Thermal compound for de-lid

[mackerel]

I've de-lidded two of my Skylakes and a pattern is emerging that after some months they seem to get hot again. Re-doing the compound fixes it, until the next time. I wonder if what I'm using is too runny and is getting pumped out. Initially I used the Noctua stuff, more recently MX-4. I might have used MX-2 in one that hasn't been redone, but can't be sure. I think MX-2 is a better working consistency, but I misplaced my tube. I could get more, or something else like it if someone can recommend?

Or should I YOLO it and go liquid metal? I've not really researched those. Are they apply once and forget, or do they need re-doing? Both systems are watercooled, and the compound of interest would go between die and IHS. Ideally if I'm going to do it again, I want it to be the last time.

 

[Woomack]

I'm using kryonaut on my 7700K and so far no issues. Can also try Gelid Extreme as it seems good too, if not better. Both had good results on chips which were cooled with LN2/dice and later went back to air/water. However I can't guarantee it will work perfect after long time. I just haven't seen results after more than half year. Of course can use liquid metal as it will give you the best results on ambient temps. I just don't like this stuff.

 

[trents]

If you are going to delid you might as well go all the way and use liquid metal so as to get the best possible temp drop. Most of the complaints I am aware of dealing with reversion to higher temps over time were from people using conventional TIM after delidding. Since all the components are now built into the die itself there is no real risk of spilling accidentally shorting out the CPU electronics by slopping some liquid metal onto the pcb.

 

[S_I_N]

I just recently went with conductonaut Thermal Grizzleys version of cool labs ultra(see warning below) with a relid after delidding so far so good not seeing the warm ups yet the blocks been remounted 4 times and loop changed as many as well. But its only been about 2 months now.

Warning on conductonaut: the syringe is somewhat harder to push than cool labs ultra to the point I managed to squirt it all over the motherboard. and while cleaning it up some ended up in socket where I ended up buggering the pins trying to get it out. My suggestion is squirt it onto a clean surface and use the included applicator q tips to pick it up off the clean surface and apply to core this way.

 

[Tír na nÓg]

If you are going to delid you might as well go all the way and use liquid metal so as to get the best possible temp drop. Most of the complaints I am aware of dealing with reversion to higher temps over time were from people using conventional TIM after delidding. Since all the components are now built into the die itself there is no real risk of spilling accidentally shorting out the CPU electronics by slopping some liquid metal onto the pcb.

Seconded.

I've delidded a few Ivy's and Haswells, best result, by far, (10c+ compared to "classic" TIMs) is with Coolaboratory Liquid Pro/Ultra.

Cover the components on the PCB with nail polish and you'll be safe

Edit: yeah, no risk anymore with Skylake/kabylake, as Trents stated.

 

[trents]

One tip I would have for applying CL's Liquid Ultra is to trim the applicator brush bristles by about 1/3 of their length. It will give you better control during the brushing on of the TIM.

 

[custom90gt]

Another tip is don't push too hard on the syringe because it will shoot out 10x more LM than you need. Ask me how I know... Another tip, you can actually use the syringe to suck up the extra amount that you applied to your desk...

 

[mackerel]

Ok, I see the votes are in for CLU so I will have to order some in. Thanks all.

 

[trents]

I would strongly suggest getting one of those delidding tools that push the IHS off the PCB. The Rocket tool seems to be sold out everywhere but you can get the equivalent knock off from ebay like I did. Works the same.

 

[custom90gt]

I would strongly suggest getting one of those delidding tools that push the IHS off. The Rocket tool seems to be sold out everywhere but you can get the equivalent knock off from ebay like I did. Works the same.

I think the OP has already delidded the CPUs, it's more of a question as to what TIM to use for longer consistent results...

 

[trents]

Got it!

But one more suggestion.

I note that Intel leaves a small unglued area around the perimeter of the lid. I presume they do this to allow air to come and go with heating and cooling of the CPU so as not to make pressure inside the lid that might cause flexing of the PCB/Lid and break the TIM seal. If you choose to reattach the lid with super glue or Permatex Ultra Black RTV (I used the latter) then I would follow Intel's example in this.

 

[S_I_N]

Yes

 

[mackerel]

I've gone and ordered conductonaut instead of CLU... Looking forward to trying it out.

To recap, I've already got two de-lid CPUs where I used normal compound. This helps in the short term, but for reasons unexplained I start to see hot cores again after some months. I'm hoping more for stability than absolute best performance, but still need better performance than stock Intel. I did find some long term comments saying CLU kinda crystallises after some years, which doesn't impact performance but could be a pain to clean up after if ever changing it.

 

[S_I_N]

Cool remember don't apply from syringe to core apply to the side and use included qtip it will jettison out of the syringe lol cost me a 250$ mobo

 

[custom90gt]

Cool remember don't apply from syringe to core apply to the side and use included qtip it will jettison out of the syringe lol cost me a 250$ mobo

Glad I'm not the only one to apply more LM than I thought was possible, sorry it cost you a MB