continued to work fine.

Left system on overnight (mistake!) and found it dead in the morning. Power button did nothing. Fiddled with PSU and power cord, and eventually hitting the power button caused everything to light up which should light up, and every fan to spin which should spin, with a single exception. The Scythe fan on the CPU cooler would spin for a second or two, then slow to a stop.

Took system home to play with it. Hooked it up, and immediately everything worked fine.

Disconnected system and reconnected my home box to the monitor. All fine, of course, but then I had reason to reconnect the vacation box once again. Copy and paste time here:
hitting the power button caused everything to light up which should light up, and every fan to spin which should spin, with a single exception. The Scythe fan on the CPU cooler would spin for a second or two, then slow to a stop.

My monitor remains blank through all of this. The computer makes a few noises like it MIGHT be thinking about booting (tho I guess I am not entirely sure what an SSD system should sound like at boot - this is much easier when you are listening to platters spinning), but the screen stays blank and the fan on the CPU cooler stays dead.

I have confirmed that everything was seated properly. I have disconnected everything but the SSD and the GPU with the same result. I have not pulled the RAM, because it is well hidden by the cooler.

I have replaced the CMOS battery. Old one read 3.0v and the new reads 3.3v. I could not find a jumper, so I shorted the CMOS pins with a screwdriver. After each change, I get the continuously blank screen and the CPU fan that spins for only a second or two.

Any ideas about what is causing this? I could try swapping the GPU from my home (Intel-based) machine, and probably the RAM as well, but it is hard to see how either change is going to give me anything other than a blank screen and a dead CPU fan.

It seems to me like this is a BIOS problem (as in, the CPU shots down when it does not detect any BIOS) but I do not know how to fix it, and the intermittent nature of the problem really does not fit with the BIO theory, does it?

 

[whovous]

Lotta views. Any ideas?

 

[MagnusTheBrewer]

Possible micro fractures.
You need to go back to basics and do a minimal install with one stick of ram, GPU and NO HDD. Rebuild stepwise, attempting to boot after each component is added. Eliminate and consolidate.

 

[whovous]

I will try that, as it does not require much more than I have already done, but I do not see how this is going to cure the complete failure to post.

I think I either need to find a way to restore the BIOS or else buy a new mobo. I wonder if it could be new CPU time as well?

 

[MagnusTheBrewer]

I will try that, as it does not require much more than I have already done, but I do not see how this is going to cure the complete failure to post.

I think I either need to find a way to restore the BIOS or else buy a new mobo. I wonder if it could be new CPU time as well?

It may not but, trying to troubleshoot a system with everything attached just adds needless complexity and you may have luck with just one stick of ram. Be sure the one stick is in the correct slot for that configuration.

The troubleshooting method I'm referring to "eliminates" possible conflicts/failures and "consolidates" known working combinations.

 

[whovous]

Ahh, now you've caught me. Which slot is the correct slot? When I have tried troubleshooting with a single stick of RAM in the past, I've been able to get to BIOS with any slot. Of course then I was attempting to diagnose Memtest errors on a different system, but I was not aware that which slot mattered. In fact, then I had to use each slot once because, as it turned out, the problem was with some slots of the mobo and not the RAM.

 

[MagnusTheBrewer]

Ahh, now you've caught me. Which slot is the correct slot? When I have tried troubleshooting with a single stick of RAM in the past, I've been able to get to BIOS with any slot. Of course then I was attempting to diagnose Memtest errors on a different system, but I was not aware that which slot mattered. In fact, then I had to use each slot once because, as it turned out, the problem was with some slots of the mobo and not the RAM.

It looks like DDR3_1 is the correct slot. That's the one closest to the CPU.

 

[pcgeek11]

Do you have a bios speaker? What is the beep code. It may be the video initialization stopping the system POST Test.

 

[whovous]

Nope, not a peep. I will try disconnecting everything with a single stick in the first slot. I will also try with the video card from this machine if that does not work.

 

[whovous]

Disconnected all drives and pulled the second stick of RAM. No change.
Swapped video cards with my home rig. Still no change, although the cabin card works fine in the home rig.

Is there anything left to try, or is it time to get out my checkbook?

Should I replace both the mobo and the CPU, or just the former? The formerly intermittent nature of the problem makes me think it is a mobo problem, though I am not entirely sure I can explain why I think that way.

 

[THRiLL KiLL]

wonder if the cpu fan killed the proc.

next steps.

remove the mb from the case. and then try the one memory stick, cpu/fan video card only. this will eliminate any grounding.

Also i would try a diffrent power supply.

 

[whovous]

Do you think the fan is dead? It is a three-wire fan and it spins when given juice, but then shuts down. I assume it does that as a 'smart' fan, not as a dead one. Of course I could be wrong. It should be a simple matter to connect it to something other than the fan connection marked CPU FAN, and if it stops then, we probably have our explanation.

The PSU seems to be powering everything else OK...

Oh, I think I was less than clear on an earlier point. I do not have a BIOS speaker, which insures I do not get any beep codes. I do not know if this board has a place for one, nor if I still have one laying about somewhere, though I suspect I do not.

 

[whovous]

One last correction, although I have pretty much decided to give up on this and replace the mobo and the CPU. The Scythe fan on the CPU cooler is a four wire fan, not a three wire fan. The CPU fan connection has four pins, so they should play nice with each other.

Just out of curiosity, I tried again with the CPU fan connected to a 3-pin connection elsewhere on the motherboard. The fan worked just fine, but it still did not even pretend to boot. The screen remained blank the whole time.

To me, this means the problem is either with the CPU, or with the mobo in the area of the CPU, or both. The fan works, but I don't think it is meant to start up until it receives a signal (i.e., power) from the mobo via the BIOS. I noticed similar conduct on my home machine, which has an Intel based Gigabyte mobo. The other fans start when the power button is pressed, but the CPU fan waits for a second to get going, then presumably gets the 'signal' and starts to spin as well.

Thanks all for the responses, but unless someone has some new inspiration from this last clarification, I am going to throw in the towel on this one.

 

[THRiLL KiLL]

have you tried a different powersupply yet?

if you have a bad connector to the cpu power, you will get the same symptoms where everything else seems like it is powering on correctly.

 

[whovous]

The system worked fine for at least a year. I am not sure I understand how a connector could go intermittent, and then fail completely. I can see a connector getting bent during installation or at other times when it is disconnected and reconnected, but I have a hard time understanding how it could go bad intermittently and then completely when never touched.

Of course, this is compounded by the fact that I am lazy. The only other PSU I have is in my home machine, which means I have to take them both apart to get the other PSU into this box. It just seems like an awful lot of work for a low percentage possibility.