9900K DEAD AFTER DELID

[AnonymousGuy]

I don't see anything physical that looks catastrophic, so my guess would be they brute-forced to get the IHS off and probably broke the bumps that connect the die to the substrate.  I believe they're solder bumps (soon to be replaced with copper but that's like 2 years out...bonus fun fact) so in theory you could try the toaster oven strategy that sometimes works to reflow graphics cards.

 

Also make sure the IHS is making good contact with the die too, which can be a problem if you didn't evenly sand off the solder.  If it isn't the chip can immediately go into thermal shutdown and appear broken.

[ntuason]

Damn the original owner probably used a cheap delid tool. It’s a goner.

[Letgomyleghoe]

On 2/4/2020 at 9:54 AM, Princess Luna said:

Why would anyone delid a soldered processor is beyond me

Direct die cooling

[Letgomyleghoe]

On 2/5/2020 at 10:16 AM, AnonymousGuy said:

so in theory you could try the toaster oven strategy

I used to have to do this on my 360 about once a month

 

wrap it in a towel, 375 for 5 mins.

[Guest]

On 2/4/2020 at 9:56 AM, SolarNova said:

Direct die cooling.

 

Die lapping.

 

And LM can in fact give u a few degrees over a soldered IHS.

LM does not do better than Solder in most cases.

For example, AMD uses Indium which has a melting point of 157c which is above reflow temperatures and why I can use heat to remove the IHS plate safely.

The Indium Solder has an 85 w/m-k while Liquid metal is 73 w/m-k while Thermal Grizzly claims 79 w/m-k. Even with this boosted number, is still lower than Indium solder.

This simply means there hasn't been a thermal compound that's been invented which would conduct and transfer heat better than solder.

 

Now, I'm not entirely sure exactly what Intel uses with the 9900K, someone would have to research the facts to have the numbers straight here. I'm not in the mood at this time.

 

@Mr.David

 

I would give this one last try. But first, you need to lap the die free of that remaining solder. Then, do NOT glue the IHS plate down. It takes up space and has potential to keep the IHS plate just high enough off the core, it'll make bad thermal contact.

Also, lap the under side of the IHS plate. Don't have to go nuts, just be sure the solder is lapped away and smooth.

When you install the LM, use a super duper thin layer. As thin as you can get it. 

At the time you clamp the IHS plate down, note the bracket slides a little. Do your best to set the plate back just a touch so it centers as the clamp slides. Once you have the IHS plate clamped, put just a little silicon glue at the edges you can access so when you remove the cpu, the IHS plate doesn't go sliding around on you.

 

Lastly, as may have been mentioned *TLDR* it would be best resulted if you where to lap the waterblock on your custom loop to meet the cpu die nice and flush. You can then use an extremely thin layer of thermal compound which will have minimal push-out. You would want the naked die kit for sure. 

 

Anyhow, good luck! And if it's really dead....Bummer, but good try! 

[Dabombinable]

1 hour ago, scuff gang said:

Direct die cooling

It's something from the days of the Pentium III and Athlons that I really miss. It was surprising at just how little heatsink you could get away with for the given TDP, compared to now. You could even get away with using bad thermal compound ($25 for a 150g tube of the stuff found in a lot of laptops even today).

[Guest]

Just now, Dabombinable said:

It's something from the days of the Pentium III and Athlons that I really miss. It was surprising at just how little heatsink you could get away with for the given TDP, compared to now. You could even get away with using bad thermal compound ($25 for a 150g tube of the stuff found in a lot of laptops even today).

I remember running socket A chips missing corners on the die. Good times. 2.0v weeeee

[Dabombinable]

7 minutes ago, ShrimpBrime said:

I remember running socket A chips missing corners on the die. Good times. 2.0v weeeee

Same with my PIII 1000 (the 667 with worse treatment by far hasn't had that happen). Cranking 2x of those (dual socket FTW) ever closer to 1.4GHz is almost an adrenaline rush.

[Guest]

1 hour ago, Dabombinable said:

Same with my PIII 1000 (the 667 with worse treatment by far hasn't had that happen). Cranking 2x of those (dual socket FTW) ever closer to 1.4GHz is almost an adrenaline rush.

Oh for sure, I had a mobile 450mhz chip but unfortunately the board I had was OEM and didn't OC. 

 

Kids now-a-days have no idea how good they got it lol. Back then these frequencies where getting fast as fuck. Slot 1 Mendocino 333mhz. Man that was so fast on OC. XD

[Beerzerker]

7 hours ago, Dabombinable said:

Same with my PIII 1000 (the 667 with worse treatment by far hasn't had that happen). Cranking 2x of those (dual socket FTW) ever closer to 1.4GHz is almost an adrenaline rush.

Here's a little something I did a long time ago......
 

[Whiro]

Still looks better than one off my xeons which I delidded  

 

[Kilrah]

F