Burned CPU!!! Why? How? Help!!

[Snowbum]

I just transplanted my PC equipment into my new Thermaltake case. This case is a beast to hook up because of all of the fan controllers and connectors for the Hardcano. As soon as I powered it on, POOF! Instantly my Athonl XP 3000+ was fried. I am sure that some of you are familiar with the distinct smell of a burned CPU.

The question that I have is how in the hell did this happen? I am very experienced with building systems and assembling HSF. The Zelman CU HSF combination powered on fine so it couldn't have been due to heat.

The Hardcano has three temperature sensors. The directions explain that one should be placed underneath the CPU. The diagram in the manual shows that the the thin wires for the leads should run between the pins of the processor and the sensor itself should be taped the center, square part of the processor without pins. I used the tape that came with the case, and placed the sensor wires between the processor pins as suggested. Is this indeed the correct way to run the temperature sensor?

• I did notice that after taping the sensor to the processor there was some oil, I guess from my fingers on the square part of the CPU where the sensor was taped. This also may have been caused by some residue remaining on my fingers from the Arctic Silver that I was using. Would this cause the chip to fry?
  
• Also, I noticed when I seated the processor that it did not sit ENTIRELY flat in the socket. There was a very slight, slight rocking of the processor in the socket. However, the ZIF lever still closed without any problems at all. Would this be the more likely cause of the fried CPU?

I just ordered another processor from Newegg and I don't want to make the same mistake twice. Please tell me what you think. Maybe I should skip the whole idea of placing the temperature sensor underneath the CPU. I heard that this is really the only way to get an accurate reading of your CPU temp though. Any thoughts would be greatly appreciated.

 

[cambio]

I've not used your heatsink before and therefore am not sure exactly how it mounts, but if you felt some rocking motion of the CPU, that sends a yellow flag that the two surfaces might not have been parallel to each other. No wonder it fried so quickly.

 

[Syphon Filter]

hrm...lemme get this straight...you booted a machine which u knew didnt have the processor seated correctly? and which u knew had grease on the CPU?

Sounds like either one of those could have been the cause...

 

[Elledan]

Originally posted by cambio
I've not used your heatsink before and therefore am not sure exactly how it mounts, but if you felt some rocking motion of the CPU, that sends a yellow flag that the two surfaces might not have been parallel to each other. No wonder it fried so quickly.

I have to go with this theory as well. Unless the mainboard and/or PSU are broken, sending power surges into the CPU, it is the only likely cause at this point.

 

[Little Grabbi]

I'd say there was some sort of dirt in the CPU socket, either blocking a necessary pin, or bridging between two of them. Either could cause the CPU to fail.

 

[Snowbum]

Originally posted by cambio
I've not used your heatsink before and therefore am not sure exactly how it mounts, but if you felt some rocking motion of the CPU, that sends a yellow flag that the two surfaces might not have been parallel to each other. No wonder it fried so quickly.

The HSF actually screws into two brackets that are mounted to the MB via the mounting holes located around the CPU. I am most certain this was not caused by a gap between the HSF and CPU.

 

[Snowbum]

Originally posted by Syphon Filter
hrm...lemme get this straight...you booted a machine which u knew didnt have the processor seated correctly? and which u knew had grease on the CPU?

Sounds like either one of those could have been the cause...

Will grease on the pin side of the CPU cause this to happen? If so, in the event grease does manage to get onto the pin side of the CPU, will cleaning it with alcohol prevent something like this from happening? What about grease on top of the CPU that is not on the center core thingy? Should this also be cleaned first?

 

[Syphon Filter]

i have built a few PC's...and i would never consider booting a CPU that had any kinda visible foriegn stuff on it...i dont know for sure if that could have done it, but i dont think its impossible...

and as for not seated correctly that too would be a no no...

 

[darktiger]

I dont think your HS was placed properly.

 

[kingsack]

I have never seen that HSF but I wonder if it is directional with respect to orientation with the ZIF Socket? If he put it
on backwards it might rock back and forth and the bolt through MB design wouldn't help at all. Little to no contact between CPU and HSF = one cooked Barton!

 

[dobbz]

Originally posted by Snowbum
The HSF actually screws into two brackets that are mounted to the MB via the mounting holes located around the CPU. I am most certain this was not caused by a gap between the HSF and CPU.

just cause you have screws and mounting holes doesnt mean your heatsink was mounted flush with the core. id bet you that a clip sink would allow for the slight angle caused by the sensor and still let it sit right.

 

[Little Grabbi]

Originally posted by kingsack
I have never seen that HSF but I wonder if it is directional with respect to orientation with the ZIF Socket? If he put it
on backwards it might rock back and forth and the bolt through MB design wouldn't help at all. Little to no contact between CPU and HSF = one cooked Barton!

I concur... never throught of it before, but I think someone in my neighborhood had a CPU die that way...

 

[Snowbum]

Originally posted by Little Grabbi
I'd say there was some sort of dirt in the CPU socket, either blocking a necessary pin, or bridging between two of them. Either could cause the CPU to fail.

There definitely was not any dirt in the CPU socket. When I said that there was some oil on the pin side of the CPU, I meant from simpy touching it with my fingers while trying to tape the thermal sensor to it. Could this cause the CPU to burn out? If so, how in the heck do I tape the thermal sensor to the pin side of the CPU without touching it?

The question that remains unanswered is if my descirption of the way I placed the thermal sensor on the processor is indeed the correct way to do it.

Also, I am certain that the HSF unit was centered on the core of the processor. The HSF unit did not rock or move at all and did indeed make contact with the CPU.

 

[ocphangaz]

when you placed the sensor under the cpu, it'll make an uneven surface on the ZIF socket.. this might caused the rocking of the proc...

when the proc is not level, your HSF will not touch the core correctly, which will cause overheating bla bla bla...

 

[CaNNoN]

you placed the sensor UNDER the CPU in the socket? or ON the CPU die? if yu put it ON the CPU die then it would fry instantly. if you put it under, you prolly got hardly any even contact between the CPU and HSF base. AMD CPU's will fry in exactly 1 second without total HS to CPU die contact. saw this on a video from toms hardware. it really takes exactly 1 second for smoke to rise.

C

 

[Tordek]

Put some links on that 1 second fried amd cpu... I would really like to see how not to screw up my clawhammer....

 

[TruthInRuin]

Your positive one second? It takes the fan on a HSF half a second to start up.. would it be way to hot by then? And wouldnt booting it up numerous times at a temp that high damage it?

 

[Syphon Filter]

the fan spinning up is not the important part for the 1st couple of seconds...its the mere fact that there is a big lump of (cool) metal sat on top of the core...just waiting to absorb the heat from the core instantly pretty much. But if the contact isnt good then yes AMD's will die...FAST...

 

[Snowbum]

Originally posted by CaNNoN
you placed the sensor UNDER the CPU in the socket? or ON the CPU die? if yu put it ON the CPU die then it would fry instantly. if you put it under, you prolly got hardly any even contact between the CPU and HSF base. AMD CPU's will fry in exactly 1 second without total HS to CPU die contact. saw this on a video from toms hardware. it really takes exactly 1 second for smoke to rise.

C

Finally, someone addressed my question. I did indeed place the sensor on the pin side of the CPU die itself. Should it be placed under the CPU on the motherboard part of the socket? Is there a correct way to do this to ensure that there is good contact between the CPU and HSF? Or is this something that most people don't do?

IF I SHOULDN'T place the sensor on the motherboard part of the socket, where is the best place to put it to get accurate CPU temp?

 

[BlownFuse]

I'm curious on the "correct?" way to do this also.
Would it be plausible to drill a tiny hole through the heatsink, towards the center bottom ( at an angle )? Then just slide the probe in there once the heatsink is mounted?

 

[mbmadness]

Were there any scorch marks on the pin side of the CPU? I'm wondering if the thermal sensor's wire may have somehow had a bare spot on it and it touched one of the pins and caused a short.

 

[cambio]

Originally posted by BlownFuse
I'm curious on the "correct?" way to do this also.
Would it be plausible to drill a tiny hole through the heatsink, towards the center bottom ( at an angle )? Then just slide the probe in there once the heatsink is mounted?

NO way, you'll be measuring the temp of the metal in the heatsinkwhich will be lower than the source.

 

[BlownFuse]

Originally posted by cambio
NO way, you'll be measuring the temp of the metal in the heatsinkwhich will be lower than the source.

Well, I mean really get the probe down there, close to where the contact is. Here is a horrible image of what I mean:

|0000000000000|
|0000000000000|
|000000000000_/
|0000000000_/0 |
|00000000_/ 000|
|000000_/ 00000|
|00000 / 000000 |
-------------------------

 

[cambio]

Cool idea and great job on the rudimentary drawing, but the density of the metal in that heatsink is so different than that of the cpu. It's just not going to provide an accurate or fast enough measurement.

 

[BlownFuse]

It's a shame that I couldn't wire up a thermocouple from plastic injection mold machines. Those puppies are durable and accurate. On a hot-runner mold, they are inside the mold itself and accurate to within +/- 1 degree at the injector tips.