Physically reattaching pins to CPU?

[KillaKilla]

My brother (on here as Gunslinger) recently snapped 3 pins off his old 2.0a, resulting in his upgrade (to a 2.6c). I offered to buy it ($10!) and he accepted. I was wondering if it's posible to reattach the pins to the CPU... even ghetto style (IE having them sitting in the socket an pressing the cpu onto them for contact.

The 3 broken pins are on the outer rim, by where the 2 missing pins usually are for CPU orientation. They are in line, next to each other. Obviously it can't work w/o these pins attached.

I do have extra threaded copper wire laying around (from a gutted PSU). He doesn't know where the 3 pins went, so of course I don't have them. Also, would it be posible somehow to sodder (solder? SP?) them back on? or would this burn the CPU? This Mobo will never use any other cpu, so having the pins stuck in the socket doesn't matter!

 

[MrSnatcher]

Hi Killa, I have the same problem with a P4 3ghz. I bought some liquid weld at radio shack, cause I cant solder worth a crap, and afraid of frying the chip. I am missing the 2 outer ones. I though of the wire thing to, but it seems to flimzy, or the other wire is to thick. So I also would like some ideas. I have been told its a lost cause, but surely someone out there has successfully did this.

 

[MercenaryForHire]

I've never heard of a successful pin reattachment, but then again, most people don't try. I don't know what to say here.

- M4H

 

[modedepe]

I heard of someone doing the ghetto "put the detached pins in the socket" method and it supposedly worked. Where exactly is it broken? Is any of the pin left? I think soldering would theortically work, but it'd probably be too damn hard to solder it.

 

[dkozloski]

The first thing to do is look at a physical drawing of the chip to see if the pins are really required. Many of the pins are redundant and the chip will work just fine without them. If the pin is required take either the pin remnant or make a replacement from suitable copper wire. Tin the end of the pin and apply a little blob of solder . Put a little liquid rosin flux on the broken end of the pin on the chip. Place the pin to the appropriate place on the chip and apply heat to the shank of the pin with a very hot iron. The heat will travel down the pin and fuse the solder and pin in place. It may take several tries to get it exactly in the right place and straight. It will be very tender and you may break it off while handling but just do it again. This type of work is duck soup for a good jeweler because he will have the tweezers and such. Maybe you can arrange a tradeout.

 

[mrweirdo]

Imo the thought of using a cpu with ghettoized reatched pins in an uper end system just scares me. It can work for awhile then you bump your case and poof you fry something like mobo, gfx, ect :/

 

[compudog]

Years ago I reattached two broken pins on a 486-DX2/66 with jeweler's tools and silver solder. Took a few tries, but it worked and ran for a long time. I don't know about the new CPUs with the smaller pins and closer spacing. Good luck!

 

[MichaelZ]

i wish u luck as well. let us know how that goes. if it works i'll try the same thing if i ever snap pins off a CPU. although that's not likely

 

[Maximilian]

I think you should just stick the copper wire into the mobo as you say it wont use another processor anyways and itll prolly work albeit maybe not forever but itll still prolly work. <--easy but might not last as long as actual soldering
Unless you can get access to accurate soldering equiptment and somone to work it unless you can solder yourself. <--long lasting but more difficult
Or hell you could just do a DIY job and melt it together yourself and if it frys the processor you can melt the whole thing into goo and pass it off as some form of new pentium

 

[KristopherKubicki]

This is the reason why Intel moved to the new LGA package no pins!

I tried soldering a pin back on once, needless to say it worked, but was not stable. I am sure by affecting the purity of just that one pin, i managed to muck up the latencies enough to totally throw the voltages off.

Kristopher