bit-tech.net

Has 'Pin Burn-out' Returned with LGA1155?

Has 'Pin Burn-out' Returned with LGA1155?

Pin burn-out evidence, provided by the TechReaction blog.

Does Intel's new LGA1155 socket suffer from the same 'pin burn-out' problems that we first saw on the company's LGA1156 socket? That's what the TechReaction blog claims; its Gigabyte P67A-UD4 and UD7 boards, which both feature Foxconn LGA1155 sockets, have suffered from socket burn in the VCC power delivery area.

However, it's worth noting that the site claims this only occurs after extreme overclocking, much like the original burn-outs on the LGA1156 socket.

It could be an isolated incident, as we've not heard of any other reports yet, and our UD4 is still working perfectly, despite being subjected to strenuous overclocking sessions. TechReaction also hasn't provided any details about how the burn-out occurred, although it does say that it contacted Gigabyte eight days ago to raise the issue, but has yet to hear back.

We've just checked a couple of other boards from our own quarters, and both the MSI P67A-GD65 and Asus P8P67 (Deluxe and WS) use LOTES sockets instead of Foxconn ones. That said, LOTES sockets weren't entirely immune to the original problems with the LGA1156 socket either.

If you own a different LGA1155 board that uses another brand of socket, then please drop us a note with the details, and we may be able to look into this further.

Could this be a serious quality-control problem with Foxconn sockets, or is it a non-issue that's only relevant to a minority of overclockers? Let us know your thoughts in the forums.

Has 'Pin Burn-out' Returned with LGA1155? Has Intel socket 'pin burnout' returned with LGA1155? Has 'Pin Burn-out' Returned with LGA1155? Has Intel socket 'pin burnout' returned with LGA1155?
Click to enlarge

29 Comments

Discuss in the forums Reply
RedFlames 17th January 2011, 11:52 Quote
Quote:
However, it's worth noting that the site claims this only occurs after extreme overclocking, much like the original burn-outs on the LGA1156 socket.

[extreme] overclocking may damage your kit... who'd have thought...
mi1ez 17th January 2011, 11:53 Quote
^+1

I would say this is nothing to worry about for the vast majority of buyers/users/overclockers.
faugusztin 17th January 2011, 11:55 Quote
To those who think this is a serious issue - no, it is not. You can achieve this by overtightening the cooler on board, which makes the board bend a little, which makes some pins not having contact with CPU, which means same current going through fewer pins, which means higher current per pin. Combine that with extreme overclocking where you use a lot higher voltage than the safe limit, and you got a deadly combination. And what a surprise, LN2, dry ice and other cooling methods often use mounting systems which tend to bend the board. Even in P55 time, the issue was the same.

Simply don't overtighten your cooler on your board and you won't have burn pins.
Snips 17th January 2011, 12:13 Quote
I wasn't even aware of the 1156 problem! Not concerned by this at all. If you extreme overclock then you already know this could be a possibility. It's one of these warnings you get like ....Don't run with scissors or Don't use virtual reality goggles while standing at the edge of a cliff.
Spraduke 17th January 2011, 12:26 Quote
I've got the UD4 and whilst my overclock is a measly 4 GHZ (2500k but still much faster than I could get my E8400) and I havent changed the voltages so all seems well so far.

It scared the bejesus out of me when I fitted the CPU because the retention plate seemed very stiff and it made a nice grating noise (assume the pins flexing) but all seems well.
dunx 17th January 2011, 12:38 Quote
Voltage isn't the issue, it's the current ^2 that causes the heating...
feathers 17th January 2011, 12:39 Quote
Some people will always be too scared to overclock. I know people who think overclocking should only be done without voltage raising (duh!). The same people who believe that applying thermal paste to a CPU is a complicated and risky business. The world is full of muppets.
faugusztin 17th January 2011, 12:43 Quote
Quote:
Originally Posted by dunx
Voltage isn't the issue, it's the current ^2 that causes the heating...

Yes, but increasing voltage increases current as well, or in other words, Ohm's law. So does the non-perfect contact, which makes less pins take more of the current, and once you cross the maximum current per pin, you get these nasty burns :).
V3ctor 17th January 2011, 12:53 Quote
I have my 2500K at 4.8ghz with 1.35v, that voltage doesn't seem that big to do these things... Better downclock, just in case...
Landy_Ed 17th January 2011, 13:28 Quote
Not everyone wants to overclock. I know people who think overclocking can only be done by raising voltage (duh!). Some people believe that thermal paste is a distateful and irrelevant task that if not done correctly doesn't actually matter. The world has some muppets in it, and their fingers will at some point get burned and/or covered in excess thermal paste!

And some people remove rev limiters, over-rev their engines to get more power at the lights then wonder why their oil suddenly looks like a latte but smells rather nasty!

:)
faugusztin 17th January 2011, 13:29 Quote
Of course 1.35V is nothing. But extreme overclocking means 1.7V and higher, plus as i said, most times it comes with overtightened coolers, which results in less pins making contact with the CPU itself. And that means that if some pins don't have contact with the CPU, then the rest must deliver more current. And this way it's easy to go over specs of those pins.

And by the way, the burned pins are those which work correctly.
wuyanxu 17th January 2011, 13:42 Quote
it's those Foxconn sockets again.....

nothing to worry about for normal consumers. i have a first generation Foxconn LGA1156 socket, and i inspect the socket everytime i re-mount the cooler, inspected last week while installing the new PSU. zero problem.

just avoid the said boards, UD4 and UD7 it seems. buy Asus? :D
lworbey 17th January 2011, 14:57 Quote
Quote:
Yes, but increasing voltage increases current as well, or in other words, Ohm's law

Err, not quite...

The CPU will require a certain amount of power to run and as P = I x V if you increase the voltage then the current goes down for the same amount of power. That said I don't know enough about the internals of a CPU to know if increasing the voltage makes the CPU use more power overall but in general increasing the voltage decreases current...
faugusztin 17th January 2011, 15:16 Quote
Quote:
Originally Posted by lworbey
Err, not quite...

The CPU will require a certain amount of power to run and as P = I x V if you increase the voltage then the current goes down for the same amount of power. That said I don't know enough about the internals of a CPU to know if increasing the voltage makes the CPU use more power overall but in general increasing the voltage decreases current...

There is one minor flaw in your logic - maybe it will surprise you, but when you increase the CPU voltage, that means you increase it's power consumption too. And because resistance doesn't usually change much, that also means higher current at higher voltage.
lworbey 17th January 2011, 15:24 Quote
Yeah hence why I said
Quote:
I don't know enough about the internals of a CPU to know if increasing the voltage makes the CPU use more power overall
as I wasn't sure how the CPU would deal with the increased voltage.

Sorry the section of my brain where ohms law was burnt to kicked in and told me it didn't sound right hehe! Still good to know, thats my learning done for the day :-)
faugusztin 17th January 2011, 15:31 Quote
Quote:
Originally Posted by lworbey
Yeah hence why I said

as I wasn't sure how the CPU would deal with the increased voltage.

Well, there is something in the new CPUs which limits the turbo modes if you go over the power consumption or temperature limits, but the fact is that this all comes down to overtightening the cooler and bending the board. Even if the bend is small, it's enough to keep few pins from having contact with the CPU itself.
Glix 17th January 2011, 16:55 Quote
So we learn that moving the pins to the motherboard has it's own crop of disadvantages, mainly that it means if your board flexes for whatever reason, you can fry your cpu/mobo.

So does this mean those big heavy blocks (w/c included ie bad piping) are bad for the long term? :>
t5kcannon 17th January 2011, 17:00 Quote
Quote:
Originally Posted by RedFlames
Quote:
However, it's worth noting that the site claims this only occurs after extreme overclocking, much like the original burn-outs on the LGA1156 socket.

[extreme] overclocking may damage your kit... who'd have thought...

haha yeah, breaking news :D
maverik-sg1 17th January 2011, 17:06 Quote
Quote:
Originally Posted by faugusztin
Combine that with extreme overclocking where you use a lot higher voltage than the safe limit, and you got a deadly combination. And what a surprise, LN2, dry ice and other cooling methods often use mounting systems which tend to bend the board.

I cant agree with this - dry ice, LN2 and phase coolers all cool the chip lower than -20degrees Celsius - to the point a motherboard needs to be properly insulated (front and back) around the cpu area - I doubt anything gets hotter than +10degs, even with an exaggerated warped board.

Anyway to get back to the socket issues- I do recall early conroe boards having the cheapest of the nastiest of sockets, whereas even pressure from 'normal' coolers and or mild voltage increases (resulting in more current that created higher temps through the pins) would cause the pins to bend/collapse/burnout.

Some etailors in their infinite wisdom <cough>ARIA.CO.UK<cough> would even say 'you must of put the cpu in wrong' and reject RMA instead of realising the issue at hand.

Once I even had a board delivered, sealed and brand new - it complete with bent socket pins at no extra cost... what a plus >.<

I sincerely hope that the lack of customer satisfaction caused by the nature of burying ones head in the sand does not rear it's ugly head again.
Ross1 17th January 2011, 17:09 Quote
Some etailors in their infinite wisdom <cough>ARIA.CO.UK<cough> would even say 'you must of put the cpu in wrong' and reject RMA instead of realising the issue at hand.

-> I had that exact problem with aria, never bought from them since.
Azayles 17th January 2011, 17:46 Quote
How come it's the pins which burn out, not the finer, more delicate bond wires in the chip itself? or am I missing something here?
Surely pin burn out would be more due to a poorly seated chip, which is having intermittent contact with a voltage and arcing, which is causing the burning, and not due to shear quantity of current flow?
greigaitken 17th January 2011, 18:22 Quote
Azayles - good question. i think it's a mixture of contact with poor integrity mixed with high power useage. the wires inside cpu will have a stable high area of contact.

The amount of current per mm^3 of pin isnt that big compared to say that on the pci strip on a graphics card so the problem cant be too much current travelling through the pin.

An impractical but theoretical solution would be to have some sort of high conductance fluid applied to each pin to ensure maximum contact area.
Of course since there are no problems under moderate overclocking then nobody needs to do anything and be happy your mb works just fine.
centy 17th January 2011, 21:23 Quote
The image shows what pops into my mind when I consider putting VTT voltage higher than VCore
thehippoz 17th January 2011, 23:56 Quote
Quote:
Originally Posted by feathers
Some people will always be too scared to overclock. I know people who think overclocking should only be done without voltage raising (duh!). The same people who believe that applying thermal paste to a CPU is a complicated and risky business. The world is full of muppets.

yeah that's a lot of it right there.. same people are scared of linux

but on the other extreme- there's guys in linux who will never see the great things they can accomplish in windows with super expensive programs xD.. they are just as bad if not worse cause they become neckbeards
Landy_Ed 18th January 2011, 12:36 Quote
^+1
VaLkyR-Assassin 18th January 2011, 21:43 Quote
I wish Intel would go back to the older style socket that AMD still uses with the pins on the chip itself - that design by default doesn't have this issue.
VaLkyR-Assassin 18th January 2011, 21:46 Quote
Quote:
Originally Posted by Ross1
Some etailors in their infinite wisdom <cough>ARIA.CO.UK<cough> would even say 'you must of put the cpu in wrong' and reject RMA instead of realising the issue at hand.

-> I had that exact problem with aria, never bought from them since.

I'm pretty sure the majority of etailers have the same policy tbh. And it's obviously a major cause for concern for us buyers. It might be down to the motherboard manufacturers refusing the reject boards though.
Xtrafresh 18th January 2011, 22:33 Quote
Look up the fine print, bent pins are not covered by warranty, however unfair this may or may not be. BURNT pins due to high overclocking will 100% be denied, as it's double-not covered.

It's a storm in a glass of water. They think they read something similar, and cry out the first relevant name they think could be the issue, and forget to check anything else.

From the blog, it seems only one of his two samples has the damage, so he is claiming to have discovered structural problems on the basis of one defect. That alone is a load of horsedung already, and looking at the pictures, i fail to see anything burnt, it's just damaged. He should blame the reviewers that had the board before him, and possibly the PR rep that failed to test the board before sending it, but crying wolf about the whole socket on a structural basis is frankly just unprofessional. Why re-report this on the front page as if it's a genuine concern? We all KNOW the vast majority of the public are impressionable muppets, and they will be telling stories of the sucky foxconn sockets years from now :p
maverik-sg1 19th January 2011, 09:47 Quote
Speaking from experience, this was a problem before and could potetniallybe a problem again.

I have built well over 30 LGA based systems, add into that cpu upgrades and we're talking 40+ in the last 5 or 6yrs. My RMA's have been from boards that have not generated faults immediately after installation, but rather weeks or months after running faultlessly - so then for some pre-pubescent Asshat to tell me that I put the cpu in wrong and this is my problem....it's insulting.

The worst board for me was the ASUS P5K and in fact like I said previously one of these came with bent pins already......I only knew that because I insepcted it thouroughly prior to use, which then led me to believe that maybe, just maybe, the other faulty boards already had bent pins also? It was not something I thought I really needed to check before.

If you compare sockets of later LGA775 boards to the earlier P5B or P5W you can see an apparent shift in robustness/quality. Sockets of any kind should be robust enough for the job at hand and it can be done.

Most of these board promote overclocking and watercooling and some even extreme cooling - so users of these boards are expecting some support if/when something like this happens - at the very least, if a warranty is too flimsy - there should be a focus on offering socket replacements at a reasonable cost (£20 plus shipping) and a fast turnaround (10days).
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums