Two cpus fried back to back -- getting no post

highdef2000

Distinguished
Jan 21, 2010
8
0
18,510
I've been dealing with this issue for a few weeks now, and got the machine running the first time with ideas off these forums. Except, I wouldn't be here posting a thread myself, if the new setup had lasted more than three days. It didn't.

Here's what happened. This is a self-built, liquid cooled computer, and had been running fine for a good four years. Last week, suddenly windows froze. A few days later it froze again. Then, upon restarting, the machine would only go into post every other time, and freeze or randomly restart some time down the road. And finally it got to where I was getting no post at all.

I went through the usual suspects first. New motherboard battery, clearing CMOS, switching RAM sticks around. Nothing changed. My local big store has an easy return policy, so I also tried a new power supply and video card. Still nothing. I then got a replacement motherboard. No change. Only with a new (bought used off of ebay) CPU that I tried last, the computer came back to life, boom! And it felt good, let me tell you.

Except then, on the third day of moderate use, there was another random freeze (Again, this computer
ran like butter for four years. No funny stuff whatsoever, linux OR windows). I was a little worried, but kept using it. For the rest of the day, the machine shut down and restarted with no problems a few more times, as I was on it.

And then, this morning, again no post. All symptoms exactly the same as last time.

I've already ordered another processor. While it's in the mail, I wanted to share this experience and see what the ideas are as to what might be going on.

To be clear, when disaster struck the second time, I was using the same power supply, same graphics card, same RAM, with a new motherboard and a new (bought used) processor.

Some thoughts of my own:
-when testing the power supply with no load, the 3V leads are a little over, the 5volt leads are a little over,
but the 12 leads, both + and -, are at about 10.8. But then, people tell me the 12V leads are for the hard drives only and have nothing to do with the processor.
-when putting in the new processor, I used some $2 thermal paste off of ebay. Should I instead have gotten something fancier like 'arctic silver' for $8?

I'd appreciate any thoughts. Also I promise to close up this thread properly by posting the solution, which seems to be overlooked on forums often.
 
whoever told you that the 12 volt is just for your harddrives is wrong the 12 volt rail powers alot on your motherboard including your CPU and 10.86 is way too low im suprised it will even post at all

edit WHAT processor you useing and motherboard
 

highdef2000

Distinguished
Jan 21, 2010
8
0
18,510
The last processor ran at 29 C fresh in BIOS, and at 39 C roundabouts under normal load. I know my cooling is tight.

Well, a story. In all innocence, I was wobbling my bicycle down the sidewalk of Magazine St, a neighborhood main street in New Orleans, on my way to get my hair cut, when I saw a computer repair shop. I figured it wouldn't hurt to ask. It's the counter guy there who said 10.6 is nothing to worry about on the PSU 12V rails.

But then, he also said that liquid cooled isn't a very good idea in general, especially in Louisiana, because of the climate, and condensation. Nevermind that the machine ran fine in Connecticut for 2 years, where it's also hot and wet as all hell. So the solution, of course, was to 'bring it in'.

Funny thing, when I was searching through these boards last time, I also saw someone say that anything below 11.7 isn't going to cut it. So I went to the Best Buy here to get another power supply, the Geek Squad guys said 10.7 was nothing to worry about (right before he tested for post with a new video card without the monitor cable hooked up). haha.

This is an ASRock 939 socket motherboard, with a dual core AMD. The one that lasted 4 years was a 4200+, the second one that burnt was 3200+ that I got for $10 on ebay just to see if the CPU was the problem.

So at this point we have two local experts saying 10.7 is good enough against two guys on these forums saying 10.7 is way the hell low. I think I'm getting the picture. Just to be sure, could we get a few more people to weigh the balance?

Thank you.


 
Well - first off if the +12V was only for the HDD then why would the 20 or 24 pin MOBO adapter have the +12V pins ?? -- not to mention the newer MOBO's that have additional 4 or 8 pin +12V connections specifically for the newer CPU's Since the MOBO connector alone can no longer provide enough +12V wattage to run everything !! (the PCI-e slot also provides +12V current to the PCI-e video cards as well - but newer cards also use one or two 6 pin +12V wires since they use more than the 75Watts the PCI-e slot can safely provide ) - so there are alot of things using the +12V ---- and running them at 10.7 Volts Max (it will drop even lower under stress if it is already that poor) is never going to be adequate. (If it was then why would they make PSU's with +12V in the first place if +11V was adequate ??) -- Sounds like the PSU is deteriorating to a point where it is no longer providing steady power to the system so I would definitely replace it !! (before it blows out and takes the components with it)
 
From the Jonnyguru PC Power Supply Basics FAQ
"CPU’s and GPU’s regulate their power off of the +12V DC rail. Also, all of the computer’s motors run off of +12V DC: hard drive and optical drive motors, fan motors, pumps for water-cooling, etc."
http://www.jonnyguru.com/forums/showthread.php?t=3264

ATX spec requires that all rails be within 5% of nominal value, aka 11.4V to 12.6V, beginning with ATX 2.0 most of the power draw was shifted to the 12V rails to allow for the use of smaller wires, you can deliver the same power on smaller wires if you use higher voltages. For ATX specs take a look at the PDF, page 22 has the voltage tolerances
http://www.formfactors.org/developer%5Cspecs%5Catx2_2.pdf

the first ASRock 939 board i could fine claims to have a 4 pin ATX 12V connector which only contains 12V lines and provides power directly to the CPU. If the power being fed to the voltage regulation circuits for the CPU is outside the allowed range it may damage the circuits or they may not be able to provide sufficient voltage to the CPU to allow it to POST.

Since its reading so low, definitely return the PSU and get one from a good brand, try for an Antec, Corsair, Seasonic, PC Power & Cooling, or OCZ as they all make high quality units. A cheap PSU can blow up or manifest itself in a variety of problems, pay a bit more and get a high quality unit.
 

highdef2000

Distinguished
Jan 21, 2010
8
0
18,510
thanks for all the responses. i'm still a little puzzled that people working in real world computer stores would all be so clueless, but i guess that's just the way it is.

i'm putting my system back together with a new PSU and will close this thread properly after the machine has been running right for a few weeks.

the PSU that's been giving me trouble is an OCZ. i'm sticking with this brand for now since four years is still pretty respectable and they have a decent reputation.

thanks again.
 

highdef2000

Distinguished
Jan 21, 2010
8
0
18,510
just confirming--the computer has been running fine the past three months. the fault was definitely with the power supply.

the main lesson i learned--any advice you are given at a computer store, whether big chain or a neighborhood one, is better ignored completely.

big thanks again to everyone who helped out here.

j.