Removing the IHS from a C2D or C2Q

[DerwenArtos12]

For the record, I'm still running my A64 with it's IHS removed, have for a couple of years now. I have built multiple C2D and C2Q machines for clients but, have not really seen teh need for one of my own yet and therefore haven't tried to pry the IHS off one yet to see whats underneath. A quick googling didn't turn anything up so, I'm asking?

Why haven't you tried it yet?

 

[Tempered81]

I saw a broken e2140 with the IHS pried off. I heard it's soldered to the die? Either way the chip was already broken, so I don't know if it's a safe thing or not.

 

[DerwenArtos12]

Originally posted by: jaredpace
I saw a broken e2140 with the IHS pried off. I heard it's soldered to the die? Either way the chip was already broken, so I don't know if it's a safe thing or not.

Given that I'm very well related to someone who works for one of the companies that packages these chips I can say with a fair ammount of certainty that it's not soldered in the traditional sense. chemically, maybe but, last I heard they were using a ceramic epoxy to attach the IHS, could be wrong though, I have to be wrong at least twice a day.

 

[PliotronX]

I can confirm the E2 series, at least made last year, use the clay TIM which makes the process a whole lot easier. It requires a paper thin razer at first, there is a much smaller gap between the shroud of the IHS and the PCB compared to previous IHS I've removed. I found it more difficult to remove than the K8's, so extra care is needed. The higher end chips are the ones attached using aluminum oxide epoxy. Removal entails a blade at each corner, suspended in the air by the blades, and heating the center of the IHS with a torch so that the CPU drops from beneath the IHS (hopefully) before damage is done.

 

[DerwenArtos12]

Originally posted by: PliotronX
I can confirm the E2 series, at least made last year, use the clay TIM which makes the process a whole lot easier. It requires a paper thin razer at first, there is a much smaller gap between the shroud of the IHS and the PCB compared to previous IHS I've removed. I found it more difficult to remove than the K8's, so extra care is needed. The higher end chips are the ones attached using aluminum oxide epoxy. Removal entails a blade at each corner, suspended in the air by the blades, and heating the center of the IHS with a torch so that the CPU drops from beneath the IHS (hopefully) before damage is done.

ooh, yikes. Do you know where the line is that seperates the clay from the aluminum? Are we talking desktop vs server, 65nm versus 45nm, dual versus quad?

 

[PliotronX]

AFAIK so far only the Allendale chips have been found to use the clay, that'd be the 65nm E2 and E4 lines (likely the further neutered Celeron E1 series as well). Everything else should be presumed epoxied.

 

[DerwenArtos12]

Originally posted by: PliotronX
AFAIK so far only the Allendale chips have been found to use the clay, that'd be the 65nm E2 and E4 lines (likely the further neutered Celeron E1 series as well). Everything else should be presumed epoxied.

guess I'm going the right direction thiking of doing a DFI LP DK P35 and a 2160 or 2180 then.

 

[PliotronX]

Originally posted by: DerwenArtos12

Originally posted by: PliotronX
AFAIK so far only the Allendale chips have been found to use the clay, that'd be the 65nm E2 and E4 lines (likely the further neutered Celeron E1 series as well). Everything else should be presumed epoxied.

guess I'm going the right direction thiking of doing a DFI LP DK P35 and a 2160 or 2180 then.

Most certainly as far as IHS removal goes. Even I would have a hard time getting myself to take a blow torch to an IHS. The upside is that the epoxy seems to make a better joint thermally than the clay so lapping the concave surface that mates with the heatsink usually results in close to the same temps as running nekkid whereas the weakest link in the Allendales is the clay TIM. Also take note you will have to remove the locking plate that normally clamps the CPU onto the spring loaded pins in the socket. This is easily done by working one side out at a time (more info here). This means you will likely have to remove the annoying plastic pins that come with most 775 coolers and use some 4/40" or 6/32" machine screws, nuts, and nylon washers for insulation.

 

[DerwenArtos12]

Originally posted by: PliotronX

Originally posted by: DerwenArtos12

Originally posted by: PliotronX
AFAIK so far only the Allendale chips have been found to use the clay, that'd be the 65nm E2 and E4 lines (likely the further neutered Celeron E1 series as well). Everything else should be presumed epoxied.

guess I'm going the right direction thiking of doing a DFI LP DK P35 and a 2160 or 2180 then.

Most certainly as far as IHS removal goes. Even I would have a hard time getting myself to take a blow torch to an IHS. The upside is that the epoxy seems to make a better joint thermally than the clay so lapping the concave surface that mates with the heatsink usually results in close to the same temps as running nekkid whereas the weakest link in the Allendales is the clay TIM. Also take note you will have to remove the locking plate that normally clamps the CPU onto the spring loaded pins in the socket. This is easily done by working one side out at a time (more info here). This means you will likely have to remove the annoying plastic pins that come with most 775 coolers and use some 4/40" or 6/32" machine screws, nuts, and nylon washers for insulation.

That step alone makes me slightly more nervous than anything else to date. I'm in the process of building a water chiller so I'll be using the bolts that come with watever waterblock I get for it. what the heck right? I wonder if it's really worthwhile if it's going to be run at 0C anyways? So what if they're a cooling loss through the ceramic, the cores will run at what 10C? I'll just crank down the temp on my chiller to -10 and get the cores to 0. I'm thinking it wouldn't be worth it at this point. I'm not so worried about the warranty on my motherboard as it'll be slathered with dielectric grease and liquid electrical tape anyway but, I really think I don't trust not having something, anything to hold the cpu in the socket besides my waterblock.

 

[aigomorla]

i heard that solder is indium. And ruby also heard the same thing.

Ruby and i however have mixed views on the melting point of that solder.

On Xbit someone recorded the melting point to be about 73C. As you guys know the threshold for the cpu is much higher.

But im going to assume you'll need about 80C on that ihs for enough heat to transfer though and melt it. Eitherway its still considered safe. However i dont think i want to try it.

Can anyone double confirm that the E2180 is clay? I lost my E8400 on friday. :[ I wanted an IHSless allendale since there cheap to hold me off until the E8500 or E8600. I got dualcore fever in overclocking now. *sigh*

Quads are gonna have to wait until i get my QX9650. :T

 

[Diogenes2]

If you must do this, why not just sand ( lap ) down to the core ?

 

[VirtualLarry]

Originally posted by: aigomorla
Can anyone double confirm that the E2180 is clay? I lost my E8400 on friday. :[ I wanted an IHSless allendale since there cheap to hold me off until the E8500 or E8600. I got dualcore fever in overclocking now. *sigh*

So 1.57v finally killed it, eh?

 

[aigomorla]

Originally posted by: VirtualLarry

Originally posted by: aigomorla
Can anyone double confirm that the E2180 is clay? I lost my E8400 on friday. :[ I wanted an IHSless allendale since there cheap to hold me off until the E8500 or E8600. I got dualcore fever in overclocking now. *sigh*

So 1.57v finally killed it, eh?

Pretty much... :T

it was fun while it lasted. Definitely stepped into phase territory on that chip.

 

[DerwenArtos12]

Originally posted by: aigomorla
i heard that solder is indium. And ruby also heard the same thing.

Ruby and i however have mixed views on the melting point of that solder.

On Xbit someone recorded the melting point to be about 73C. As you guys know the threshold for the cpu is much higher.

But im going to assume you'll need about 80C on that ihs for enough heat to transfer though and melt it. Eitherway its still considered safe. However i dont think i want to try it.

Can anyone double confirm that the E2180 is clay? I lost my E8400 on friday. :[ I wanted an IHSless allendale since there cheap to hold me off until the E8500 or E8600. I got dualcore fever in overclocking now. *sigh*

Quads are gonna have to wait until i get my QX9650. :T

I'll see what all the engineer knows at my source. He only gets a certain ammount of information on what he's working on as he only helps design the machines that make the heatspreaders so he's pretty far out of the loop on what is actually going on further down the line. He was very sure it was a ceramic or clay though based on the chemical composition designed into the heatspreader itself. An aluminum solder would not have held well in the long term in. I'll have to ask about indium as that may be a different case all together.

 

[PliotronX]

I'll be a monkey's uncle. No need to bother your buddy, because Intel holds a patent for coupling IHS with semiconductors using indium solder. I've always thought of solder as forming an electrical joint between two metallic surfaces, not caring much for flowing over glass Here is Intel's page about their use of indium. It mentions a "low" melting point as 157 degrees :Q

 

[DerwenArtos12]

Originally posted by: PliotronX
I'll be a monkey's uncle. No need to bother your buddy, because Intel holds a patent for coupling IHS with semiconductors using indium solder. I've always thought of solder as forming an electrical joint between two metallic surfaces, not caring much for flowing over glass Here is Intel's page about their use of indium. It mentions a "low" melting point as 157 degrees :Q

if thats Fahrenheit then thats only 69.5C.

 

[PliotronX]

Oh sorry, they qualified it as 157 Celsius. Although I wonder if they meant Fahrenheit as 69.5 would fit in with aigomorla's experience.

 

[DerwenArtos12]

Originally posted by: PliotronX
Oh sorry, they qualified it as 157 Celsius. Although I wonder if they meant Fahrenheit as 69.5 would fit in with aigomorla's experience.

157C would very certainly cause damange to the core and burn the silicon packaging, so I tend to think it's either a misprint or mis-information.

 

[Idontcare]

Originally posted by: DerwenArtos12

Originally posted by: PliotronX
Oh sorry, they qualified it as 157 Celsius. Although I wonder if they meant Fahrenheit as 69.5 would fit in with aigomorla's experience.

157C would very certainly cause damange to the core and burn the silicon packaging, so I tend to think it's either a misprint or mis-information.

They likely don't heat the entire package to the solder temp. Just as when you solder anything else, only very local heating is applied (typically). Same with welding. You don't heat up the whole skyscrapper to 1500C just to weld two beams together.

157C is certainly in the right ballpark. To get it any lower you'd need to add gallium, a lot of gallium.

http://www.indium.com/products/sorted_by_temp.pdf

 

[ixtapalapaquetl]

I removed the IHS from my e8400. I have lots of experience with removing IHS's on AMD CPU's, but was a little nervous with all this talk of "solder". The reality was far from the hype. I used a razor to cut through the black rubbery adhesive, starting at the corners then working toward the middles. My plan was to heat it with an oven pan to melt the "solder". But before I got to that point, it popped off on its own just from incidental pressure from the razor blade. It is my conclusion that "solder" is far too strong of a term. I would go with "mild adhesive" - at least in my case.

Unfortunately, the cooling effects were negligible. Mounting without the clasp on the CPU holder is a pain. All in all, I am glad I did it, but am not sure it was worth the effort.

 

[DerwenArtos12]

Originally posted by: ixtapalapaquetl
I removed the IHS from my e8400. I have lots of experience with removing IHS's on AMD CPU's, but was a little nervous with all this talk of "solder". The reality was far from the hype. I used a razor to cut through the black rubbery adhesive, starting at the corners then working toward the middles. My plan was to heat it with an oven pan to melt the "solder". But before I got to that point, it popped off on its own just from incidental pressure from the razor blade. It is my conclusion that "solder" is far too strong of a term. I would go with "mild adhesive" - at least in my case.

Unfortunately, the cooling effects were negligible. Mounting without the clasp on the CPU holder is a pain. All in all, I am glad I did it, but am not sure it was worth the effort.

Thank you for that. Very informational. I doubt I'm even going to bother given the mounting post-removal. It'll be running at freezing anyways, what's 5-6 degrees when it's that cold anyways?