P55-UD6 socket burn

[STEvil]

touched what table? LN2 to LHE isnt much of a difference and LN2 is fully capable of causing something to become brittle.

[Nelly]

Just to add I'm in a similar situation, I'll explain.

I have installed 4GB of RAM (2x2GB) in slots A1 & B1 as per the manual, but the Bios & OS using Windows 7 64bit only sees "1.99 GB Usable."

The manual says I can only use either A1 OR B1 to boot, so I tried slot A1 on it’s own and the system boots showing 2GB of RAM in the Bios & Windows - Correct, so now if I use B1 on it’s own the system will not boot up with any 4 sticks of RAM I try using - I get AMI BIOS POST CODE: EA which isn't in the Manual to explain what’s wrong – the RAM works perfectly in another system by the way.

Now if I populate all 4 RAM DIMM slots 8GB of RAM (4 x2GB) I have 4GB out of 8GB show in Bios & Windows.

I then began to switch out RAM & found that if I put RAM in slots A1 & A2 - 4GB of RAM (2x2GB) I have 4GB of RAM show correctly, but using slots B1 & B2 fail to show any RAM, so it must be the two faulty DIMM slots out of a process of elimination.



I did some more investigating of this problem & it seems that the issue can arise to misaligned pins, bent pins, missing pins on the socket.

Upon close inspection of the socket this is what I found, all the other pins are totally fine nothing wrong.



I've stated this in my RMA which has been recieved back today to where I purchased, I will let you know the outcome but I'm not very confident.

I'm really really worried that a Novatech Tech/Engineer will just say "ahhh end-user damage, warranty not covered" because Scan & Ebuyer have done this with customers I have looked it up! OcUK have replaced. (omg amazed!!)

With the amount of exact same problems arising I think theirs more to it than it seems, I'll probally take further action if that is the outcome as I simply cannot afford to replace the motherboard.

Well around 3 months later I finally got my RMA sorted out, it was a total nightmare because the retailer as well as the supplier who sort out RMA for Biostar in the UK refused warranty, so I took legal advice from Trading Standards which is part of the UK goverment.

I spoke to Trading Standards/Consumer Direct as I read in the Sales Act that in the first 6 months of a purchase if the damage cannot be proven to be done by the end user then a warranty claim still exists.

Finally after stating this & quoting the Trading Standards Reference number to Novatech I got a call from them saying they would offer a refurbished Asus P7P55D worth £113, my original motherboard cost £138 so I refused this & asked if they would agree if I could purchase a brand new Gigabyte GA-P55A-UD4 for (£75ish) around 50% of its purchase value to which after another day they rang back & agreed.

I think part of the problem is the Foxconn retention bracket, the new P55A Gigabyte models all use the Lotus retention bracket - all in all this took over 3 months to sort out but now I'm happy.

All goes well I'll be installing my new motherboard at the weekend, definitly sticking with Gigabyte all the time now.

Also to note, I also spoke to Gigabyte & their Tech said if it was as issue with their motherboards they would issue an RMA anyway. - http://forum.giga-byte.co.uk/index.php/topic,917.0.html

The future is bright, its blue & white, its Gigabyte!

[saaya]

i have 2 X P55 death,,,,
1 x P55 UD6
1 X P55A UD6
the worst...it killed 3 CPUs i7 670
I hate 1156...i have no lucky with this socket

wow, wth?
so this problem still exists and intel and the mainboard makers arent really doing anything about it? guess the people who push the hw on ln2 and kill the boards and cpus are too few to make them worry about it? :/

[overclocking101]

mine has foxconn socket EVGA P55 FTW. no problems here when it comes to burning. it does bork up the mem controller every now and again if I tighten my block down to hard. I cant seem to find anywhere how much pressure should be used on 1156 setups anywhere. i dont know if its just my board or what but I have to run my cpu block super loose or it only shows 2GB of ram.

[Stefan Payne]

mine has foxconn socket EVGA P55 FTW. no problems here when it comes to burning.

Yet...

[massman]

Killing boards != socket burn. As far as I can see, many of the socket burned mainboards are still working.

[Eeky NoX]

Even with an extra -20 pins rofl! ...gat the Lotes one on FTW, well I'll see when Wprimin under SS though :/

[overclocking101]

does anybody know if there is any documentation on how much pressure the cpu's should have?? i know the venomous-x does up to 70lbs but if I tighten my cpu block tighter than about 20lbs it screws the memory controller and only shows 2gb of ram. im not exactly sure why but after I have to either re seat the cpu or loosen the block till its not even touching for it to show the whole 4gb of ram.

[STEvil]

Its not the memory controller, its probably the motherboard warping.

[STEvil]

POST, you mean. Letting something get beyond that with no heatsink is just asking for things to burn up.

[Leeghoofd]

Is the mobo out of the case ? check for warping, bend pins... you could have got unlucky with eVGA too... nothing to do with a particular brand...

[Eldonko]

Anyone get socket burn while using dielectric grease? Im thinking it could be condensation shorting the cpu and grease could stop it.

[Stefan Payne]

I prefer silicon paste

[Jor3lBR]

Is the mobo out of the case ? check for warping, bend pins... you could have got unlucky with eVGA too... nothing to do with a particular brand...

I've tested two evga's p55 and they are by FAR the most sensible board with this issue.

I guess manufactures take 'warping' event as a real world factor and evga just didn't do the research as much as the others.

If you warp or tighten a p55 evga too much or too little be ready for hell (ram problems and errors, ff-d4 reboot loop, bsod's etc).

The problem there is when enough is not enough?!

[jetroho]

This is from Intel tech doc 322167 . Intel I7 800 and I5 700 Thermal/Mechanical Spec's .
As per the spec's the heatsink min presure is 0lbf min to 50lbf max

It appears that the ILM does most of the socket loading . If the the heatsink manufacturers are making the heatsink spring load from 70lbf to 125lbf . Then in my opinion you would need to either mod the stock ILM plate to provide from 10lbf to 60lbf max . Or remove the load plate completely and mount the heatsink with a spring force of 80lbf to 185lbf . This would then make the static load of the ILM and heatsink match the total static load spec for the socket .

1156 socket load spec.jpg

j

[STEvil]

Someone should get a micrometer and a dead 1156 board and measure the deflection amount of the board with a few different sinks mounted... would be nice to have pressure statistics too but not sure how we could measure that..

[Kougar]

Anyone noticed similar discoloration of LGA1366 CPU contact pads... just without any apparent socket damage? For any reason whatsoever? Also, any idea what sort of voltage/current/heat would be required to cause this discoloration of the contact pads?

I ask because I appear to have a board that turned nasty after one year of good hard use. At first I thought I cooked my 920 with my overclock and over a year of 24/7 Folding@home load... but then I saw what it did to my three week old Core i7 930 I replaced it with. With the 930 I was using 1.36vcore (by CPUZ), 1.60v RAM, and 1.225QPI/VTT @ 4Ghz under an Apogee GTZ.

The photos don't show it well, but the discoloration is more clearly apparent in natural light and can even be visibly seen in a few of the same pads in the 920. The 930 is on the right, which is astounding given I've used the 930 for less than 3 weeks... much of that time not even overclocked.



Suffice to say the 920 is no longer stable at stock frequencies regardless of voltages, and the 930 is no longer stable at 3.5GHz+. The 920 seems to only respond to 1.42+ VTT, and worryingly the 930 seems to require unusually high VTT now as well just to run stable. The ASUS Rampage Extreme II is defaulting to 1.36v VTT just for stock operation.... when I first tested the 930 before installing it into the Gigabyte UD5, the Rampage defaulted to <1.20v VTT. Not happy at all just now...

[STEvil]

What board was that on that the damage was caused to the 920 and 930 originally? Maybe its a board issue as the board should use the default VTT of the CPU and not pick something it wants to use itself..

[overclocking101]

well couple months going on strong. no burns nothing. but with the issues of this thread it still seems to me intel rushed this whole socket replacement thing. i fixed the issue of the memory probs and no posts. simply tightened the crap out of my HK 3.0 backplate that sraightened the board right up. no probs now. using foxconn socket. maybe its a luck thing and foxconn QC isnt that great. I mean its well known the job cuts and layoffs at foxconn, maybe QC got the worst i guess.

[Kougar]

What board was that on that the damage was caused to the 920 and 930 originally? Maybe its a board issue as the board should use the default VTT of the CPU and not pick something it wants to use itself..

The Gigabyte EX58-UD5 in my sig caused the damage to both CPUs. It's the one at fault here... I am trying to determine how to proceed. I am certain it was the motherboard, I never once used voltages or temperatures that could cause such vivid discoloration. The Core i7 930 spent half of those 3 weeks at stock settings! CPU socket pins appear pristine, you would never know it kills CPUs.

The ASUS board does not have "NORMAL" or "STOCK" settings, if I select failsafe defaults it will feed 1.85V to my 1600MHz RAM, and 1.36v VTT to the damaged 930 with the CPU at stock settings. I have to manually set each voltage setting to a number... score one for the Gigabyte BIOS. Before the 930 was damaged, the ASUS board would auto to ~1.18v VTT as reported by its BIOS.

[STEvil]

Honestly the discoloration worries me.. really the only thing that could cause it would be the traces in the cpu substrate packaging being too small causing excessive resistance, or the pins which touch the cpu pads having poor contact also causing resistance and either one of those could cause fires.

[bartx]

I see that mainly it's a problem of Asus and Gigabyte boards. I have MSI P55 GD65, I was using it on ln2 with i3 530, it's also my daily board. No issues at all. Even high voltages about 1,4v on air cooling doesn't make any problems. MB has Foxconn socket. I haven't seen in this thread any MSi board, which had burned socket. Maybe only few people use it, due to problems with voltage (above 1.7v vmod is necessary), so it can be a reason.

Anyway, my friend had burned socket in his Asus board, cpu was fried a bit too (i5 750). And he received new parts from RMA, although this combo has been still working.

[Kougar]

Well, it isn't just cosmetic discoloration. The 920 is 100% unstable at stock settings, the 930 can't even overclock anymore. Before I put it in the Gigabyte UD5 I tested it on the ASUS Rampage.... it did 4Ghz easy using only a minor hike in CPU voltage. Ever since I put it in the UD5 and the discoloration appeared it no longer overclocks at all... vcore doesn't change a thing and only absurd levels of VTT stabilize it. The Ud5 has "killed" the 920 and crippled the 930...

Any thoughts on how to approach Gigabyte about this? The board is easy enough to RMA, but I've just lost $600 in CPUs and they should replace one of them. I highly doubt Intel would warranty it given the discoloration (and my heavy overclocking on the 920 for over a year anyway negates any warranty).

[overclocking101]

gigabyte will do the board but no way will they replace a cpu, they just wont.

[devil_mcry]

there are a new foxconn clip?

i read this in a review of p55a ud7, new clip fix the old problem... is it real?