7700k delid with CLU - should I redo it?

[RichUK]

Here's mine - you can see the bubbles (non mirror finish):

 

[Czachol]

Here's mine - you can see the bubbles (non mirror finish):

Thanks for the picture. Your result really is similar to what achieved! Have you remounted the heatsink in the past months, after you delidded the 7700K? I wonder if it is ok to do without need to repeat the delid.

 

[Valantar]

Thanks for the picture. Your result really is similar to what achieved! Have you remounted the heatsink in the past months, after you delidded the 7700K? I wonder if it is ok to do without need to repeat the delid.

That looks like every single picture I've ever seen of a CLU/liquid metal application, so if yours looks like that, you should be good.

As for removing the heatsink, I would think that's okay as long as the CPU retention bracket is holding the IHS in place (it shouldn't be possible to lift the IHS at all without ripping out the socket, after all), but someone with actual experience delidding should probably verify this.

 

[Czachol]

That looks like every single picture I've ever seen of a CLU/liquid metal application, so if yours looks like that, you should be good.

As for removing the heatsink, I would think that's okay as long as the CPU retention bracket is holding the IHS in place (it shouldn't be possible to lift the IHS at all without ripping out the socket, after all), but someone with actual experience delidding should probably verify this.

My thoughts exactly. If caution is taken not to rip the heatsink right out but first twist and move it in order to break the adhesion force between it and IHS, it should be fine. In theory at least. However, the layer of liquid metal between the die and IHS is so thin that who knows how it could turn out...

 

[DrMrLordX]

On a PGA socket (FM2+), I had bonded the IHS so tightly to the heatsink base that the IHS just popped off the weak adhesive bond (I left some of the black epoxy stuff on the PCB as a guide plane) and stayed stuck to the heatsink. It was a pita getting it off the heatsink later. Of course, that was a lapped IHS bonded to a lapped copper heatsink base by CLU. That's a pretty strong bond.

If you use normal TIM it shouldn't be that big of an issue, especially if you heat up the CPU with a good stress test shortly before removing the HSF.

 

[Czachol]

On a PGA socket (FM2+), I had bonded the IHS so tightly to the heatsink base that the IHS just popped off the weak adhesive bond (I left some of the black epoxy stuff on the PCB as a guide plane) and stayed stuck to the heatsink. It was a pita getting it off the heatsink later. Of course, that was a lapped IHS bonded to a lapped copper heatsink base by CLU. That's a pretty strong bond.

If you use normal TIM it shouldn't be that big of an issue, especially if you heat up the CPU with a good stress test shortly before removing the HSF.

You mean my removing the HSF shouldn't mess up the die to IHS bond?

 

[Valantar]

You mean my removing the HSF shouldn't mess up the die to IHS bond?

It really shouldn't. The only way it could as long as the retention bracket is mounted, is if it somehow had enough adhesion to the centre of the IHS to make it flex. That's not happening.

And @DrMrLordX, what gave you the idea to put CLU between two copper surfaces? That sounds like a rather poor idea

 

[DrMrLordX]

It really shouldn't. The only way it could as long as the retention bracket is mounted, is if it somehow had enough adhesion to the centre of the IHS to make it flex. That's not happening.

Exactly. Also a lot of delid/relid plans call for rebonding the IHS to the PCB if you are satisfied with what you have on the die. In my experience, CLU will last a long time in there.

And @DrMrLordX, what gave you the idea to put CLU between two copper surfaces? That sounds like a rather poor idea

Actually I've done that quite a bit, ever since I lapped my NH-D14 in . . . 2009? That HSF was the gift that kept on giving. Anyway I figured I'd lap the IHS for kicks since it was already off the CPU. Easiest lap job ever. Copper -> copper bonding via CLU is actually pretty awesome. You can paint it on extremely thin for amazing heat transfer. If you have a surface that has already been "corrupted" by CLU, it's even easier. My old NH-D14 is all eat up with that stuff.

 

[Valantar]

Actually I've done that quite a bit, ever since I lapped my NH-D14 in . . . 2009? That HSF was the gift that kept on giving. Anyway I figured I'd lap the IHS for kicks since it was already off the CPU. Easiest lap job ever. Copper -> copper bonding via CLU is actually pretty awesome. You can paint it on extremely thin for amazing heat transfer. If you have a surface that has already been "corrupted" by CLU, it's even easier. My old NH-D14 is all eat up with that stuff.

I get that, but isn't it an absolute nightmare in terms of removal/remounting? Not that that's something you do every day, of course.

 

[IndyColtsFan]

Please forgive my ignorance, but I have a question. I always thought delidding was where the IHS was completely removed, allowing direct contact with the die for cooling. That's kind of why I've avoided doing stuff like that.

However, I've seen a few posts lately indicating that delidding seems to be removing the IHS, removing Intel's TIM and replacing it with something with higher thermal conductivity, and remounting the IHS. It seems that will be the approach Silicon Lottery will use with Skylake-X as well. Is this more or less what delidding really is and it is rare to do direct die contact for cooling?

 

[DrMrLordX]

I get that, but isn't it an absolute nightmare in terms of removal/remounting? Not that that's something you do every day, of course.

In non-delid situations, it's manageable. I had copper->copper contact on my Athlon x4 635, and while it was a bit . . . cantankerous when removing it, it didn't rip out the socket or anything. The 635 was (to the best of my knowledge) a soldered chip. I did lap it, but I did not delid.

If you do delid and re-epoxy (or otherwise fix the IHS back to the PCB with a firm mount), it shouldn't be an issue.

Please forgive my ignorance, but I have a question. I always thought delidding was where the IHS was completely removed, allowing direct contact with the die for cooling. That's kind of why I've avoided doing stuff like that.

However, I've seen a few posts lately indicating that delidding seems to be removing the IHS, removing Intel's TIM and replacing it with something with higher thermal conductivity, and remounting the IHS. It seems that will be the approach Silicon Lottery will use with Skylake-X as well. Is this more or less what delidding really is and it is rare to do direct die contact for cooling?

Direct contact with the die is risky, especially with HSFs that can have high pressure mounting and mounting systems that might not work so well once you've performed the delid operation. Noctua's SecuFirm system has stops that have to be lowered/shortened to work with a bare die setup. Watercooling fanatics love bare die though.

 

[Valantar]

Please forgive my ignorance, but I have a question. I always thought delidding was where the IHS was completely removed, allowing direct contact with the die for cooling. That's kind of why I've avoided doing stuff like that.

However, I've seen a few posts lately indicating that delidding seems to be removing the IHS, removing Intel's TIM and replacing it with something with higher thermal conductivity, and remounting the IHS. It seems that will be the approach Silicon Lottery will use with Skylake-X as well. Is this more or less what delidding really is and it is rare to do direct die contact for cooling?

Delidding can mean both, but bare-die cooling is generally avoided due to the very high risk of cracking the die. I suppose we might be returning to the AMD Duron days of CPU shims if Intel keeps up this shenanigans. Delidding and relidding with a better TIM is becoming surprisingly popular simply because you easily lower temperatures by 15-20 degrees on an i7-7700K. Which is totally crazy.

 

[IndyColtsFan]

Delidding can mean both, but bare-die cooling is generally avoided due to the very high risk of cracking the die. I suppose we might be returning to the AMD Duron days of CPU shims if Intel keeps up this shenanigans. Delidding and relidding with a better TIM is becoming surprisingly popular simply because you easily lower temperatures by 15-20 degrees on an i7-7700K. Which is totally crazy.

Thanks guys. I'm thinking if I go with SKL-X or even Coffee Lake, I may look at buying from Silicon Lottery and have them do the dirty work.

 

[Valantar]

Thanks guys. I'm thinking if I go with SKL-X or even Coffee Lake, I may look at buying from Silicon Lottery and have them do the dirty work.

If I lived in the US and was considering an Intel build, that would definitely be considered. Buying a K-sku CPU and not overclocking is a bit silly (and buying non-K cuts down clocks a lot), but if the 7700K is any indication, the 8700K(?) will probably need delidding to overclock well. Buying from SL at least cuts down on both work and risk, and if they only charge $50 for it, that's hardly more than the price of the delidding tool itself.