MP 1,1-5,1 Safest delidding process

[macsforme]

Sorry to bring up a topic again that has been beaten to death...

I had one of my dual X5677 CPUs fail (lost a few RAM slots) a while ago, so I am trying to replace it. They run about $15 each on eBay, so I figured it would be no problem. I somehow just killed five Intel X5677 CPUs in a row trying to delid them to get one working one...

The first two I used the razor blades and butane torch method (which I used to delid both of my original X5677s a while back), but both CPUs failed. I ordered three more from a different seller, and this time I tried the clothes iron method after hearing that this is gentler with the heat. Two of them failed, then on the third I reverted to the butane torch method again, and that one failed as well. Most of the CPUs I tested beforehand and verified that they were working properly before the delid process.

I figure the problem must have to do with the heat, or else the bowing of the board the CPU chip is mounted to when the razor blades are inserted (is does bend quite a bit). I tried only using two razor blades at a time, and I tried working more slowly with the heat, but I’m still just killing the chip every time.

When I attempt to use these CPUs after the delidding process, the machine just doesn’t chime, and sometimes but not always it shows the red CPU fault LED. Also, sometimes but not always the backplane shows the CPU overtemp error light when I try to boot up.

Are these CPUs now just so fragile after so many years of use that they can’t tolerate the heat while desoldering the IHS? Or is there likely something wrong with my process? I see there are other methods for delidding, like the vise/wood block, and the special screw tool designed for this purpose, but can you even use those processes on CPUs that have the IHS soldered down?

 

[fhturner]

Yes, you absolutely can. Vise method has worked great for me a half dozen times or more.

 

[JedNZ]

Vice method has worked for me every time. I actually broke my cheap vice last time, so I bought a more solid one yesterday and am about to do another two X5680s for my newest cMP project. Hopefully the new vice stands up to the pressures better (it’s made of metal, not cheap alloy).

I used a special green tape (made for sticking to components - easily peels off without leaving glue residue) to border each CPU side, the edges and the precious rear (leaving the top IHS exposed) so it’s protected from any knocks or unintended scratches etc.

 

[Macschrauber]

Vice method plus some heat (300 celsius) to warm up the solder.

About 10 Xeons, including the ones to practice, no fail.

Guess the problem of the TS is somewhere else. Maybe still damaged pins on the socket.

Inspect it very closely. Macro camera, microskope, something alike.

 

[Enricote]

My experience with a very simple vice and the procedure described by SolidCake was Excellent. Easier than I thought. I simply put my processors on the heating radiator for a couple of hours (wrapped in a cotton cloth) before using the vice. I don't know if that helps or not, but I did it.

Look at his video:

Simple things use to be the best. This procedure is simple and cheap, so it is excellent.
Regards from Spain

 

[Slash-2CPU]

Vice, but you have to go the right direction. Pay close attention to that. Right direction almost guarantees a perfect deld. The other three possibilities guarantee you'll destroy the chip.

 

[s.m.t.]

I just bought a pair already delidded... Didn't want to risk it, the price most people charge to delid is less than the cost of one 5690. Break one and you might as well have paid for a service...

 

[majus]

What kind of adhesive is the heat shield attached with?

 

[jscipione]

What kind of adhesive is the heat shield attached with?

Vice method worked for me. Heat shield is attached with solder. You'll have to shave off the excess solder with a razor blade after removing the heat shield, which is the trickiest part.

 

[tpivette89]

Vice method worked for me. I've done a pair of X5690s, 4 x X5677s, a pair of X5650s, and am about to do a pair of X5675s tomorrow.

I did have a few failures with X5677s, but this was due to an inferior vice. After switching to a better quality one, no problems at all.

 

[Slash-2CPU]

What kind of adhesive is the heat shield attached with?

It's soldered to the CPU die with a very soft low-temperature solder. You can cut into it with a fingernail.

The heat shield is also glued to the green CPU substrate around the edges with black silicone-type adhesive.

 

[macsforme]

Vice, but you have to go the right direction. Pay close attention to that. Right direction almost guarantees a perfect deld. The other three possibilities guarantee you'll destroy the chip.

Which direction is the correct one?

 

[MIKX]

I used four razors but the most important thing using these is to ANGLE THE BLADES 5° UPWARDS. This obviates slicing anything important.

Then I used a standard butane lighter to warm up the solder.

The lid just dropped off.

 

[JedNZ]

Post #5 above shows the correct position. Basically, with the IHS facing downwards, and the notches on the sides closest to you, the left side edge of the CPU will be towards the top edge of the left side of the vice, and the right side of the IHS will be on the edge of the right side of the vice. If that doesn't make sense, check the video.

I did mine two X5680s today. First one was a bit stubborn (soft release of the IHS), and the second one put up less of a fight (came off with a ping).

Tomorrow I'll install them in my dual cMP 4,1>5,1 8-Core (2.26GHz) tray. I'll check out the Northbridge rivets while I'm giving it a clean up.

 

[macsforme]

Thanks, guys. I bought a $20 vise from Home Depot and followed the instructions in the video to pop off the IHS, and the CPU now works with no problem!

 

[Stormageddon]

Hi! I am new here – I registered finally with the sole purpose of trying to find out: I just delidded two x5675s for my Mac Pro 5,1(originally 4,1). I used the vice method, but in the process of detaching the black silicon, I scraped a capacitor on both of them. I find mixed opinions about capacitor damage, some people seem to have experiences where cpu:s may work very well even with capacitors missing. In some cases the opinion is that even a slight damage will render the cpu useless.

Other than the small scrapes, I did a reasonable job. However, I am in dire need of a definitive answer, if I try them on, and should they be damaged beyond operating, can they break something else in my tray?

Opinions off course are also most welcome!

EDIT: this is the worst damage – more than a scratch..

 

[MIKX]

Post #13

4 razor blades ANGLED SLIGHTLY UPWARDS to avoid cutting anything important.

Work slowly, be patient, take a few coffee breaks if needed.

Then you can use a butane lighter while wiggling the razor blades with a pair of pliers over a rigged frame.

My IHSs just fell off onto a soft cloth.

I don't trust the vice method. Too forceful for my taste.

 

[JedNZ]

I too have touched (applied pressure against them with the blade) the capacitors while scraping off the black silicon gasket. No visible damage and I’ve never had any issues.

If you knocked the capacitor lid off one then I’m pretty sure that would render it unusable. But a slight knock like I’ve done should be fine.

 

[Macschrauber]

Measure if the remaining thing makes a short circuit (hard to do but this is no excuse). You will need a microscope or a very good camera.

If its shortcut then scrape further until not.

Noone knows what happened with this thing.

And of course there is a chance to damage the board with damaged CPUs.

There is also the chance that not.

Most of the caps will "clean" voltages, so its not too catastrophical if one misses, but who knows...

 

[Snow Tiger]

The safest "delidding" method of a processor is the one you never need to perform . There is no thermal benefit to using integrated heat spreader-less processors in cMPs , anyways . I exhaustively looked into it . Apple has a long history of developing mostly failed but amazingly creative methods of cooling HEDT and personal workstations . As a builder , I have to deal with and even improve Apple's factory cooling methods .

The safest way of upgrading the processors of a Dual 2009 cMP is to install off the shelf Xeons ( with factory heatspreaders still retained ) and add 2 mm nylon washers on all eight CPU heatsink standoffs , add 2 mm of thermal gap material on both heatsinks and liberate both heatsink's fan connectors so that they can be independently attached to the Tray's PCB connectors . That's it . No need to worry about damaging anything . In case you wish to get technical , there is precisely a 2.3 mm difference in thickness between the factory's processors and a retail X56xx Xeon .

You made a good choice with the upgrade processors . Cheap and powerful . I built the first Dual X5677 cMP years ago and it was used by a CAD CAM machine designing motors for construction vehicles in Texas .

 

[Stormageddon]

Thank you for your replies. Even though I was equipped with second hand -knowledge of the delidding process after reviewing many videos about it, there were some surprises. The bit with the vice did not seem at all risky, in fact?! Cutting the black glue was - for me anyway. I wasn't really sure whether the black glue was the one maintaining the hold after the ihs started moving, or did the solder maintain some elasticity and hold. That was a surprise that probably made me do razor-work more than necessary.

I placed an order for two more of the same x5675s and will delid them in a few weeks and probably choose the two most perfect results for installing.

Having said that, I do very much value any further input re the capacitor damage – I added an image about it in the original post #16.

 

[tsialex]

Thank you for your replies. Even though I was equipped with second hand -knowledge of the delidding process after reviewing many videos about it, there were some surprises. The bit with the vice did not seem at all risky, in fact?! Cutting the black glue was - for me anyway. I wasn't really sure whether the black glue was the one maintaining the hold after the ihs started moving, or did the solder maintain some elasticity and hold. That was a surprise that probably made me do razor-work more than necessary.

I placed an order for two more of the same x5675s and will delid them in a few weeks and probably choose the two most perfect results for installing.

Having said that, I do very much value any further input re the capacitor damage – I added an image about it in the original post #16.

You ripped off a capacitor almost completely. Even with that, maybe it can work, but since dual trays are so expensive I'd only test with a single CPU tray.

 

[Macschrauber]

For a ruff test I just pop in the CPU without heat sink, if te mac chimes and the efi done LED is lit when pressing diagnose the cpu is at least not broken.

Works of course not for dual 4.1 as theres no pressure to give contact.

Dont run the Cpu more than a few seconds without heat sink.

Only can second that, a single cpu tray, even without heat sink is a good test tool. As a test tool dont matter than if 4.1 or 5.1 - if fans go crazy.

 

[Stormageddon]

Got a new pair of X5675. Delidded them with no damage visible to the eye. Tried them on, but wasn't able to get a chime out of them. I changed the 2,26 gig cpus back, had a bit of trouble finding the right tightness, but finally got the cMP to boot. With the x5675s I got either a steady white led light or a blinking one, but no chime. I think I tried both of them in the A slot.

Now: as to delidding, a tip: after removing the lid from 4 x5675s, I think I can say that the process of the lid coming loose is not straightforward. It doesn't just snap off. Instead it seems that the solder has some elasticity and you have to go back and forth for several times with the vice, until the bond is finally loose. At least with the x5675s this seems to be the case, perhaps with others?

Here is a question: with the structure of the 2009 dual tray, with delidded cpus, is it a realistic scenario that one can go so tight with the heatsink, that one can break the tray? My feeling was that when eventually I managed to get both old cpus online (as well as all the ram), I perhaps had to go tighter than I had gone with the x5675s. One of these days, I think I´ll give it another go, but would love to know more.

 

[Snow Tiger]

Got a new pair of X5675. Delidded them with no damage visible to the eye. Tried them on, but wasn't able to get a chime out of them. I changed the 2,26 gig cpus back, had a bit of trouble finding the right tightness, but finally got the cMP to boot. With the x5675s I got either a steady white led light or a blinking one, but no chime. I think I tried both of them in the A slot.

Now: as to delidding, a tip: after removing the lid from 4 x5675s, I think I can say that the process of the lid coming loose is not straightforward. It doesn't just snap off. Instead it seems that the solder has some elasticity and you have to go back and forth for several times with the vice, until the bond is finally loose. At least with the x5675s this seems to be the case, perhaps with others?

Here is a question: with the structure of the 2009 dual tray, with delidded cpus, is it a realistic scenario that one can go so tight with the heatsink, that one can break the tray? My feeling was that when eventually I managed to get both old cpus online (as well as all the ram), I perhaps had to go tighter than I had gone with the x5675s. One of these days, I think I´ll give it another go, but would love to know more.

I assume your Mac has been upgraded to a MP5,1 model , as a 4,1 doesn't recognize six core chips or the quad X5677 .

Why don't you examine the pin array of the sockets for bent pins ? For larger sized sockets than your LGA1366 , I'll use both a DSLR camera's super micro and an USB microscope . With the LGA1366 I can use the naked eye and an AA flashlight's beam pointed from several directions .