My 6600K DELID steps and overclocking results! (With advice from Hoonigan)

[Fizzl]

gallium->aluminium

Mercury and aluminium is strange one, worth a google.

[CAT-THE-FIFTH]

One thing I'm curious about - I thought the heatspreaders were aluminium, and so unsuitable for use with gallium-containing liquid metals?

Both use nickel plated copper. However,the funny thing is the shape of the Intel IHS actually impedes things as they could actually shape it to have more surface area:

https://www.gamersnexus.net/guides/3...pu-rockit-cool

GN tested an aftermarket copper IHS and temperatures dropped a few degrees. With the AMD APUs it made much less of a difference as the normal IHS is larger in surface area anyway:

https://www.gamersnexus.net/hwreview...ven-ridge-apus

So its not only the TIM they use,but also the fact the IHS could also have a larger surface area which is not helping Intel.

[Ravens Nest]

Very interesting Hoonigan !

I would assume with the 8700k i7 part and the possible upcoming anniversary special edition intel Central Processor that 'delidding' and liquid metal may be really close to a necessity. More cores! Same tech! Same old 'engineering'!

I put that in quotes since Intel should have fixed Spectre and Meltdown many years ago, they were warned REALLY early and didn't do a thing.


I want my 7nm Ryzen and I want it now to be perfectly honest

I reckon that liquid metal or going back to solder will be there secret weapon after Ryzen 2 comes out.. then again something else might be waiting in the wing.

Nah, the vast majority are nickel plated copper. But you're right about the gallium->aluminium being a bad thing.

Its the Heatsink's you have to watch for, i would love to put liquid metal under mine but seems too risky. (Its a coolermaster TPC800 which is supposed to have a polished copper base)

Both use nickel plated copper. However,the funny thing is the shape of the Intel IHS actually impedes things as they could actually shape it to have more surface area:

https://www.gamersnexus.net/guides/3...pu-rockit-cool

GN tested an aftermarket copper IHS and temperatures dropped a few degrees. With the AMD APUs it made much less of a difference as the normal IHS is larger in surface area anyway:

https://www.gamersnexus.net/hwreview...ven-ridge-apus

So its not only the TIM they use,but also the fact the IHS could also have a larger surface area which is not helping Intel.

5 Degree extra drop in temperature, then again £20 for 5c drop is expensive vs the 20c drop for delidding.

Still tempting though..

Thought it was interesting on the Ryzen results that no adhesive holding the IHS was lower temperature than a relidded IHS?

[CAT-THE-FIFTH]

I reckon that liquid metal or going back to solder will be there secret weapon after Ryzen 2 comes out.. then again something else might be waiting in the wing.



Its the Heatsink's you have to watch for, i would love to put liquid metal under mine but seems too risky. (Its a coolermaster TPC800 which is supposed to have a polished copper base)



5 Degree extra drop in temperature, then again £20 for 5c drop is expensive vs the 20c drop for delidding.

Still tempting though..

Thought it was interesting on the Ryzen results that no adhesive holding the IHS was lower temperature than a relidded IHS?

Higher Z height due to the layer of adhesive??

[Hoonigan]

Higher Z height due to the layer of adhesive??

That's exactly it. The gasket seal around the base of the IHS increases the distance the heat has to travel through a TIM between the IHS and the silicon. Every little helps when getting the heat away from the CPU.

[Ravens Nest]

Higher Z height due to the layer of adhesive??

I guess it acts like an oven if you have a larger area under the IHS.

That's exactly it. The gasket seal around the base of the IHS increases the distance the heat has to travel through a TIM between the IHS and the silicon. Every little helps when getting the heat away from the CPU.

I have seen different ways people relid there IHS some use a small dot of superglue in each corner
(big no no, the superglue can melt at high temperature and doesn't come off very easily if you need to delid again and can cause damage to the CPU)

Tiny dab of gasket seal in each corner just to stop the IHS moving or falling off
(if i was reliding i would do it this way)

Gasket seal around the whole inside edge of the IHS
(Not good idea as you need a small gap to vent gases and excess air from reliding)

Gasket seal nearly all the way around the inside edge of the IHS leaving a small gap
(Good way but i guess as has been shown temperatures are higher as it raises up the IHS)

Thats the general idea im getting.

[Hoonigan]

I have seen different ways people relid there IHS some use a small dot of superglue in each corner
(big no no, the superglue can melt at high temperature and doesn't come off very easily if you need to delid again and can cause damage to the CPU)

I spoke to 8Pack at OcUK regarding relidding the CPU. A dab of superglue (Cyanoacrylate) in each corner will not melt until around 190°C, by which point your CPU will be toast. Cyanoacrylate also has a low shearing strength, and is often used as a temporary bond between two products, which makes it perfect if you need to delid the CPU again.

So I'm sorry, but whatever you have based your "superglue" facts upon, must be on a different type of glue and not cyanoacrylate, which is generally accepted as "superglue".

[Pob255]

Don't forget that one of the other reasons for the IHS is to protect the chip, over tightening a cpu cooler could crack the cpu, killing it.

Interestingly lapping the IHS was a big thing on the early intel IHS, because the soldering casuse warping of the IHS leaving them concave, this doesn't happen with the pasted IHS and lapping has died out (although direct heat pipe coolers are also a lapping no no)

[Pob255]

The superglue is also much easier to break due to its lack of elasticity, if you need to delid it again for any reason.

Although I'm betting that it also has a greater chance to damge the surface of the substraight (that is the correct term for the pcb the chip and IHS is on isn't it?)
I'd say a dab of silicon at the corners i probably best, stick it into the motherboard slot before it cures and the pressure from the retension bracket will do a good job of squishing it as flat as possible.

[Hoonigan]

Although I'm betting that it also has a greater chance to damge the surface of the substraight (that is the correct term for the pcb the chip and IHS is on isn't it?)
I'd say a dab of silicon at the corners i probably best, stick it into the motherboard slot before it cures and the pressure from the retension bracket will do a good job of squishing it as flat as possible.

I was only following the advice of one of the world's top overclockers, that's all

You may very well have a point about eventually damaging the PCB though, but that's all part of the risk really. However, in my experience, breaking the bond of a cheap superglue isn't that tricky.

[Pob255]

I was only following the advice of one of the world's top overclockers, that's all

You may very well have a point about eventually damaging the PCB though, but that's all part of the risk really. However, in my experience, breaking the bond of a cheap superglue isn't that tricky.

I wasn't saying it's not easy to break, super glue is brittle, the question is how easily the substraight surface will break?
And I'm guessing it's more of a case as to what you have to hand, most people will have superglue not silicon in a small applicator.

[Hoonigan]

And I'm guessing it's more of a case as to what you have to hand, most people will have superglue not silicon in a small applicator.

This probably has more to do with it, if it's readily available. It's just a couple of small dobs of superglue to hold the IHS in place until you've got it under the retention mechanism.

[peterb]

Need to be careful what silicon sealant you use - some liberate acetic acid as they cure - not good for electronic components.

[Ravens Nest]

I spoke to 8Pack at OcUK regarding relidding the CPU. A dab of superglue (Cyanoacrylate) in each corner will not melt until around 190°C, by which point your CPU will be toast. Cyanoacrylate also has a low shearing strength, and is often used as a temporary bond between two products, which makes it perfect if you need to delid the CPU again.

So I'm sorry, but whatever you have based your "superglue" facts upon, must be on a different type of glue and not cyanoacrylate, which is generally accepted as "superglue".

I just read i few horror stories, one was using a delid tool to delid a superglue covered IHS again and it took some of the PCB off killing the CPU, another was putting superglue under the IHS and getting it onto the contacts and under the pins, and the last was the superglue eating into the PCB (But saying that he wasn't using real superglue, so there is a lesson there i guess)

Don't forget that one of the other reasons for the IHS is to protect the chip, over tightening a cpu cooler could crack the cpu, killing it.

Interestingly lapping the IHS was a big thing on the early intel IHS, because the soldering casuse warping of the IHS leaving them concave, this doesn't happen with the pasted IHS and lapping has died out (although direct heat pipe coolers are also a lapping no no)

Very interesting, i never knew that and makes total sense.

This probably has more to do with it, if it's readily available. It's just a couple of small dobs of superglue to hold the IHS in place until you've got it under the retention mechanism.

Superglue Gel looks interesting as a way to relid though, so only a drop on two corners would be enough do you think?

Need to be careful what silicon sealant you use - some liberate acetic acid as they cure - not good for electronic components.

Any ideas what a good and preferably cheap one can be found?

[peterb]

Cheap and good don’t always go together!

The instructions usually state if they liberate acetic acid (often bathroom sealants). They also smell of vinegar!

[Pob255]

If I remember correctly you want the clear stuff not the white stuff.