Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz

Page 8 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

HondaCop

Member
Aug 4, 2012
42
0
0
I just finished glueing my IHS to the pcb with some automotive, black silicone. It's the Permatex Black Silicone Adhesive Sealant to be exact.

I placed a thin bead all along inside edge of the IHS, the edge where there is always leftover glue from the original Intel adhesive. When I pressed the IHS firmly into the pcb, no silicone whatsover spread out under the IHS ans thus now the chip looks stock.

From what I can see, the IHS looks even and no different from when I purchased it. I will give it 24 hours and then I will mess around with it to see if it holds together strongly.

BTW, the silicone is rated up to 230C. No problems with the CPU's heat. ;-)

- Grammar *censored*, this was sent from my phone.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
I just finished glueing my IHS to the pcb with some automotive, black silicone. It's the Permatex Black Silicone Adhesive Sealant to be exact.

I placed a thin bead all along inside edge of the IHS, the edge where there is always leftover glue from the original Intel adhesive. When I pressed the IHS firmly into the pcb, no silicone whatsover spread out under the IHS ans thus now the chip looks stock.

From what I can see, the IHS looks even and no different from when I purchased it. I will give it 24 hours and then I will mess around with it to see if it holds together strongly.

BTW, the silicone is rated up to 230C. No problems with the CPU's heat. ;-)

- Grammar *censored*, this was sent from my phone.
Unless I'm mistaken, the stuff in the tube you're using has something like an acetic acid base. It sure smells like it. Once it dries -- no problem. But I had thought about that -- silicone adhesive-sealant -- and was thinking to incline myself more toward the "Pit Crew" adhesive for the reason I mention here. If it doesn't do anything to the green coating on the PCB, it's probably fine.

I notice here that everyone participating in this thread -- LIKE ME -- is second-guessing, picking fly-shit out of pepper and anticipating every little thing that could damage this $350 CPU.

Which . . . . is probably a good thing . . . I guess . . .
 

HondaCop

Member
Aug 4, 2012
42
0
0
Unless I'm mistaken, the stuff in the tube you're using has something like an acetic acid base. It sure smells like it. Once it dries -- no problem. But I had thought about that -- silicone adhesive-sealant -- and was thinking to incline myself more toward the "Pit Crew" adhesive for the reason I mention here. If it doesn't do anything to the green coating on the PCB, it's probably fine.

I notice here that everyone participating in this thread -- LIKE ME -- is second-guessing, picking fly-shit out of pepper and anticipating every little thing that could damage this $350 CPU.

Which . . . . is probably a good thing . . . I guess . . .
We are the guinea pigs! :D This silicone is low odor and I used a minimal amount. As a matter of fact, I can't smell a thing. I'll tell you tomorrow if it holds together well, though visually, you would never think that CPU was de-lidded.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
We are the guinea pigs! :D This silicone is low odor and I used a minimal amount. As a matter of fact, I can't smell a thing. I'll tell you tomorrow if it holds together well, though visually, you would never think that CPU was de-lidded.
I'd never seen it damage a painted surface -- or anything, for that matter. I just thought about the vinegary smell and the acetic acid because we're all goosey paranoid about de-lidding and "what's under the lid." Once dry, it would just peel off like rubber. I doubt that it would be as easily removeable as the original adhesive, demonstrated by IDC, his dremel and Isopropyl.

As for "sealing" the IHS, I just figured nobody would do that. IDC had his misgivings about using water for lapping, given the gap in the bottom of the IHS. My thought was to apply some adhesive sparingly to replicate the original "Intel job."
 
Last edited:

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
I was thinking that replacing the original adhesive with a new adhesive having similar properties would allow for close contact with the die, while providing some protection against excessive pressure of the IHS spring-loaded assembly. I wasn't thinking that Intel did much different than what I describe, although maybe they did.
You can bet that in Intel's production line they have a specific pressure they apply during the IHS seating, a pressure that was dialed in to give them the IHS-to-CPU gap they desired while being repeatible within minimal variation across samples.

If you knew that pressure and replicated it when setting your IHS then you'd probably be getting close enough to be good enough.

If you plan to just wing-it and go by instinct, which is about all you can do (its all I'd imagine myself doing), then its still just a guess as to whether or not you left a large enough gap to prevent the IHS from sitting on the die, squeezing out all the TIM in the process (as happened in my first round of testing after delidding).

The challenge is doubly so because you have no way of inspecting the internals after you re-adhere the IHS.

I suppose there is one way to do it, you build your own calibration table of known shim gap versus operating temperature prior to reglueing the IHS down, then glue it down and take a temperature measurement. THen look on the graph of calibration data and you'll know your final gap height. If its too small for comfort then you delid and try again with less pressure and/or more adhesive.

In fact I think that would be a cool experiment. Need more shim thicknesses though.

BTW, I want to buy some of that IC Diamond stuff for my tests, where is a good place to get it from? Preferable a vendor who will sell it for $10 a tube without charging another $10 S/H :confused:
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
You can bet that in Intel's production line they have a specific pressure they apply during the IHS seating, a pressure that was dialed in to give them the IHS-to-CPU gap they desired while being repeatible within minimal variation across samples.

If you knew that pressure and replicated it when setting your IHS then you'd probably be getting close enough to be good enough.

If you plan to just wing-it and go by instinct, which is about all you can do (its all I'd imagine myself doing), then its still just a guess as to whether or not you left a large enough gap to prevent the IHS from sitting on the die, squeezing out all the TIM in the process (as happened in my first round of testing after delidding).

The challenge is doubly so because you have no way of inspecting the internals after you re-adhere the IHS.

I suppose there is one way to do it, you build your own calibration table of known shim gap versus operating temperature prior to reglueing the IHS down, then glue it down and take a temperature measurement. THen look on the graph of calibration data and you'll know your final gap height. If its too small for comfort then you delid and try again with less pressure and/or more adhesive.

In fact I think that would be a cool experiment. Need more shim thicknesses though.

BTW, I want to buy some of that IC Diamond stuff for my tests, where is a good place to get it from? Preferable a vendor who will sell it for $10 a tube without charging another $10 S/H :confused:
www.heatsinkfactory.com

www.sidewindercomputers.com

Last I looked, it was about a $1 per carat . . . Lemme see . . . Sidewinder has it now for $5.99. Don't know about the shipping/handling. The "factory" has it priced close to $7.

You could try making it yourself, which I did, but you'd have to buy the diamond-dust from an industrial supplier. Then make yourself a little mixing container with a paper-clip paddle which you'd run through a hole in the container-top and turn around like a hand crank with the dust and some run-of-the-mill silicon-grease. If the diamond "loading" is up to snuff -- as thick as ICD7 -- you'll get about the same result.

But -- really -- it's still cheap, even if you only get enough for four or five applications . . .
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
If you plan to just wing-it and go by instinct, which is about all you can do (its all I'd imagine myself doing), then its still just a guess as to whether or not you left a large enough gap to prevent the IHS from sitting on the die, squeezing out all the TIM in the process (as happened in my first round of testing after delidding).
Of course . .. instinct . . . It depends on the TIM whether it would "squeeze out." My thought was that you could just replace the cap with a thin bead of adhesive on the flange until you could just feel the TIM make contact. With a rubbery substance like the automotive silicone adhesive or the "Pit Crew," it would just solidify and you'd probably come close to what they do at the factory . . . or so I might suppose.

Maybe the shim is the better idea. But out of all this trouble you've gone through, we could take any contributions here and make a "handbook-guide" to IB de-lidding and TIM replacement. You'd want to suggest the simplest operations and resources . . . then write your disclaimer "Not responsible if the reader crushes his IB CPU" . . . . :biggrin:
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
www.heatsinkfactory.com

www.sidewindercomputers.com

Last I looked, it was about a $1 per carat . . . Lemme see . . . Sidewinder has it now for $5.99. Don't know about the shipping/handling. The "factory" has it priced close to $7.

You could try making it yourself, which I did, but you'd have to buy the diamond-dust from an industrial supplier. Then make yourself a little mixing container with a paper-clip paddle which you'd run through a hole in the container-top and turn around like a hand crank with the dust and some run-of-the-mill silicon-grease. If the diamond "loading" is up to snuff -- as thick as ICD7 -- you'll get about the same result.

But -- really -- it's still cheap, even if you only get enough for four or five applications . . .
Cool, ordered the 5g tube from sidewinder for $20 shipped. Thanks :)
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
Been VERY happy with IC Diamond since I started using it a little over a year ago. Great stuff.
If God made us serve time in Purgatory just for doing stupid things, I'd get a thousand years. I was so "enthusiast-ically" wrapped up in experimenting with cooling, that I dropped about $80 on a little vial of industrial diamond powder. And the reason it was more than stupid, I don't think it was synthetic stuff, or it would've been cheaper.

Some people are miffed at how thick ICD is. If you try to spread it too fast, you'll tear holes in the layer. I use a razor-blade, and take my silly-a**ed time. The only thing that performs better is any one of the metal products with gallium, indium, etc.

As I said before -- if you are going to use it between the IHS and silicon die, you'd best spread it on the mating surface or bottom of the IHS -- try and get an even, consistent layer . . . . maybe even put a drop of cheap silicone-grease on top of that. then -- mate the surfaces.
 

AtenRa

Lifer
Feb 2, 2009
13,772
2,932
136
IDC i would like to ask if you could measure the Core i7 2600K at default settings and the Core i7 3770K (better TIM between the die and IHS) at default settings but with the default Heat-sink+fan in both of them .

I would like to see whats the impact of the smaller die size in operational temps if any at default settings and cooling solutions.
 

Diogenes2

Platinum Member
Jul 26, 2001
2,151
0
0
....

BTW, I want to buy some of that IC Diamond stuff for my tests, where is a good place to get it from? Preferable a vendor who will sell it for $10 a tube without charging another $10 S/H :confused:
MicroCenter sells it for $6.99 ..
 

PlasmaBomb

Lifer
Nov 19, 2004
11,815
1
81
You can bet that in Intel's production line they have a specific pressure they apply during the IHS seating, a pressure that was dialed in to give them the IHS-to-CPU gap they desired while being repeatible within minimal variation across samples.

If you knew that pressure and replicated it when setting your IHS then you'd probably be getting close enough to be good enough.

If you plan to just wing-it and go by instinct, which is about all you can do (its all I'd imagine myself doing), then its still just a guess as to whether or not you left a large enough gap to prevent the IHS from sitting on the die, squeezing out all the TIM in the process (as happened in my first round of testing after delidding).

The challenge is doubly so because you have no way of inspecting the internals after you re-adhere the IHS.

I suppose there is one way to do it, you build your own calibration table of known shim gap versus operating temperature prior to reglueing the IHS down, then glue it down and take a temperature measurement. THen look on the graph of calibration data and you'll know your final gap height. If its too small for comfort then you delid and try again with less pressure and/or more adhesive.

In fact I think that would be a cool experiment. Need more shim thicknesses though.

BTW, I want to buy some of that IC Diamond stuff for my tests, where is a good place to get it from? Preferable a vendor who will sell it for $10 a tube without charging another $10 S/H :confused:
Could you not just measure the total height before delidding and after and compare the two?
 

MadScientist

Platinum Member
Jul 15, 2001
2,122
31
91
I just finished glueing my IHS to the pcb with some automotive, black silicone. It's the Permatex Black Silicone Adhesive Sealant to be exact.
This would be my adhesive choice. Great stuff. Use it for multiple purposes, mostly on my cars. It contains stearic not acetic acid. Stearic acid is a much weaker acid than acetic. It's used in foods, detergents, cosmetics, shampoos, etc.
http://www.permatex.com/documents/msds/01_USA-English/81158.pdf
 
Last edited:

HondaCop

Member
Aug 4, 2012
42
0
0
This would be my adhesive choice. Great stuff. Use it for multiple purposes, mostly on my cars. It contains stearic not acetic acid. Stearic acid is a much weaker acid than acetic. It's used in foods, detergents, cosmetics, shampoos, etc.
http://www.permatex.com/documents/msds/01_USA-English/81158.pdf
Yup and the results were astonishing. After 24 hours, the IHS is as firm as it was when it came from Intel. And if you want to de-lid it again, you can use a blade to cut it too.

- Grammar *censored*, this was sent from my phone.
 

felang

Senior member
Feb 17, 2007
594
1
81
Great Job guys!

You´re making me want to get an Ivy just so I can try this, hopefully without destroying it. Although then I´d want to get a z77 board to go with it and some high speed ddr3 and... my wallet would be taking a huge hit... again.
 

Denithor

Diamond Member
Apr 11, 2004
6,300
23
81
IDC i would like to ask if you could measure the Core i7 2600K at default settings and the Core i7 3770K (better TIM between the die and IHS) at default settings but with the default Heat-sink+fan in both of them .

I would like to see whats the impact of the smaller die size in operational temps if any at default settings and cooling solutions.
One problem - you cannot get an apples-to-apples comparison between Sandy and Ivy because the cooling interface is different. The solder under the IHS on Sandy transports heat away from the cores at a different rate than the TIM (of whatever variety, whether stock or aftermarket) under the IHS on Ivy.

The only way to get a true comparison would be to also de-lid a Sandy processor and apply the same TIM you were using on Ivy. You would also have to make sure to set the gaps exactly the same, etc. Then set speeds to the same, use the same cooler and see how the heat transfer differs.

EDIT: Actually, upon thinking a bit further, you would be best just taking the whole IHS out of the picture. Run SB and IB cores naked, same TIM direct to the same cooler and monitor temps under those conditions. That would yield the best picture of how much (if any) the thermal density inhibits heat flow out of the cores.
 
Last edited:

Rvenger

Elite Member <br> Super Moderator <br> Video Cards
Super Moderator
Apr 6, 2004
6,283
5
81
IDC i would like to ask if you could measure the Core i7 2600K at default settings and the Core i7 3770K (better TIM between the die and IHS) at default settings but with the default Heat-sink+fan in both of them .

I would like to see whats the impact of the smaller die size in operational temps if any at default settings and cooling solutions.

I am interested in hearing this too. Ideally the same clockspeeds and voltage. If IVB runs cooler, it would invalidate Intel's statement.
 
Last edited:

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
This would be my adhesive choice. Great stuff. Use it for multiple purposes, mostly on my cars. It contains stearic not acetic acid. Stearic acid is a much weaker acid than acetic. It's used in foods, detergents, cosmetics, shampoos, etc.
http://www.permatex.com/documents/msds/01_USA-English/81158.pdf
That's interesting! I made the assumption off the top of my head because it smelled like vinegar! And -- yeah -- you'd think that Permatex might have the ingredient-spec at their web-site.

I've used the stuff when I had to solder fan wires or LEDs -- just for insulation. Takes a little "artful" application, because it leaves a black (or clear) lump on the wire.

IDC is concerned about precision, which is a good thing. I felt that since the stuff hardens to stiff silicone-rubber, and IF it doesn't expand in hardening (not likely), then it would counteract the pressure of the HSF spring-loaded assembly. If it set while the IHS was close enough to the die to accommodate a thin layer of (whichever) TIM is chosen, you would get the desired combination. OF course -- you don't know until afterward, and any mistakes would mean going through the entire razor-blade de-lidding process again . . .
 

Rvenger

Elite Member <br> Super Moderator <br> Video Cards
Super Moderator
Apr 6, 2004
6,283
5
81
IDC, if you want I can donate you my Antec Formula 7 TIM. LMK if you are interested.
 

AtenRa

Lifer
Feb 2, 2009
13,772
2,932
136
One problem - you cannot get an apples-to-apples comparison between Sandy and Ivy because the cooling interface is different. The solder under the IHS on Sandy transports heat away from the cores at a different rate than the TIM (of whatever variety, whether stock or aftermarket) under the IHS on Ivy.

The only way to get a true comparison would be to also de-lid a Sandy processor and apply the same TIM you were using on Ivy. You would also have to make sure to set the gaps exactly the same, etc. Then set speeds to the same, use the same cooler and see how the heat transfer differs.

I am aware that SB IHS is soldered with the die and cannot be delidded. I believe that even with a better TIM at default settings the IvyBridge temps will be close to the SB temps (if not lower).


EDIT: Actually, upon thinking a bit further, you would be best just taking the whole IHS out of the picture. Run SB and IB cores naked, same TIM direct to the same cooler and monitor temps under those conditions. That would yield the best picture of how much (if any) the thermal density inhibits heat flow out of the cores.
If we could do that and install the same default Intel heat-sink on the die without the IHS, im sure that IvyBridge would have higher temps than SB due to smaller die. The IHS plays a significant part by spreading the heat evenly across the entire IHS surface making it easier for the Heeat-sink to cool the die.
 

Yuriman

Diamond Member
Jun 25, 2004
5,530
141
106
If we could do that and install the same default Intel heat-sink on the die without the IHS, im sure that IvyBridge would have higher temps than SB due to smaller die. The IHS plays a significant part by spreading the heat evenly across the entire IHS surface making it easier for the Heeat-sink to cool the die.
This is nonsense.

"Let's transfer heat to the copper cooling plate more quickly by putting a second copper cooling plate between the first one and the die."

"Heatspreaders" are made of the same materials as heatsinks and don't spread out heat any better than the bottom of a heatsink would. "Lid" or "die protector" I think would be more politically correct terms.
 

HondaCop

Member
Aug 4, 2012
42
0
0
This is nonsense.

"Let's transfer heat to the copper cooling plate more quickly by putting a second copper cooling plate between the first one and the die."

"Heatspreaders" are made of the same materials as heatsinks and don't spread out heat any better than the bottom of a heatsink would. "Lid" or "die protector" I think would be more politically correct terms.
There is a reason why Intel calls them heat spreaders and not die covers or die protectors.

- Grammar *censored*, this was sent from my phone.
 

BonzaiDuck

Lifer
Jun 30, 2004
15,308
1,187
126
There is a reason why Intel calls them heat spreaders and not die covers or die protectors.

- Grammar *censored*, this was sent from my phone.
Whatever they call it, it doesn't matter. It is (1) an extra thickness of nickel-plated copper between the die and the heatsink, (2) it requires two layers of thermal interface material instead of one. All of these things reduce the transfer of heat per unit of time into heatpipes or a waterblock.

Intel knows this; we know this. Whatever Intel wants to call the IHS, it's there to protect the die first, and "spread" the heat as second priority.
 

ASK THE COMMUNITY