For sure I've heard enough of horror stories to be aware of the issue, including first-hand description from my mate Visionary. But I decided that I had to try it with both the hard and soft approach. Basically, it was quite clear cut, as long as I could heat the heatspreader to the temperature the solder melts, I should be able to remove the heatspreader. We've thought of quite a few ideas, from boiling water to torches to TECs... in the end, I decided to go with a hot air gun.

My friend Vinnzz was helpful enough to lend me his, the real deal, Bosch, good control, focused heat, the works.

Of course, I was in no hurry to try this on the latest most powerful Conroe X6800 we have on hand. So I took out a Pentium 4 3.0E from my workstation PC and started cutting through the black rubber linings. It was easier than what it was on the Athlon 64s, and since all the SMD components are located on the back of the core, so I did not really have to worry too much about cutting into those.

And so in less than 5 minutes, it was clean of the rubber linings, all that is holding the heatspreader on is just the solder between the core and heatspreader. So I moved the processor onto a bench vice I used to clamp it on so that it would be easier to blow hot air over it and pry the IHS off quickly before the solder hardens.

I tried sandwiching in my "pry tool", a very thin flat head screwdriver and it subtly fit in between spreader and PCB. To buy some insurance for the integrity of the processor, especially the small SMD components on the back og the CPU, I squeezed the clamped processor against a sack of wet cloth so that the back doesn't get too hot.

My sources have told me that the solder used on the IHS typically melts at around 85 degrees celcius. So I intended to bring the heat spreader up to that temperature slowly, trying to pry off IHS softly in between. I needed a way to monitor the temperature. So I used a laser pointer temperature reader:

So I started heating up the heat spreader with the hot air gun, aiming and focusing on the center of the core itself. Went to 50C, tried prying. Went to around 60C, tried prying. Went to 75C tried prying. Went to 80C, pried again. It seemed to be loosening significantly, so I pried a bit more at the edges....

Another one to add to the statistics. How the hell did that happen? Did the core contact soften before the solder?? Well since the core is off anyhow, I decided to blast the hot air onto the removed heat spreader to see what temperature the solder "runs" (melts) at. Indeed, my source was quite correct, it gets softie at around 80C and at 85C to 90C it starts "running like butter on hot toast".

Rather convinced that I was close and that I was aiming too centered on the core such that the contacts got softened, I went on to try this on the last LGA CPU I had in my home, a Pentium 4 670. Moving fast, I had the rubber linings cut in a minute, making sure that the rubber lining was cut through all around so I just had to deal with the solder later.

I repeated the hot air process, moving the hot air blower around the entire IHS and not staying focused at the center spot. When my temperature reader read around 65C, I tried a few pries around. Still felt rather stiff. So I blew more hot air, till about 80C. Now one side felt really about to come off and the other side still felt tight. So I blasted the heat onto that side and tried prying again. Softer... went back to the first side, pried... came off...

Did I say one? Well make that two... Well it was a weird feelling having killed 2 CPUs within an hour. I am beginning to think this is just about as easy as a heart transplant. Thinking back, I would think that my main mistake is in prying. This was a habit I was accustomed to when removing IHS from AMD CPUs. Logic tells us that the core contact points with the PCB should not melt before the solder on the IHS so staying around the IHS solder temperature should not cause the core to fall off. The only reason I think is that I was prying when the solder was only slightly softened. The next time I try, I will not be doing any prying but will be pushing the IHS from one side. When the solder gets "runny" enough, the IHS should slide off the core. Let's hope at least. And next time, I probably should TRY to do this on a dead CPU :O

Meanwhile, I'll put the core back on the PCB. I think it should work again... what say you??