Black-Green Screen / Grapich Card Dead

[yacard]

Please allow me start with a big apology to the readers and moderators for creating another thread about Green Screens. Yes, there are many out there, but I've not been able to find any similar to my case, and none of those solutions apply to my case. If you know (after reading my post) a similar thread like this one, please do not hesitate in redirect me there, I need help, really doesn't matter how or from where.
With that being said, let's get into business.
3 weeks ago I started to see like a blink on my screen, like a very thins green lines horizontally across the screen. In all the cases, this happened while I was playing World of Warcraft, but all happened real quick, not a big deal, I checked all my connectors, being everything perfectly ok. That was in a weekday, Same week, in the night from Friday to Saturday, again, playing World of Warcraft, the system just crashed, dead, no signal, no sound, just a green/black screen.
See pictures how the screen looks like after the failure, same looking screen with moth cards.
Screen after failure - album with all pictures in this thread

After going through the RMA process with EVGA, they sent me another card, guess what, I installed it last Wednesday. Right after installing it I saw the blinks again and by Friday night, the card crashed again.
I've been talking with EVGA technical support, the representative said that the card I sent did not pass their tests, was completely dead. I'm in the process of receiving another RMA from them, but I had the same problem with 2 different cards, so I'm starting to assume that is not problem of the GPU, something else is causing the damage on the graphic card. EVGA has replaced 2 cards, I dont want to take the risk of a 3rd one.
Going into details.....
RIG:
Motherboard: MSI Z68A-GD55(B3), socket 1156
CPU: Intel I5-2500k
RAM:
* 2 x 4Gb DDR3-1333 Corsair Vengeance , Black Edition working in Dual channel
* 1 x 4Gb DDR3-1333 Kingston
Storage: OCZ Agility 4 SSD 256Gb (SO) + Seagate HDD 500Gb (Massive storage)
PSU: OCZ 700W MXSP 80+
GPU: Nvidea GTX 660 from EVGA, reference design.
2 Monitors 22" in extended desktop mode, using HDMI interface for the main monitor (where I play all my games) and DVI-D for the 2nd monitor.
Windows 8.1, EVGA Experience, and Precision X (software used to monitors GPU temperature)

For your information, since the first card failed, I improved the cooling inside the case, just in case the problem came from overheating and after installing the RMA card I started to monitor the GPU card at every moment. FOr that I used Precision X, software provided by Nvidea to tweak their GPUs. The highest temp recorded was 66 Celsius degrees whiles I was playing Company of Heroes 2. In my conversation with EVGA tech support, the representative said that 66 degree is not a critical temp for that card and should not be that the reason for the problem. He suggested checking the 12V rail of the PSU.
To monitors the voltages I use MSI Control Panel, provided by MSI with its motherboard.
Voltages recorded by the motherboard
For the 12V channel, the shown voltage is only 2, 12.313V and 12.408V, changing from one value to the other in NO specific interval of time, there is NO pattern in the variations.
I also contacted OCZ tech support requesting help to diagnose the health of my PSU. Waiting their response.
Folks, sorry for the length of this post, I'm trying to present all the details for you so you can get a better idea and hopefully help me finding the possible cause of the problem.
Last, but not least, I want to talk about some else weird thing that I noticed while I was dealing with the problem. See the next pictures. All these pictures is to show you the wobble I noticed in the PCI-e slot and the locking system in the chasis does not help either

slot wobble - card lifted
slot wobble - card by its own
same problem, but seeing from the back of the case
rear view upper position
rear view by its onw position
FYI, if you let the card by its own, its stays in the lower position (gravity, lol), the higher position happens when I use my finger to evaluate the magnitude of the wobble.

After all these, my 2 guesses/question about:
Is the PSU causing the damage?
Is this wobble in the slot which the case locking system does not lock the cause of the problem?
What could cause a damage like that in the graphic card?

Ok guys, really wish you can help me with this problem. Please feel free ask for more details on any end/item that you may need for your evaluation, gladly I will provide as many/much possible.

 

[Tom Tancredi]

....

After all these, my 2 guesses/question about:
Is the PSU causing the damage?
Is this wobble in the slot which the case locking system does not lock the cause of the problem?
What could cause a damage like that in the graphic card?

Ok guys, really wish you can help me with this problem. Please feel free ask for more details on any end/item that you may need for your evaluation, gladly I will provide as many/much possible.

Long story short: Great deduction and yes and YES! I would presume to guess a combination of both factors. If the card slot mechinism to secure it lets it move that much you DID NOT SEAT IT CORRECTLY, but if you are 'sure you are' then the mechinism is broken and that would mean a NEW MOBO (built in can't be replaced normally). I would secondly suspect the PSU as well, but this mechanical issue would still be the main culprit and without it resolved, worrying about the PSU doesn't help (your still causing a short by having a 'wobbling' GPU).

 

[yacard]

Long story short: Great deduction and yes and YES! I would presume to guess a combination of both factors. If the card slot mechinism to secure it lets it move that much you DID NOT SEAT IT CORRECTLY, but if you are 'sure you are' then the mechinism is broken and that would mean a NEW MOBO (built in can't be replaced normally). I would secondly suspect the PSU as well, but this mechanical issue would still be the main culprit and without it resolved, worrying about the PSU doesn't help (your still causing a short by having a 'wobbling' GPU).

Yes, I'm 200% sure that I installed the card correctly in the slot. That case is a Cooler Master HAF 932 Advance, and the way that case locks (secures) the expansion card in place is by inserting a plastic tip through the hole and ideally putting enough pressure to avoid any kind of movement
Notice the plastic tip to the left of the card and the hole in the backplate
Open clamp, before locking the expansion card
Closed mechanism.
After closing the clamp
Later today I will update this comment to post the result of moving the card to another PCI-e slot and evaluating the wobble in that other position.
Ok, understand your point...will try to solve it, or at least find the way to secure it in place. I may be able to put threat in that hole and use a screw to lock the card. do you think the PCI-e is broken/damaged?

 

[Tom Tancredi]

Long story short: Great deduction and yes and YES! I would presume to guess a combination of both factors. If the card slot mechinism to secure it lets it move that much you DID NOT SEAT IT CORRECTLY, but if you are 'sure you are' then the mechinism is broken and that would mean a NEW MOBO (built in can't be replaced normally). I would secondly suspect the PSU as well, but this mechanical issue would still be the main culprit and without it resolved, worrying about the PSU doesn't help (your still causing a short by having a 'wobbling' GPU).

Yes, I'm 200% sure that I installed the card correctly in the slot. That case is a Cooler Master HAF 932 Advance, and the way that case locks (secures) the expansion card in place is by inserting a plastic tip through the hole and ideally putting enough pressure to avoid any kind of movement
Notice the plastic tip to the left of the card and the hole in the backplate
Open clamp, before locking the expansion card
Closed mechanism.
After closing the clamp
Later today I will update this comment to post the result of moving the card to another PCI-e slot and evaluating the wobble in that other position.
Ok, understand your point...will try to solve it, or at least find the way to secure it in place. I may be able to put threat in that hole and use a screw to lock the card. do you think the PCI-e is broken/damaged?

Please note the bold part of what I am saying, your looking at the case as the part that 'secures' the card, the problem I am pointing to is the "other end of the Mule" where it secures into the PCI-E slot. That Video Card slot has a CLIP that holds the card in place, if it is not 'seated' correctly it will wobble too (not just because of the case clips). The case clips are supposed to secure the card into place that when the fans get really moving the cimtripical force doesn't move the card out of the slot.

Ideally (as they do in the labs) you would lay your motherboard 'flat' to the floor, then insert the cards 'from above' the MOBO, to secure them into the spot. In some cases this is the ONLY way to optimumly 'secure' the card in the case and not do it while it is 'sitting on the floor' as you have it in those pictures. Just a lessoned I learned tip for you.

 

[yacard]

Please note the bold part of what I am saying, your looking at the case as the part that 'secures' the card, the problem I am pointing to is the "other end of the Mule" where it secures into the PCI-E slot. That Video Card slot has a CLIP that holds the card in place, if it is not 'seated' correctly it will wobble too (not just because of the case clips). The case clips are supposed to secure the card into place that when the fans get really moving the cimtripical force doesn't move the card out of the slot.

Ideally (as they do in the labs) you would lay your motherboard 'flat' to the floor, then insert the cards 'from above' the MOBO, to secure them into the spot. In some cases this is the ONLY way to optimumly 'secure' the card in the case and not do it while it is 'sitting on the floor' as you have it in those pictures. Just a lessoned I learned tip for you.

Ok, double check won't hurt, I will do it the way you explained it in both slots and will see if there is any different. Thank by the way for following up this thread.

 

[yacard]

As mentioned, 2 attempt dowith the installation.
These 2 videos are for the same reason, to show you the wobble. My apologies for the quality, I don't have an studio, neither the proper light, nor a good camera. This shooting was done using my phone with the case lying over the desk and with the light of the room; I even used a flashlight to improve visibility, but I think it's good enough for the purpose, show you the problem.

Video - Problem with the locking system in the case
Video - wobbling card even being well installed.

While installing it I heard the click of the lock in the PCI-e slot. Card is well sit inside the slot. Tried using both slots, still the same wobbling.

 

[Tom Tancredi]

Okay watched the video very closely (WELL DONE! BRILLIANT BTW!). From what I can see (if you look closely) the blue clip at the bottom (the part I was pointing to) should be visibly 'locked' into the card notch that is located at the PCI-E part (you said you hard a click, just visually check it with your nice LED light to make sure). That is good, now about the wobble. Yes the card shouldn't move THAT much, especially as you push the frames the FAN increases and thus causes force to move making the pins at the bottom connect / disconnect (which is also a semi power source for the card as well). That said, I se you can keep the 'snaps' in the up position, I would recommend keeping them in the up position and instead use the proper size LARGE HEAD screws for those spots. To my visual interpetation, the clips 'peg' that fits into the 'screw hole' seems small, and as both points on the hard are not simple drill holes (you see the wide gaps to allow easier compatibility with the myriad of cases that exist) but gaps to allow flexibility when inserting the double wide card. The pegs and the pressure on the clamps do not make a 'tight' seal enough to prevent 'wobble'.

I would try this FIRST, then if your screws still don't keep it tightly inplace, then either the design of the case just won't line up for this type of card properly or the Mobo design is a couple mm off for the PCI-E card slots so you can't get a firm connection. Next test to 'check' would be try a different case with either the card itself (does it fit securely in your friend's PC but not yours?) then try with the Mobo in the second case (just screw it into place no connection made then install card and see if again does it fit the 2nd case AND your mobo fine).

If the screws work fine then lets test video out with a different (again friend?) monitor and cable. Does it work now? Then test your cable on friend's monitor? Does it work now? Now test with your monitor and cable? Does it work now?

If the card is securely screwed in, the cable and monitor all prove same performance no matter who's cable and monitor is used, and we STILL have issue, then proceed to the assumption of the PSU frying the hardware / your power source (house outlet / incoming power to the house). At this point 1) make sure your using some power strip or BEST use a UPS (Uninterrupted Power Supply) as that also protects not only against SPIKE (overload of power) but DROPS (underload of power) that maybe occuring, and usually lets you know when it happens so you can deduce the issue better. 2) replace the PSU, maybe a 750-800?? (look up calculate computer watt usage to determine best solution for your specific system) does it still have a issue?

Hope this helps

 

[yacard]

Sorry for the delay, the last days of the week were very hectic at work...

[quoteFrom what I can see (if you look closely) the blue clip at the bottom (the part I was pointing to) should be visibly 'locked' into the card notch that is located at the PCI-E part (you said you hard a click, just visually check it with your nice LED light to make sure)][/quote].
It's confirmed, the card was/is properly installed.

That is good, now about the wobble. Yes the card shouldn't move THAT much, especially as you push the frames the FAN increases and thus causes force to move making the pins at the bottom connect / disconnect (which is also a semi power source for the card as well).

Holy Cow!! now you mentioned Fans and made recall something important,...After both failure (both graphics cards) all the fans connected to the motherboard started spinning way faster, WAY WAY faster that they Max. speed. You could heard them, I was only able to stops them by rebooting the system.

That said, I se you can keep the 'snaps' in the up position, I would recommend keeping them in the up position and instead use the proper size LARGE HEAD screws for those spots.

Using the flashlight I noticed that there is thread in those holes, so yes, I can put screws in there to secure the card. Indeed, I already did it, and the movements got visibly reduced.

To my visual interpetation, the clips 'peg' that fits into the 'screw hole' seems small, and as both points on the hard are not simple drill holes (you see the wide gaps to allow easier compatibility with the myriad of cases that exist) but gaps to allow flexibility when inserting the double wide card. The pegs and the pressure on the clamps do not make a 'tight' seal enough to prevent 'wobble'.

Yeah, I noticed the same. Think I discovered the threats a bit late, lol.

If the screws work fine then lets test video out with a different (again friend?) monitor and cable. Does it work now? Then test your cable on friend's monitor? Does it work now? Now test with your monitor and cable? Does it work now?

My both monitors and cables have been tested before w/o any problem. The AOC with the DVI-D interface is the one I'm using with the on-board video output. No signal of problem, everything run smooth. 2nd Monitor, Asus, with HDMI. Was working w/o any problem before. I'm doing, as you suggested, a 2nd round of tests on them, I've them as video output option in my laptop, no problem so far.

If the card is securely screwed in, the cable and monitor all prove same performance no matter who's cable and monitor is used, and we STILL have issue, then proceed to the assumption of the PSU frying the hardware / your power source (house outlet / incoming power to the house). At this point 1) make sure your using some power strip or BEST use a UPS (Uninterrupted Power Supply) as that also protects not only against SPIKE (overload of power) but DROPS (underload of power) that maybe occuring, and usually lets you know when it happens so you can deduce the issue better.

I'm a bit skeptical on this one. Assuming there is a supply problem. Did happen 2 times? (remember 2 cards have been installed and both are dead)? Not only that, only happened when the video card was installed? The PC works perfectly using the on-board video.
I already contacted OCZ and tomorrow I will bve checking the voltage readings out of the PSU to confirm that are within the 5% recommended by them.

2) replace the PSU, maybe a 750-800?? (look up calculate computer watt usage to determine best solution for your specific system) does it still have a issue?

I've a 700W PSU, based on EVGA specifications a GTX660 based system only requires 450W PSU to be fully functional.

Another approach...I've a second post like this one in another forum. Over there, one of the suggestion that came up was a possible damage/failure in the Southbridge. I'm not familiar with the architecture of LGA 1155 / Z68, but still find a bit odd that a failure in the Southbridge only gets manifested in the graphics card. What you think? Could be this "Southbridge failure" a possible cause?

 

[Tom Tancredi]

Sorry for the delay, the last days of the week were very hectic at work...

>> Is fine and normal no problems

That is good, now about the wobble. Yes the card shouldn't move THAT much, especially as you push the frames the FAN increases and thus causes force to move making the pins at the bottom connect / disconnect (which is also a semi power source for the card as well).

Holy Cow!! now you mentioned Fans and made recall something important,...After both failure (both graphics cards) all the fans connected to the motherboard started spinning way faster, WAY WAY faster that they Max. speed. You could heard them, I was only able to stops them by rebooting the system.


>> This is a further symptom, EITHER when the GPU is / as it fails causes a backload short across the entire Mobo, or is actually causing the short inside the GPU itself and 'melting down' causing even more heat. Either cause would cause the exact symptom (all the fans kick on really high) your describing.

To my visual interpetation, the clips 'peg' that fits into the 'screw hole' seems small, and as both points on the hard are not simple drill holes (you see the wide gaps to allow easier compatibility with the myriad of cases that exist) but gaps to allow flexibility when inserting the double wide card. The pegs and the pressure on the clamps do not make a 'tight' seal enough to prevent 'wobble'.

Yeah, I noticed the same. Think I discovered the threats a bit late, lol.

>> MY concern is now we have a defective port / GPU we are trying to resolve with a solution too late after the damage is permanent now. I would likely suggest A) move the card to the other slot you mentioned then secure it down with the screws: does the card now work? B) remove card, use a different card in the 2nd slot, then test in the 1st slot. Does the 2nd card work fine? If so, the the GPU will be #3 broken / dead, if not (2nd card doesn't work) will depend on which slots (hopefully not both) it fails to perform in (aka PSU and/or Mobo fail at that point).

If the card is securely screwed in, the cable and monitor all prove same performance no matter who's cable and monitor is used, and we STILL have issue, then proceed to the assumption of the PSU frying the hardware / your power source (house outlet / incoming power to the house). At this point 1) make sure your using some power strip or BEST use a UPS (Uninterrupted Power Supply) as that also protects not only against SPIKE (overload of power) but DROPS (underload of power) that maybe occuring, and usually lets you know when it happens so you can deduce the issue better.

I'm a bit skeptical on this one. Assuming there is a supply problem. Did happen 2 times? (remember 2 cards have been installed and both are dead)? Not only that, only happened when the video card was installed? The PC works perfectly using the on-board video.
I already contacted OCZ and tomorrow I will bve checking the voltage readings out of the PSU to confirm that are within the 5% recommended by them.

>> To Summarize, you installed twice card into a slot. The card initialize then as used increase fan actions to cool the card the card wobbles breaking and reconnecting over and over and over causing undetermined power for the computer to use. The computer keeps trying to compensate, adding more power, etc. till eventually either the card shorts out and 'melts down' inside the GPU or causes (worse) a back surge in the power into the slot, across the Mobo (and arching to other parts in it's pathway) till it surges back into the PSU as well.
Again this is based totally on 100% perfect power conditions (I had no clue really about where you are, the type of building your in, the age of the wiring, the source of the power, how reliable the company is, etc.). I have seen in MAJOR cities a simple old 'retrofited building' that they jipped on the wiring which caused all the hardware to keep frying, to a simple suburb house 'out where the cow pasture used top be' that the local power company didn't factor in all the power used by the new suburban area, causing 'measurable dirty power' constant dips then over compensations. So to eliminate that I would ENSURE we have clean power with a higher end power strip OR best yet a simple UPS (plenty of cheap small ones at BestBuy, Walmart, etc.).

That said, as twice the card failed while connected to this mobo (making a metal connected pathway for the power to run), is the Mobo reliable in that port? (see test above) Is the Mobo at all also to blame now (your suggestion down below) because 'across' the I/O Chips got measurably damaged now too? Is the PSU safe (since you have a direct power to card from PSU allowing again a direct path not just across the Mobo, for the backsplash of the overload to damage the PSU)? So to test these, first make sure the input power is clean, then test the PSU in this reliable condition as suggested to determine other fail points.

2) replace the PSU, maybe a 750-800?? (look up calculate computer watt usage to determine best solution for your specific system) does it still have a issue?

I've a 700W PSU, based on EVGA specifications a GTX660 based system only requires 450W PSU to be fully functional.

>>> Yeah that is nice, but if you did as I suggested (https://www.google.com/search?q=calculate+computer+power+usage&oq=calculate+computer+&aqs=chrome.1.69i57j0l5.7224j0j1&sourceid=chrome&espv=210&es_sm=122&ie=UTF-8) you would get pages like this:

http://www.extreme.outervision.com/psucalculatorlite.jsp
http://support.asus.com/powersupply.aspx << Better!

which calculate ALL components using power. So while you may have a 700W PSU, and a 450W need, your using 749Ws because you installed 6 different HDDs and a extra RAID card to boot. See what I am saying?


Another approach...I've a second post like this one in another forum. Over there, one of the suggestion that came up was a possible damage/failure in the Southbridge. I'm not familiar with the architecture of LGA 1155 / Z68, but still find a bit odd that a failure in the Southbridge only gets manifested in the graphics card. What you think? Could be this "Southbridge failure" a possible cause?

<<< See above >>>

 

[yacard]

OK guys, after several days of back and forth with RMA services from OCZ and TradeSquare I finally got them approving my returns and replacements are on theirs way back to me. This is the main reason why I have not posted anymore here.
Big THANK to Tom Tancredi!! thk man for following my problem and helping with your advices.
Hopefully with new PSU and Mobo wont have this problem anymore.