P55-UD6 socket burn

[RPGWiZaRD]

More likely Windows 7 mysterious reboot problems..
Could be drivers like chipset, audio, GPU etc..
Powersuply problems like bad contact, rail issues..
Incompability software..

My problem is as follows:

Randomly will restart when I'm about to login to Windows without any warning/BSOD or when playing games (well I have only played UT3 and Borderlands lately basicly and it happens for both of them), sometimes I can play for 1 - 2hrs and sometimes it restarts 3 times in 1hr. No minidumps received but evenlog reports "critical" issues: "Kernel-power" Event ID 41, and basicly means unexpected powerloss or something like that.


Well here's everything I've tried:

- Running Windows XP [no change]
- Reinstall of Windows 7 due to I used a vlited Win 7 copy before [no change]
- Reverted to default BIOS settings + some increased vcore to ~1.265v and 1.2v VTT to make sure it's not a degraded CPU [no change]
- Swapped PSU to my backup Thermaltake ToughPower 750W [no change]
- Stability testing; OCCT PSU tester, LinX, MemTest [OK]
- Temperatures are fine, not using a comp case, CPU @ 1.4v: 55~59C typical for gaming, 68~72C in LinX, GPU @ 60~75C depending on game (fan @ 61%), mobo sth like 32C.
- Tried different drivers for GPU, NIC and audio card [no change]
- Removed CPU to check for burnt pins or bad contact => no burnt pins and all pins seemed to made contact (my eyes after systematicly checking 1156 pins O_o).

However since it happens mostly when playing UE3 based games I think it's most likely a GPU failure, if it was bad contact or mobo voltage regulation issue of some kind it would more than likely cause errors in stability tests as well. The funny or should I say unlucky thing is that when I tried to give my backup card 8800GT a try then ofc the thing freezes in UT3 after a few mins and basicly broke as after the restart I had white dots all over the bootup dos screens and Windows 7 won't install drivers for it cause it says the device is not workign properly and now it doesn't even boot into Windows with it, damnit!

Then the only other gfx card I have here is an old 6800GT and playing UT3 is NOT funny with a 6800GT and a i7-860 with very stuttering gameplay at 640x480 and lowest quality (7 ~ 30 FPS) but it's like a video that swaps between slowmotion and fast-forward mode, so it's really a pain to even whitstand running some test sessions for say 3hrs to see if it's not rebooting. However with this card it doesn't ever seem to fail at login, I've already made like 10 successful logins in a row.

How common would it be to recieve such issue (kernel-power) of a gfx card though? This EVGA GTX 260 SC 55nm is not following standard design AFAIK so it's more likely to happen IMO. In fact I managed to find another user with a EVGA GTX 260 SC that's got this "kernel-power" issue as well but for him it also can happen while browsing web and there were no reports if issue got fixed and how he fixed it in the thread.

Just dunno if I should buy a new GPU right now or wait for 8800GT RMA to be completed which prolly will take more than a week.

[Dahmer]

Doesn't that make sense?

Pins are attached to socket plastic; pins get hot and heat up plastic. Plastic can't hold heat and melts.

makes a lot of sense to me =o

problem solved i guess

[RPGWiZaRD]

In my case it turns out to be GPU clock instability lol. I've run these clocks at over 6 months perfectly stable and then suddenly it starts rebooting at them? Normally I would expect seeing artifacts or bsods if GPU overclock stability degrades, not getting random reboots without any warning and "kernel-power" issues. I ran UT3 for like 4hrs without a prob already at standard clocks which hasn't been possible the last 1½ weeks.

[Marco André]

safaxs, how did you checked the contact between both parts?

[cowpuppy]

Yes there was a problem with the manufacturing of these sockets but I don't believe it's exactly what they are saying. Look at the socket and exactly how the pins make contact. If you notice all the contacts around the outside of the cpu socket point inwards, leaving the only way to make contact for the outside edge to be made by the little straight piece of contact that moves and raises to contact the cpu right next to the angled pins(this is what puts the dents in the bottom of the cpu not the angled fingers). I marked a pic with red on the pins that I am talking about. If it was as they are saying that after the cpu heats up the plastic is warping, making the angled fingers loose contact then there would not be any issues before the combo is even ran pins would all make contact and then start loosing contact as the parts heated and warped. Don't settle for the answers they are giving us at this point and time.

[matose]

The good news is that the new P55A series from Gigabyte has LOTES socket. Let's freeze them !

[cowpuppy]

So after doing a lot of looking at the pics I have a few questions that you guy's will be able to investigate. I attached another pic and marked areas of concern. The area that is marked you can see little impressions where the ejector pins of the tool that produced the socket are sunk into the part. (pins are to long or the part was ran to fast causing the pins to push into the part and this would cause distorted/warped sockets) When this happens ejector pin flash is not far behind. Any flash in this area would cause the cpu not to sit properly flat in the socket making it harder for the pins to make contact. I can see flash on the top left but I'm not sure how close the cpu is to the these pins (top left and right)

[So_BAD]

After big delay, last friday received my Core i5 750.

To be sure that nothing was wrong, before I did a contact test between pins and CPU socket. I fit it with my water Block, for few hours.

I dismantled them and made these pictures HIRES in MACRO.

Notice the circled areas, which are the marks mess electrical contact...







Well, I am now terrified!
.
.
.
.
.
.

[Leeghoofd]

Did you also talke pictures before you had fitted the CPU ? were there no marks on it already ? or doesn't Intel pretest these CPU's beforehand... c'mon lads move on... The issue has been discovered and analysed, with thanks to Massman, SF3D and Raju. It should be corrected in the current range of mobo's...

[So_BAD]

Did you also talke pictures before you had fitted the CPU ? were there no marks on it already ? or doesn't Intel pretest these CPU's beforehand... c'mon lads move on... The issue has been discovered and analysed, with thanks to Massman, SF3D and Raju. It should be corrected in the current range of mobo's...

No, I didn't.

But, I have had dozens of Intel CPU (775, 1366) and don't saw this marks before. If Pretest or not, I don't know. But no marks are visualized before.

By the way, double marks in contacts, is not caused by two pins contacting the same electrical contact into CPU, like a afraid. It makes when the pins is bended over fit pressure (probably).

I "discover" it right now, because a tooke some HIRES MACRO pics of older PentiumD 915 and Pentium4 531, both 775. And in this are the same marks (double contact) and also, NO CONTACT MARKS IN SOME CPU CONTACTS...
(I will upload this pics ans post here latter)

But these CPU is perfect working for many years... any troubel is detected...

Well, I've more confused now....


P.S.: sorry my poor, very poor english...

[massman]

NO CONTACT MARKS IN SOME CPU CONTACS...
(I will upload this pics ans post here latter)

Doesn't really suprise me. I reckon Intel adds more VCC pads than the bare minimum just as a safety precaution.

[So_BAD]

Doesn't really suprise me. I reckon Intel adds more VCC pads than the bare minimum just as a safety precaution.

Ah, ok. Understand.

As said:

Pentium 4 531





Pentium D 915

[Mikael]

By the way, double marks in contacts, is not caused by two pins contacting the same electrical contact into CPU, like a afraid. It makes when the pins is bended over fit pressure (probably).

No, you're looking at the wrong marks. The highly visible marks are not caused by insertion in the socket, those are there from the factory. If you look closer at your own pictures, you can see faint dark spots in the middle of most of the pads on your CPU. Those are the marks that your socket has made.

There is no way for the pins in the socket to cause the sharp marks that some of you seem to be looking for.

[So_BAD]

No, you're looking at the wrong marks. The highly visible marks are not caused by insertion in the socket, those are there from the factory. If you look closer at your own pictures, you can see faint dark spots in the middle of most of the pads on your CPU. Those are the marks that your socket has made.

There is no way for the pins in the socket to cause the sharp marks that some of you seem to be looking for.

You is 100% right!

I "paint" with permanet pen (pen to write in DVD's) some CPU electrical contacts and remount. When dismantled it, there is marks, exacly you told.
Good job, Mikael !

Well, return to zero point of my observations... and more frustade. Tomorrow I will UP the system, will see!

By the way, I make contact with LOTES headquarter, talking about this socket 1156 issue. Me and some friends want to buy some complete LOTES LGA 1156 socket (PN: ACA-ZIF-082-K01), to replace Foxconn. Ms. Hellen Chen, tell me that they haven't a officer in Brazil. So they couldn't sale any parts here. See below.

But they others sales representative officers, in Asia, Europe and USA. If we could buy a hundred of units, its solve the problem of many people, with the same afraid.
Someone would contact they, to talking about this issue?

De: Hellen Chen <sa11@lotes.com.cn>
Para: Marcelo Gomes <sobad_br2@...>
Enviadas: Segunda-feira, 2 de Novembro de 2009 4:59:12
Assunto: RE: Socket 1156 Issue

Hi, Marcelo,

Thanks for your e-mail. We don't have any sales representative in Brazil

For our business model, we will not direct business to end -user or retailers. Please contact the motherboard maker or shop for the replacement.

Should you have any questions, please let us know.

Thanks & Regards,
Hellen Chen
Marketing & Sales Dept.
Lotes Guangzhou Co., Ltd.
DID: 86-20-8468 6190
Fax: 86-20-8468 2478
E-mail: sa11@lotes.com.cn
Website: www.lotes.cc

Maybe togheter we could help us.


Thanks!!

[Mikael]

You is 100% right!

I "paint" with permanet pen (pen to write in DVD's) some CPU electrical contacts and remount. When dismantled it, there is marks, exacly you told.
Good job, Mikael !

No problem.

By the way, I make contact with LOTES headquarter, talking about this socket 1156 issue. Me and some friends want to buy some complete LOTES LGA 1156 socket (PN: ACA-ZIF-082-K01), to replace Foxconn. Ms. Hellen Chen, tell me that they haven't a officer in Brazil. So they couldn't sale any parts here. See below.

But they others sales representative officers, in Asia, Europe and USA. If we could buy a hundred of units, its solve the problem of many people, with the same afraid.
Someone would contact they, to talking about this issue?

The problem is that the socket can't just be replaced. All of those 1156 pins in the socket are connected to the motherboard, probably through a ball grid array or similar. Removing the socket from the board would require heating the socket and board so that the solder points melt. You'd then need to restore the balls before putting on the new socket and once again heating it all up.

This process is very delicate and cannot be done without expensive machinery. Simply put, it needs to be done at a PCB assembly house. The problem is that it would be expensive and I doubt anyone would even take on such a job. Besides, the likelyhood of failure (i.e. destroying the board in the process) is high.

So, unfortunately, the best solution is to simply get a board with a LOTES socket already on there.

[krogen]

2 weeks ago I got a P55-UD3R.

Same exact issue.

I ran it with an i5 750 at 4GHz (20x200), now it is no longer stable at that frequency. Can't run it at anything higher than 160x20... It also won't recognize all ram (sees 2GB out of 4).

At least it still (kind of) works.

[Leeghoofd]

and your issue is directly related to the socket you think ? What did you test ?

[Hyperhorn]

First case I discovered at pcghx.de: http://extreme.pcgameshardware.de/ko...ml#post1258061

Settings used: 4.3 GHz, 1.43 V Vcore, 1.25 V IMC @ Prime (water-cooled)
According to the user he switched to 4.6 GHz and then BIOS/CPU-Z-detection went wrong. Then he looked at the bottom side of the CPU:

Pic 1
Pic 2

Edit: Mainboard = Asus Maximus III Formula

[krogen]

and your issue is directly related to the socket you think ? What did you test ?

I can't say that that this is the issue with complete confidence.

What I can say is that the board worked overclocked for two weeks no problem, Prime95 stable and was able to pass 10 iterations of intel burn test on a high setting. Then one random day I get BSODs out of nowhere, and the motherboard will no longer recognize all RAM. Thinking that my RAM has gone bad, I try a few other sticks, but I have the same issue. Flashing the BIOS to the newest beta one doesn't help.

So I do a google search on why it wouldn't recognize all my memory. I find a few P55 threads saying that the pins on the socket might be bent. So I take off my heatsink and the CPU and what do I find... Warped pins. I can't take any pics since I don't have a camera, but some of the pics in this thread seem awfully close to what I saw.

[chew*]

Lotes.

I think this is far from over yet. Question is what is really causing this now.....1156 power design? Faulty cpus? or intels 1156 socket spec?

http://www.ocforums.com/showthread.p...63#post6295863

[massman]

Note: some companies used the Lotes holddown, but Foxconn socket.

[chew*]

Note: some companies used the Lotes holddown, but Foxconn socket.

Thought EVGA p55 FTW was strictly Lotes?

Here is my guranteed 100% foxcon.



And here's the one on this particular board in above post.




Note: The molds are completely different.

[Mav451]

Crap, I had a sneaking suspicion that this might come up. Now what? Ahhhhh

[chew*]

Crap, I had a sneaking suspicion that this might come up. Now what? Ahhhhh

Ehh, I would not take that screenshot I posted as an end all........

The location of the burn is rather "interesting" to say the least as it differ from all the others.

[Mav451]

Yeah I noticed the molding is different. Even still, people put tons of voltage in C2D (s775) and PhenomII's (AM2+/AM3) - I don't remember hearing a single burn story on the scale we are seeing from this new socket. The different location could be different...vulnerabilities? I dunno. I don't know the native language of that original link.