Missing Pentium 4 Pins

January 28, 2006 4:25 PM

I am missing two pins from my Pentium 4 3.06 (478)CPU... Can it still work?

I am missing Two pins called W1 in cpu pin assignment (which is also called A#[29] Source Synch(Signal Buffer Type) - Direction Input/Output

Also B26 in cpu pin assignment (which is also called VSS) Power/Other.

That is all what Intel pdf guide says they are...

Would those pins essential for smooth operation?

Following link shows intel pdf and it starts from page 53

http://www.intel.com/design/pentium...ts/24988703.pdf

posted by curiousleo to computers & internet (20 comments total)

prolly not. i had an AMD k5 with one missing pin. didn't work at all. didn't fry my mobo either, but i can't guarantee you won't.

at any rate, i would plug it in and give it a shot. worst-case scenario: you fry yr motherboard and everything else, requiring you to upgrade yr WHOLE SYSTEM! YESSSS!
posted by herrdoktor at 4:31 PM on January 28, 2006

Hmm..it's possible. Did you clip the pins? I know there are voltage/overclocking tricks you can do my clipping specific pins, but I'm blanking on which ones.
posted by Bucket o' Heads at 4:36 PM on January 28, 2006

I'd be really surprised if it worked at all. If it does work, I'm sure it will crash your OS from time to time. Those pins were there for a reason -- some data flows in or out on them. Chip engineers don't put pins on for fun.

Just pop it on and find out -- I'm 90% certain it will just fail to function. But only trying will tell you for sure.
posted by teece at 4:38 PM on January 28, 2006

i am still trying to unbend rest of the pins now...
but it is very hard to get the cpu to fin into the socket yet...
posted by curiousleo at 4:38 PM on January 28, 2006

Find some local techie geek who loves to play with miniature soldering irons. I had a friend in Romania - a land where you simply do not throw anything out - who did amazing things with totally trashed computer hardware like this.
posted by zaelic at 4:44 PM on January 28, 2006

If you read the signal description, pins in the A# range are defined as the memory address bus. This is used to transmit locations in memory to read and write from. It is absolutely essential. Since it's A[29], there's a possibility your system will work if you have less than 256 MB (or 512 MB?) of memory, but I doubt it.

B26 is one of the chips many, many Vss signals. The worst case here is that it will mean part of the chip isn't powered up, though I'm not sure if they're wired like that. It might not be a problem at all.
posted by cillit bang at 5:01 PM on January 28, 2006

I second the idea that every single pin is essential for smooth operation, unless there are some meant just to act as stabilizers. And even then...

Regardless, you likely will have crashes and problems. What did you do to the chip that involves unbending all the pins and missing two? Chances are, whatever it was cracked the die, in which case, you're out of luck anyway.
posted by disillusioned at 5:01 PM on January 28, 2006

After purchase, it came loose from a box and rattled around with the heatsink.... that is why it was bent...
two missing pins were result of me trying to unbend it..
posted by curiousleo at 5:03 PM on January 28, 2006

o.k. i have successfully inserted the cpu in the socket...
i will put the computer together again and see if it works...
but i stil like to know if those two pins are really that important...
posted by curiousleo at 5:09 PM on January 28, 2006

leo, that's almost exactly like installing an engine that's missing a cylinder. Just because you don't understand how something works doesn't mean you can take random pieces off and expect it to continue working.

You are exhibiting wishful thinking of the highest order. The cost for that thinking may be a dead motherboard as well as a dead CPU. The chance of blowing up the motherboard is much higher than the chance of the chip working.... which makes plugging it in a really stupid idea.

If the pins are broken off or not salvageable, throw that CPU away and buy a new one.
posted by Malor at 6:02 PM on January 28, 2006

http://www.motherboardrepair.com/index.php?sec=procrep

i found this guy...

what do you think?
posted by curiousleo at 6:08 PM on January 28, 2006

Leo, I think you're going to have to suck it up and buy a new CPU...
posted by anthill at 6:47 PM on January 28, 2006

ahhhhhh.....
posted by curiousleo at 6:53 PM on January 28, 2006

You can definitely lose some pins and still function! They are not all essential, despite what some here are saying. Take a look at the tualatin in BX board mods people used to do.

The VSS pin is probably not important, I doubt it would power down a hole section of the chip, it might though. It can't hurt to try... The missing memory address line is critical though.

I have two taulatin's, both missing a pin or two, both working fine. I had to 'replace' a couple of pins to make the chips work...

What you have to do is replace the pin. You aren't ever going to get the pin to reattach to the chip, so forget about that (I think there are services out there that will try to do this for you). Instead, put the replacement pin in the proper hole in the CPU socket (steps for this below). Consider the new socket 775, it has springloaded pins in the socket, which is basically what you are trying to do. You can't replicate the spring, but as long as it is only a couple of pins the spring in the plastic socket will be enough.

Steps:

1. Replacement pin: If you still have the pins that broke off you are in business. Otherwise you will have to find replacement material, I believe I used a tiny piece of lead from a resistor...
2. Prep pin: Add a little ball of solder to the top of the replacement pin, about 0.01" in diameter if you can manage it, maybe a little bigger (I had it easier than you will because the pin spacing on socket 370 is much larger...). This will make the pin a little longer, which will help to contact the chip. Also, solder is much softer than the lead or pin, so it will deform a little and it is less likely to damage the chip. Finally, you need a little flare to keep the pin from falling down the hole.
3. Place pin: put the replacement pin in the correct hole on the CPU socket, do a close visual inspection, make sure it won't short on any other pin when you place the CPU, that would be disaster!
4. Place chip: Put the chip in the socket, hold it down tight while you use the lever to lock the chip in place. I have sometimes left the lever up until the heatsink is attached, allowing the heatsink pressure to help contact with the replacement pin. I don't think that is necessary though, and for a P4 it would be pretty hard to do. Attach heatsink and you are pretty much done.
5. Testing: Make sure you do extensive testing. You are the only person in control of reliability now, don't make a fool of yourself :P
6. Don't give up: Just repeat the procedure if it doesn't work the first time, sometimes you will be unlucky.
7. Caveat: It might fail at any time due to little bumps, corrosion, who knows what. This isn't going to be a mission critical PC ever again

posted by Chuckles at 8:04 PM on January 28, 2006

Err.. I don't exactly mean "the spring in the plastic socket"...

The plastic socket will have an inherent springiness which you will take advantage of. By pushing down on the chip hard while closing the lever you should be loading the socket, when the lever comes down the chip is held strongly in place, which allows the socket to push the replacement pin up a little. I don't really know if that is the critical effect that makes my trick work... It is an added force in the right direction, so it must help a little...
posted by Chuckles at 8:08 PM on January 28, 2006

I always waste time rereading answers after they are posted...

Anyway, I shouldn't be saying "the VSS pin is probably not important"... That is far too strong, but there is a good chance that you can get away without it.

VSS pins are common because all parts of the chip need to have a nice clean VSS signal. That allows maximum clocking of the chip, and reduced power consumption (and who knows what else, I'm an analogue guy, I don't do chip design). So, maybe you have to under clock a little or something, because missing a VSS pin makes it a tiny little bit less stable, it still isn't garbage... Of course cillit bang could be right, it might power down a hole section of chip...


Oh, and disillusioned, man you have to be more optimistic! Maybe we should trade user names for a while or something... I mean "Chances are, whatever it was cracked the die"... It is possible I guess, but it really doesn't seem likely, especially on a CPU with bonded heatspreader...
posted by Chuckles at 8:24 PM on January 28, 2006

VSS is ground, and yes you need lots of ground pins to avoid the phenomenon called "ground bounce", but missing just one ground pin should be okay. its not going to power down any section of the chip.

intel processors don't have integrated memory controllers, so this address pin is not for the main memory. DRAMs are addressed in a multiplex fashion; you have to send 2 addresses, sort of like an X and Y location on a map to pinpoint a particular address. so even if you have 2^n bytes of memory, you don't need n address bits. you need m bits, where m*2 is greater than n.

the address pin in question is part of the so-called FSB (frontside bus) and its very likely not going to work at all without that pin. the physical memory map is not contiguous and just because you have less than a certain amount of ram is no guarantee that there isnt something out there that requires a particular address pin to be asserted.

it looks like the address bus is bidirectional because the motherboard manufacturer can strap pins high/low at power on to program certain processor features, but A29 is not one of these pins. still, without tying that pin to VSS it is floating and even if there is nothing in the address map requiring A29=1, no one is tying it to 0 either, unless there is a resistor pack on the motherboard to provide pulldowns/pullups.

so short answer: repair the A29 pin, or get a new processor...
posted by joeblough at 9:39 PM on January 28, 2006

A29 is absolutely necessary - repair that and it may work. But given that you're bending pins I'm gonna take a wild stab and say that it's sustained some ESD (static) damage from the handling.

Vss not so much if they're all wired together internally BUT be careful. Missing a Vss pin will mean higher current densities in parts of the chip perhaps not designed for it, which causes metal migration and earlier failure. There'll also be poorer noise immunity and increased ground bounce in some part of the chip, probably leading to really mysterious reliability issues.
posted by polyglot at 1:02 AM on January 29, 2006

ometimes there are cuts in the power rings for I/O power vs. core power, so there might be 2 flavors of Vss. but i'm sure there is plenty of design margin such that missing just one Vss pin would not be a problem.
posted by joeblough at 4:45 PM on January 29, 2006

Chuckles—You're right about the heatspreader thing... I was thinking more of the slightly older AMD chips that have an exposed die which would crack by looking at it.

Meanwhile, he was mentioning unbending other pins, which made me think perhaps it had some pressure exerted on the chip against something hard.

I'd be yelling at whomever shipped it to me, or looking to get my insurance money, personally, but I also spent $700 on my chip and would have been spitting blood if it arrived broken or bent in any way.

We could play a fun game and see with how few pins said chip could operate with. That might be fun.
posted by disillusioned at 5:22 PM on January 29, 2006