• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Ivy Bridge Delid, Delidding, Delidded

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

I.M.O.G.

Glorious Leader
Joined
Nov 12, 2002
Location
Rootstown, OH
In the recent past, Intel has been using an fluxless soldering method between CPU die and IHS. The soldering method they use(d) is patented, and described generally here:
http://www.google.com/patents/US7009289?printsec=description#v=onepage&q&f=false

Someone with a dead Ivy Bridge delidded and took a picture, stating nothing about a soldered connection, but only TIM between core and IHS:

546200_10150762676959940_521849939_9360350_683757774_n.jpg

Is anyone else delidding already, or has anyone else seen evidence of the thermal interface between die and IHS on Ivy Bridge?

The suspicion we have been discussing in light of higher heat levels being reported with Ivy Bridge, is that they haven't been using the fluxless soldering method to ensure a more solid bond between die and IHS, so while Intel claims the increased heat is due to transistor density... There may be other variables at play as well.

Anyone else with thoughts or references? Also, when you get your IB and delid it - post here. :D
 
Wow, why would they do something like that?
Seems like a really dumb thing to do, and my thermodynamics knowledge is rather limited.
This is really interesting, now to find folks rich enough to wreck Ivy cpu's :D
 
In the past, delidding has been shown to improve temps on some processors (I forget which models exactly, someone else can chime in on which chips delidding was popular with in the past). The particular chip pictured was already dead, so it was delidded just to see whats underneath - you can see they damaged the PCB at one corner of the IHS when removing it.

Regarding thermal management, there is only 1 way an IHS can actually improve heat transfer - if the bond between CPU die and IHS is SO good, that it is incredibly more efficient at transfering heat to the IHS than a direct die to HSF/WB interface. On a basic level with thermal transfer, a longer heat path or more thermal interfaces does NOT help heat conduct. The IHS is just a slab of copper, and there is no good reason it spreads heat better than if the HS/WB were in direct contact with the die... Unless the interface between die and IHS is immensely superior. If the heat interface is not immensely superior there, then its probably better for heat transfer to delid and cool the die directly.

This isn't a testament that anyone should delid - but we're curious now that we've seen one delidded and it does not appear to be soldered. So hopefully this starts some discussion and gets some more input and experience, or maybe people actually testing with their IBs.
 
So fluxless solder is better than a medium specifically made for transferring heat (I assume so as this worked on older processors, but when I say it like that it sounds counter-intuivitve)?
 
Solder is metallic and metallic bond > TIM for heat transfer, that's how I'm thinking.

I SO want to test this out, but CPU is too expensive for that. If I came across free Ivy CPU, I'd try ripping the lid off in a heartbeat.
 
The real question is: Why didn't they solder it?

Can the processor not stand up the heat required to solder the IHS without dying or having a large increase in failure? Cost? Complexity? They didn't just do this for no reason and they originally switched to a soldered IHS for a reason. This seems very odd.
 
Well, if they can solder IHS to CPUs while staying under the TJmax of the specific CPU, then it should definitely be able to handle the heat.

Definitely odd.
 
In the past, delidding has been shown to improve temps on some processors (I forget which models exactly, someone else can chime in on which chips delidding was popular with in the past).

I know Opteron 165's were one of them i delidded 2 of them myself :D temps went from mid 50's to high 30's under full load
 
The real question is: Why didn't they solder it?

Can the processor not stand up the heat required to solder the IHS without dying or having a large increase in failure? Cost? Complexity? They didn't just do this for no reason and they originally switched to a soldered IHS for a reason. This seems very odd.

Well, if they can solder IHS to CPUs while staying under the TJmax of the specific CPU, then it should definitely be able to handle the heat.

Definitely odd.


Heat isn't a problem during the soldering process exactly, the soldering process is done without any voltage going through the CPU, and the heat tolerances in those conditions are different - TJMax doesn't have relevance to the soldering process. Reading the Intel patent, they talk about really high temps in the ovens through the soldering process.
 
I was thinking along the lines of it affecting the longevity of the processor, not it outright killing the processor at solder time.
 
In the past, delidding has been shown to improve temps on some processors (I forget which models exactly, someone else can chime in on which chips delidding was popular with in the past).
Not some, all. Every single one. The IHS can not help cooling only hinder. I guess I sound like a broken record :D
Regarding thermal management, there is only 1 way an IHS can actually improve heat transfer - if the bond between CPU die and IHS is SO good, that it is incredibly more efficient at transfering heat to the IHS than a direct die to HSF/WB interface.
We need to break this down - if you solder the IHS to the die, yes you would get better heat transfer from die to IHS than using say, a TIM we could apply on a bare die. Even though the heat transfer may be greater from die to IHS (if soldered) than from die to WB or HSF that does NOT mean that you can cool the DIE better under an IHS. Period end of story. Would it be easier to cool a feverish person by dressing them in a ski outfit? :)

I know Opteron 165's were one of them i delidded 2 of them myself :D temps went from mid 50's to high 30's under full load
Exactly, temps will always go down when delidding if proper die/hsf/WB contact can be made.
 
Heat isn't a problem during the soldering process exactly, the soldering process is done without any voltage going through the CPU, and the heat tolerances in those conditions are different - TJMax doesn't have relevance to the soldering process. Reading the Intel patent, they talk about really high temps in the ovens through the soldering process.

Since I wasn't sure about the soldering process, I just knew that if the soldering heat was under TJmax then the CPU would, without a doubt, be okay; but it seems they can handle more heat when not operating. So, I guess it could be a heat issue when soldering.
 
Soldering is expensive.
TIM is cheap.
Presto!


Even soldered chips have better temps delidded, just not that much. The CPUz WR was held (up till BD) by a delidded soldered celeron.
It's just that you don't gain nearly as much by delidding a soldered chip and your risk of killing the chip is far, far higher.
 
Not some, all. Every single one. The IHS can not help cooling only hinder. I guess I sound like a broken record :D

Delidding isn't something that has been done on all processors though. It was only particularly popular on certain procs... On others, the risk of damage in delidding compared with the potential benefit was never perceived as worthwhile. On certain procs though, delidding was regarded as especially beneficial - rpkole's is one example of that kind of processor.

The rest of your statements, we've covered in private discussions, and I agree with. ;)
 
Delidding isn't something that has been done on all processors though. It was only particularly popular on certain procs... On others, the risk of damage in delidding compared with the potential benefit was never perceived as worthwhile. On certain procs though, delidding was regarded as especially beneficial - rpkole's is one example of that kind of processor.

The rest of your statements, we've covered in private discussions, and I agree with. ;)
On soldered IHSes the risk of killing the CPU was not worth the temp benefit, that is correct. I still stand by the fact that no matter how an IHS is affixed, the cpu can be cooled better without the IHS on, and that is my main point.
 
My guess would be because there is too little surface area to use the solder. I can only guess that the heat used to bond maybe too much for such a smalll area, possibly tweaking the ihs's shape too much, ie makeing it unevenly concave.
 
Back