troubleshooting Diagnosing dead parts after ripping CPU out of socket

[SpacePumpkin]

Hey, so I ripped my CPU (Ryzen 9 3900X) out of its socket the other day when trying to swap the cooler. I then spent two days trying to separate the two (it was like I used cement instead of thermal paste) and bend the pins back in place, eventually it came off after I heated the cooler up with a hair dryer and applied A LOT of twisty force. I've put the CPU through hell to get it off so I honestly don't expect it to have survived. After plugging it back in and giving it power, the lights on the motherboard turn on (Asrock SteelSeries X570) but the power button does absolutely nothing. Not even case fans spin. Behaviour is the same with and without the CPU installed. I'm wondering if this means the motherboard is dead as well?

 

I have a spare system on the same platform but am not willing to try swapping parts because I need *something* for my job. Aside from this I am also cautious of damaging parts through association (or also ripping the CPU out of the donor system's socket too...). I'm compiling a list of parts for a new build and was thinking of just taking the plunge to upgrade everything, but a part of me wants to salvage what I can. I've seen similar posts in the past with mixed success (I'm surprised its anything higher than 0%). Is it a safe bet to assume that both CPU and mobo are dead?

 

Thanks :)

[191x7]

12 minutes ago, SpacePumpkin said:

Hey, so I ripped my CPU (Ryzen 9 3900X) out of its socket the other day when trying to swap the cooler. I then spent two days trying to separate the two (it was like I used cement instead of thermal paste) and bend the pins back in place, eventually it came off after I heated the cooler up with a hair dryer and applied A LOT of twisty force. I've put the CPU through hell to get it off so I honestly don't expect it to have survived. After plugging it back in and giving it power, the lights on the motherboard turn on (Asrock SteelSeries X570) but the power button does absolutely nothing. Not even case fans spin. Behaviour is the same with and without the CPU installed. I'm wondering if this means the motherboard is dead as well?

 

I have a spare system on the same platform but am not willing to try swapping parts because I need *something* for my job. Aside from this I am also cautious of damaging parts through association (or also ripping the CPU out of the donor system's socket too...). I'm compiling a list of parts for a new build and was thinking of just taking the plunge to upgrade everything, but a part of me wants to salvage what I can. I've seen similar posts in the past with mixed success (I'm surprised its anything higher than 0%). Is it a safe bet to assume that both CPU and mobo are dead?

 

Thanks

Nothing should be dead.

Have you checked your PSU power switch?

Connected everything?

The CPU isn't that easy to kill, maybe you're missing a pin or one is still bent, recheck.

Make sure to re-seat the RAM and clear the bios.

Feel free to try the CPU in the spare PC, nothing can go wrong.

[Middcore]

It is vanishingly unlikely you could have damaged any other part through what you did to the CPU. 

[lafrente]

What thermal paste is that? We should make sure to avoid it wholly as a brand.

[Crunchy Dragon]

1 minute ago, lafrente said:

What thermal paste is that? We should make sure to avoid it wholly as a brand.

It's less of an issue with thermal paste and more of an issue with AMD's socket design. This has been a problem dating way back into the late 90s and early 2000s on AMD sockets up until AM5.

 

Because of how AMD's CPUs and sockets were designed, if the thermal compound isn't warmed up a little bit before the cooler is removed, it's possible to rip a CPU out of the socket when you remove the cooler. The only ways to avoid this problem on AM1-AM4 are usually to heat up the thermal compound in some way before removing the cooler(such as running a stress test on the CPU) or forgoing thermal compound entirely.

[Poinkachu]

9 minutes ago, lafrente said:

What thermal paste is that? We should make sure to avoid it wholly as a brand.

Pretty sure it's just mostly surface tension & the lack of heat before removing cooler.

Concave ihs might also contribute in strengthening the surface tension
 

[PDifolco]

Happened to me as well with my former 3900X, it worked ok after unbending pins, cleaning the paste that spread all over when it slipped from the block after being ripped off, and when I finally put it back in the socket the right way, I was so pissed off I didn't slot it properly at first ...

[Tan3l6]

Seems like that been would have that installed beforehand would have avoided the problem:

Spoiler

Or maybe not, perhaps it would have crushed the small parts on the sides of IHS when trying to lift "glued" cooler.

[SpacePumpkin]

In a shocking turn of events, as I was packing it away I reinserted the CPU and heard a crunch, it felt more stable in the socket than before. So I attempted power on again and it came to life, so I deconstructed the spare I was using this weekend, put its original parts back in and installed that the new cooler. It POSTs and has been online for the last 30 mins. Time will tell I suppose, I remember reading someone's story where the chip was only able to do like 5% of its original workload, I ran a passmark benchmark and got around 29k for the CPU alone (about 10% under the average). 

 

I'm gonna go make the most of the last of my stressful bank holiday weekend. It's been a rollercoaster. Thanks for the advice! I'm very paranoid when it comes to these things, would always rather cut my losses than risk breaking more stuff (although by the sounds of it, this was never a risk).

 

In terms of the thermal paste used, I'm guessing it was either the stuff that came with the stock cooler or Noctua NT-H2. I would agree with replies on this thread that its probably a mix of attempting the swap cold (lesson learnt there) and that its been near enough 100% usage 24/7 for the past 3 years. It had been edging 90C in recent weeks and sometimes the fan needed help to spin and would prompt thermal shutoffs (hence the need to replace the cooler)