AMD 754 and 939: identical pin dimensions?

[VulcanTourist]

I have an Opteron 185 with one broken pin , on the perimeter, that I intend to try to repair. AMD predictably refused to offer any helpful information, citing "liability" crap. What I still don't have is another cheap CPU that I can cannibalize for a good pin to use. Someone locally is offering a Socket 754 CPU really cheap.

Do Socket 754 CPUs have pins of the exact same dimensions (length, diameter) as those on Socket 939? Are there any Intel CPUs of certain generations that might also have identical pins?

(If anyone wants to volunteer other advice or links on how to best attempt this, it's all welcome at this point.)

[iguana]

[kosmokrator]

igor have right.,.....
try to test the cpu may work just fine...

many pins are same...like ground ,nc ,or possitive voltages....

[iguana]

If the task to made this CPU working (not to sale nice looking & working device)... I used the follow method:
I didn't try to solder missed pin.
I just put a piece of elastic wire directly in the socket hole where there are no pin on CPU. If you'll guess wire lenght it will "made a contact" and the CPU will work.
Several PCs working till now after that procedure
PC's user have to know that it is prohibited to extract CPU from MB

[Neon]

I hope you don't plan to buy a working 754, only to rip its pins off. If you just need pins, try to find a dead cpu at a computer shop or recycler.

If that fails, I could send you some pins for cost of mailing.

[VulcanTourist]

Well, honestly I did shudder at having to trash a working CPU to do this, but I advertised locally for a dead one and was coming up empty! I know there's at least one out there (it's a big county), but that person wasn't hearing me.

It's definitely necessary... that pin is an active and distinct one. I'm too much the purist to like MacGyvering like stuffing a loose wire in the socket. Other people have tried that with the addition of a dab of good thermal paste (which is electrically conductive ), but again that offends my senses. I guess I'll do it if it becomes obvious that reattaching a pin is impossible without equipment I can't afford, but I really wanna try this.

I know this will almost certainly require a eutectic solder, but I'm guessing good old 63/37 lead-tin isn't going to cut it because of the high melting point. There are eutectic solders based on indium and other substitutes for lead that have much lower melting points, but do I really want to use a 70C solder on a chip whose thermal output might exceed that? I need something in between that and 63/37, I'm guessing?

Then there's the matter of the iron to use. I have a Princess 10W iron, which is intended for delicate electronics work, BUT it's not temperature controlled. Just how narrow would my thermal margin be here? What if I pre-chilled the CPU by setting it on an ice pack, for instance?

[gshv]

Try to put it into the motherboard. The processors have a lot of duplicate ground and voltage pins, and will work fine if one of those pins is missing. Also, socket 939 has two memory channels - if you break a pin for one of those channels then there is a good chance the CPU will still work in a single channel mode. I currently have two processors missing one pin each - socket AM2 Sempron 3800 and socket 939 FX-60 - they fail to post in dual-channel memory mode, but in single-channel memory mode both work fine and pass all tests.

Gennadiy

[VulcanTourist]

Again, it's an active and necessary pin. Even if the pin was associated with the memory controller and it did still work in single channel mode, I don't think I'd want that IF I had the option of repairing it. I don't have that choice, in any case.

I still have the choice of jamming a loose pin in the socket and praying the big HSF sitting on top won't wiggle it out of contact, but that choice gives me the creeps....

[Neon]

[VulcanTourist]

Would the loose pins include the solder base?

[Neon]

For the most part, no.

These are pins that broke off as I attempted to straighten them.

[VulcanTourist]

Ah, but see I need the whole pin, including the stubby little base that is the attachment point where it is soldered. If my crazy project is to succeed at all it depends upon having the complete pin. That's why I was shopping for dead CPUs, so that I could desolder one as a "spare part" myself.

So how much does it cost these days to ship dead CPUs intact with no further bent pins?

[Neon]

Eh, if you promise to be kind to animals, then I could ship one gratis.

You can PM me your address. Let us know how it goes.

[VulcanTourist]

Thanks for the very generous offer! It may not be going anywhere, though, if I can't get some reasonably expert input from some source other than AMD (which has already thrown in with the Devil). I was hoping maybe someone here had been involved with the production side of fabbing and could answer some of those questions nagging me.

Barring that, I might call Indium Corp. on Monday; they make a bunch of those specialized eutectic solders and might be more helpful than AMD was....

[VulcanTourist]

There's been long lapses of time (this is typical for me), but I'm still working towards getting that Opteron repaired. Neon was kind enough to ship me a dead CPU, so I had something to cannibalize, and last week I found a new tip for my old Ungar Princess 10W iron.

With the Princess iron and new tip, I've managed to remove a single complete pin from the dead CPU. It honestly was not easy, and I'm doubtful that my 10W iron has the muscle to do this properly. I was barely able to get the new tip tinned in the little tip tinning container I have. I only managed to get the one pin; when I tried to repeat the success with two other pins, they failed to release. I was able to melt a dot of Kester Sn63 solder onto the empty pad, but only just barely with effort.

It seems I really do need to find out more about the type of solder used to adhere them in the first place. AMD wasn't forthcoming, so maybe I'll try to pry some details out of a specialty solder supplier like Indium Corp.