Question About Soldering..

dandapice

Weaksauce
Joined
Nov 14, 2005
Messages
122
I have read multiple forum threads and articles about de-lidding the Haswell, but has anyone tried, or does anyone know of any information in regards to soldering the IHS to the core? Maybe a flux-less paste / hot air process?
 
I have read multiple forum threads and articles about de-lidding the Haswell, but has anyone tried, or does anyone know of any information in regards to soldering the IHS to the core? Maybe a flux-less paste / hot air process?

Long story short, what you are thinking of doing isn't going to work the way you think it will..You will at best create a cold solder joint that performs terribly, and at worst destroy the chip..

When Intel solders an IHS in place, both the IHS and the package are both heated to 115C+ for a short period..
 
Hey ccityinstaller,
That figures. They must do some sort of wave / oven type of soldering then. Must be their way of keeping us from doing it ourselves. I guess I'll have to look to IVY-E when they come out. Thanks for the info by the way...
 
Long story short, what you are thinking of doing isn't going to work the way you think it will..You will at best create a cold solder joint that performs terribly, and at worst destroy the chip..

When Intel solders an IHS in place, both the IHS and the package are both heated to 115C+ for a short period..

Lead-free solder melting points are actually on the order of ~220C. For IHS attachment, you'd ideally want to use a SnAg (Tin-Silver) solder which has a higher thermal conductivity. SnAg alloys tend to melt around 220C.

The only way to do this properly is with a temperature-controlled oven that allows for fairly precise temperature control and ramp rates. You would have to come up with an approximate reflow profile to get the job done, involving several minutes of slow temperature ramp-up to pre-heat the part, a quick ramp up to the reflow temperature, a short period of time at max temperature that is still long enough to properly reflow the solder, and then a cool-down ramp to get the part of the high temperature zone slow enough to not introduce stresses due to uneven or rapid cooling. If you're crazy you could try to pull this off with a hand-held hot air rework station, but I'd at least recommend a hot air bath preheater.

Solder thermal conductivity is decent but not great. Even the best solder is going to conduct heat 1/3 as well as Aluminum, for example. Therefore you want to make the solder layer as thin as possible under the IHS. This requires two things:

1) Even mechanical clamping pressure on the IHS during the reflow process. You want to squeeze the IHS down on to the die, but this must be done evenly and gently such that it doesn't crush the die or bend the IHS.

2) The proper volume of solder under the IHS. Use too little, and it won't cover the whole surface. Use too much, and it will squeeze out the sides. Actually that might not be too bad, given that there isn't much for the solder to come in contact with if it is extruded out of the die/IHS interface.



The real problem you have to solve, though, is how to properly flux both surfaces and introduce solder to the joint without creating any voids. The obvious (and typical) solution is a vacuum chamber to just eliminate the possibility of oxide formation. But that's not going to happen on a DIY scale. Instead, you're going to be stuck with a solder paste, which contains a bunch of tiny solder balls suspended in a paste that contains about 50% flux by volume. It looks awesome under a microscope, BTW. When you reflow, the flux will outgas significantly as it melts, and those gases are going to be trapped in your solder interface, creating voids.

This is a tough problem to solve over such a large, confined area such as the die/IHS interface. Some quick Googling reveals this thesis paper solving exactly that problem, however: http://scholar.lib.vt.edu/theses/available/etd-05012001-165030/unrestricted/JUNHYUNGKIM_thesis.pdf It also covers the basics of reflow soldering and the challenges involved in using solder for thermal interfaces. This is as close of a (good, scientific) DIY guide as you're going to get.



Okay, now someone go do it. :D
 
I wish intel would go back to soldering

That would be like having a car with the hood welded shut. I wish Intel would use a different socket and mounting so that it would be easier to run without a heatspreader.
 
Lead-free solder melting points are actually on the order of ~220C. For IHS attachment, you'd ideally want to use a SnAg (Tin-Silver) solder which has a higher thermal conductivity. SnAg alloys tend to melt around 220C.

The only way to do this properly is with a temperature-controlled oven that allows for fairly precise temperature control and ramp rates. You would have to come up with an approximate reflow profile to get the job done, involving several minutes of slow temperature ramp-up to pre-heat the part, a quick ramp up to the reflow temperature, a short period of time at max temperature that is still long enough to properly reflow the solder, and then a cool-down ramp to get the part of the high temperature zone slow enough to not introduce stresses due to uneven or rapid cooling. If you're crazy you could try to pull this off with a hand-held hot air rework station, but I'd at least recommend a hot air bath preheater.

Solder thermal conductivity is decent but not great. Even the best solder is going to conduct heat 1/3 as well as Aluminum, for example. Therefore you want to make the solder layer as thin as possible under the IHS. This requires two things:

1) Even mechanical clamping pressure on the IHS during the reflow process. You want to squeeze the IHS down on to the die, but this must be done evenly and gently such that it doesn't crush the die or bend the IHS.

2) The proper volume of solder under the IHS. Use too little, and it won't cover the whole surface. Use too much, and it will squeeze out the sides. Actually that might not be too bad, given that there isn't much for the solder to come in contact with if it is extruded out of the die/IHS interface.



The real problem you have to solve, though, is how to properly flux both surfaces and introduce solder to the joint without creating any voids. The obvious (and typical) solution is a vacuum chamber to just eliminate the possibility of oxide formation. But that's not going to happen on a DIY scale. Instead, you're going to be stuck with a solder paste, which contains a bunch of tiny solder balls suspended in a paste that contains about 50% flux by volume. It looks awesome under a microscope, BTW. When you reflow, the flux will outgas significantly as it melts, and those gases are going to be trapped in your solder interface, creating voids.

This is a tough problem to solve over such a large, confined area such as the die/IHS interface. Some quick Googling reveals this thesis paper solving exactly that problem, however: http://scholar.lib.vt.edu/theses/available/etd-05012001-165030/unrestricted/JUNHYUNGKIM_thesis.pdf It also covers the basics of reflow soldering and the challenges involved in using solder for thermal interfaces. This is as close of a (good, scientific) DIY guide as you're going to get.



Okay, now someone go do it. :D

Very nice post;.The 115C was a typo on my part, I meant 215, but you caught my error before I did heheh..
 
This is some really great information, however, I don't have the equipment or access to such equipment. I think I'm going to definitely wait to see how the Ivy-E CPUs do and make a determination then. Thanks for all the input everyone..
 
Intel used an Indium based solder TIM with a melting point of 157C, according to a spec sheet from years ago. It may very well be possible to solder it yourself if you can get a hold of the proper solder and the right equipment to do it.
 
Intel used an Indium based solder TIM with a melting point of 157C, according to a spec sheet from years ago. It may very well be possible to solder it yourself if you can get a hold of the proper solder and the right equipment to do it.

This is really good info. I wonder if we could find a low qty source.
 
Wanted to bump this and see if anyone has any new info about user-base IHS soldering success stories.

I was only able to find 1 thread at another forum where the guy actually bought some Indium based solder to try, but he was running on LN2 and I didn't see any results with a typical air cooled HSF setup.
 
I have read multiple forum threads and articles about de-lidding the Haswell, but has anyone tried, or does anyone know of any information in regards to soldering the IHS to the core? Maybe a flux-less paste / hot air process?
The solder isn't really the end-all be-all of TIM's either. I had my Q6600 IHS and TRUE lapped to mirror finishes and still improved temps, voltages, and clocks by delidding with a blowtorch and using good paste directly to the TRUE.
 
Hi guys, I just recently watched a video about removing the cap of the processor to replace the thermal paste. So I thought I would search it on here to find out how long it takes for the paste to become hard or unusable? And after removing the cap could you hold it back on with something like Phobya component glue?
 
Intel states the TIM under the IHS has a 7 year life expectancy, which is well past the 3 year retail / 1 year OEM warranty of the processor itself.

As for reattaching the IHS, I don't have an answer for that...perhaps someone can chime in.
 
Last edited:
It occurred to me that the same time they stopped soldering the IHS on, was when they introduced the 3D Trigate transistor.
I wonder if some part of the process, perhaps the high temp, damages the transistor or some of the materials used around them.
 
Intel states the TIM under the IHS has a 7 year life expectancy, which is well past the 3 year retail / 1 year OEM warranty of the processor itself.

As for researching the IHS, I don't have an answer for that...perhaps someone can chime in.
My MB is 7 years old. It would be horrible to be forced to delid my CPU. BTW is E7200 soldered?

It would be nice if Intel sold processors with no IHS.

Are you aware some people are installing CPUs with a hammer?
 
Are you aware some people are installing CPUs with a hammer?

Can't fix stupid. Warranty doesn't cover physical damage anyways. So, a non-issue. I believe an IHS'less version should be offered to those that want it.
 
It would be nice if Intel sold processors with no IHS.
Exactly! I hate the noobshields with a passion. They have done worse for cooling ever since their inception. The only time they may help is with HDT coolers because not all of the pipes would contact the naked die.

There was a time, I think in the Pentium 3 or 4 days, when there was the option to buy the mobile version that was sans noobshield. They had a tendency to OC way better than the models with noobshields...
 
Even back in the Tbird days, a lot of people ended up crushing the core of the CPU because they didn't have IHS's on top. I'm sure this also happened to P3's.

I have to imagine a lot of why there are IHS's installed is to protect the core, not just for thermal reasons.
 
That is why they are noobshields. I've never cracked a core with bare die, it is an ID10T error. Thermal reasons had nothing to do with it because they are much worse for thermal conductivity.
 
Even back in the Tbird days, a lot of people ended up crushing the core of the CPU because they didn't have IHS's on top. I'm sure this also happened to P3's.

I have to imagine a lot of why there are IHS's installed is to protect the core, not just for thermal reasons.

Only Idiots did that..In fact several different companies sold copper "shims" to prevent this from happening..

That is why they are noobshields. I've never cracked a core with bare die, it is an ID10T error. Thermal reasons had nothing to do with it because they are much worse for thermal conductivity.

Exactly.
 
Intel doesn't give a damn about what we enthusiasts would prefer or that most of us could install a heatsink on an IHS-less CPU with no issue. I'd bet that Intel started using IHS back in the day because the Dells and HPs of the world wanted a more robust CPU that could tolerate more assembly abuse. This would allow them to speed heatsink installation without worrying about broken or chipped cores due to the less precise force being used with a quicker installation method. Mobile CPUs stayed IHS-less because the IHS thickness was undesirable and the drop in thermal performance is considerably more acute in a cramped mobile environment.

All Intel cares about is that the IHS performs well enough at stock speeds as not to overheat the CPU with a typical OEM heatsink and that the TIM will last the warranty period.
 
Back
Top