3770k IHS Removals - CPU temp dropped from 79C to 71C

[roger_h99.nor]

[QUOTE=tdbone1;5117472]
so my Q again is:
does it make the case temps higher running at 75-80C then SB running at 65-70C?
see i dont think so because the SB had more of a surface area (meaning it puts out a larger heat pattern)

ok so why do this mod?/QUOTE]

Hey, come on, you're at extremesystems. It's done because it's doable.

[tdbone1]

[QUOTE=roger_h99.nor;5117501]

ok so why do this mod?/QUOTE]

Hey, come on, you're at extremesystems. It's done because it's doable.

now that i can understand and that is prob the biggest reason why i would do it.
just wanted to make sure there was no "real good reason" to do it besides just doing it to do it.

[cab]

Why should you- the customer have to re-engineer & re-build the product? Intel could just admit that the package has a thermal defect and replace them. With Sandy Bridge they had a defective chipset and they replaced millions of them. They can issue a voluntary recall (or some lawyer makes them do it), you email them your purchase invoice, Intel sends you a replacement CPU and a prepaid box & label to return the defective one.


//

[theseeker]

Why should you- the customer have to re-engineer & re-build the product? Intel could just admit that the package has a thermal defect and replace them. With Sandy Bridge they had a defective chipset and they replaced millions of them. They can issue a voluntary recall (or some lawyer makes them do it), you email them your purchase invoice, Intel sends you a replacement CPU and a prepaid box & label to return the defective one.




//

1. That would make too much sense.
2. They are selling plenty of chips and they probably don't care. If they do, they will just change the process and call it an upgrade in their quality control.

[ToxicBug]

Why should you- the customer have to re-engineer & re-build the product? Intel could just admit that the package has a thermal defect and replace them. With Sandy Bridge they had a defective chipset and they replaced millions of them. They can issue a voluntary recall (or some lawyer makes them do it), you email them your purchase invoice, Intel sends you a replacement CPU and a prepaid box & label to return the defective one.


//

I heard that Intel already said they won't recall them. They don't give a . What are you gonna do? Boycott Intel? LOL! I'll still buy their next CPU and say thank you Intel for being so awesome!

[MikelMolto]

I tell You what I do. I will not buy Haswell and I will not buy the Tick from Haswell. This way Intel sell less to me and this way I punish them.
The best on this is, that i don´t have any problem with this and I save some money, what I can put in other Hardware (Monitor, SSD.....)
Intel save some Cent with using a cheap thermal paste and loos some money with stopping me to upgrade in my usual way.

[CrazyNutz]

Intel could just admit that the package has a thermal defect and replace them.

I find it rather strange you believe this constitutes a defect. It's not a defect in any way if it performs normally at stock.

[cab]

I find it rather strange you believe this constitutes a defect. It's not a defect in any way if it performs normally at stock.

In western countries, buyers decide what constitutes defect, and if there are enough of them you have a class action.

[ToxicBug]

In western countries, buyers decide what constitutes defect, and if there are enough of them you have a class action.

I think you might be on to something. Let's do it!

[cab]

Here's another removal

[OC Nub]

Ive had the IHS off for 2 months now, been running it on SS almost every day with no issues. It took less than a minute to remove it, the video above should be a "How not to" video.

[andressergio]

Ive had the IHS off for 2 months now, been running it on SS almost every day with no issues. It took less than a minute to remove it, the video above should be a "How not to" video.

i agree OC nub its very easy to do and i have no changes at all on temps

please DO NOT FOLLOW ABOVE with punches as with the smallest razor will come very easy that will lead to damage the die or other thing

[owikh84]

Yeah very easy to remove the IHS I just used this stationary knife for that job


Before:


After:


So that's around 5-8C improvement

[cab]

Those of you that removed it, was the original stuff liquid TIM like Arctic Silver or was it dry-solid stuff? If it's liquid-paste can't it get squished out of the way by heatsink pressure and then you could get an air gap between the IHS and die???

[OC Nub]

^ The paste was pretty hard. I removed the retention bracket from the board, not using the IHS.

[corross]

thermal conductivity

Coollaboratory Liquid Ultra: 38,4(W/(m∙K))
Coollaboratory Liquid Ultra: 32,6(W/(m∙K))
silver oxide based thermal compounds: ~10(W/(m∙K))
silicon based thermal compounds: ~2,8(W/(m∙K))
Sn96.5Ag3.0Cu0.5 solder, melting point 221°C: 85(W/(m∙K))

My tought is to put the chip back to the IHS/or a custom milled copper block by using solder. You can purchase Sn96.5Ag3.0Cu0.5 as solder paste with flux here:
http://circamicro.com/p-919087-smd29...f-5cc-syr.aspx

..and there's the cooking recipe - see page 2:
http://www.peromatic.ch/d/produkteli...01Feb07%29.pdf

[OldChap]

Rumour has it that the reason that intel chose to use paste rather than solder is that these tri-gate chips are not so good at handling the heat of the soldering process.

That said, there are some metals/solders with melting points below 200° such as indium (157° and heat co-efficient of 0.86) which may prove to be good candidates if you can set up to control the temperature of the process to just a little beyond this.

If you can determine the thickness needed in a good joint then it should be possible to apply the indium by weight (this will be less than 1 gram)

My understanding of this is that it does not react with silicon and it will wet surfaces naturally without fluxes (maybe someone will step in to confirm this)

Cheapest source I found was ingots of pure (.999) and in wire form (if you can find a small enough amount)

Solders containing indium are available but I know little about their melting temps and heat conductivity...... Maybe a indium gallium tin mix would be good as gallium has a much lower melting point although the thermal conductivity is only half as good too ........so good luck looking if you are interested

EDIT: A quick look suggests that for performance (heat transfer) and low temp one might consider a 97In/3Ag mix solder (melts at 143° heat co-efficient of 0.73) . From what I have read going much lower in melting temps adversly effects the heat co-efficient

http://www.indium.com/low-temperature-solders/

[Church]

OldChap: it's a bit hard to believe that rumor, when intel has rised temps at which throttling or shutoff starts. Imho simplest answer is - economy. At stock speeds/voltages IB even with awful TIM1 as is does better then SB, so why not save LOT of money knowing big volumes of cpus intel makes on probably 10$ cheaper TIM vs solderable one. Only problem is ruining our overclocking expectations with K cpus .. but my guess that those make up 1/10 of all IBs sold, so not worth to backout decision and not save on TIM1. BTW, it also leaves headroom for simple future i7 3870 release (3770 + intel's solderable TIM1 ), after all, purposedly gimping product for marketing differentiation decisions is not that foreign concept for computing hardware making.

[cab]

If I remember my thermodynamics, simple contact is all that's needed for heat transfer, so what's wrong with bare-metal? Both IHS and die are pretty flat from what I've seen. These CPUs don't make much heat anyway, and if you OC too much the die has not enough surface area to cope.

[OldChap]

Whichever way one looks at it, IF, and that is a big IF, there is some problem that could be encountered using higher temperature solders, reducing that temperature by some 50-60° can only be a good thing.

This ensures that the contact is indeed metal to metal (to metal) and imperfections are then of no consequence. The area available for heat transfer is fixed so beyond experiments with pastes, solders etc. we cannot improve on this.

[cab]

Whichever way one looks at it, IF, and that is a big IF, there is some problem that could be encountered using higher temperature solders, reducing that temperature by some 50-60° can only be a good thing.

This ensures that the contact is indeed metal to metal (to metal) and imperfections are then of no consequence. The area available for heat transfer is fixed so beyond experiments with pastes, solders etc. we cannot improve on this.

Thank you, are you a metallurgist? If the soldering fails, you have a $250-350 scrap cpu? Wouldn't that take the fun out of experimenting?

[andressergio]

i dont really understand why to solder? it works amazing just letting it loose retention mech takes care of the rest...and if you gonna sold it just use a lit glue or silicon to seal it

soldering for me is nonsense at all

[OldChap]

Thank you, are you a metallurgist? If the soldering fails, you have a $250-350 scrap cpu? Wouldn't that take the fun out of experimenting?

There you have me my friend, because that is the reason I have not taken the step to try this. Metallurgist? No. I'm a hobbyist

[cab]

There you have me my friend, because that is the reason I have not taken the step to try this. Metallurgist? No. I'm a hobbyist

Well then, one experiment worth doing is to compare stock (with paste) vs. bare metal. If bare metal does better, then I'd ask Intel for a replacement under consumer laws.

[Jonny5isalivetm]

How hard is it to get the Retention bracket off what tool did you guys use ?

Cheerz