MSI HD 7970 Lightning water block

[Callsign_Vega]

Hm, I tighten my screws pretty tight and I've had no issues with back plates or anything on my four Lightnings. Not sure what you guys are doing.

[l88bastard]

I've noticed that on 2 of my lightnings the stock back plate can get easily bent and cause a short around the vrm area. Caused them to seem completely dead. After messing with it a bit and ensuring there was no more contact they worked fine. You could try sticking a sheet of paper under the back plate to make sure.
Notice any light colored areas on the back plates where it seems the paint burned?

I completely removed the backplate and still have a dead card. Card #9 arrives on friday, if I cant get it to work then I am done with this setup

Hm, I tighten my screws pretty tight and I've had no issues with back plates or anything on my four Lightnings. Not sure what you guys are doing.

I also have sexual relations with my cards before install but I dont think thats the problem.

[jomama22]

2 things i suggest.
1. Make sure the TIM you are using is non-conductive. If you put any on the vrms,they can easily run over onto the pcb, i have had a card short because of this.
2. check the gpu die itself for cracks/breaks. Since AMD likes to just expose the dye, it can break easy as well.

If you sill have a dead card on you, i would just comb the pcb with a magnifying glass and see what is going on. There is most likely a common denominator between these problems that is being overlooked.

It is possible your mb is eating them. Have the cards all been in the same pcie slots?

[l88bastard]

2 things i suggest.
1. Make sure the TIM you are using is non-conductive. If you put any on the vrms,they can easily run over onto the pcb, i have had a card short because of this.
2. check the gpu die itself for cracks/breaks. Since AMD likes to just expose the dye, it can break easy as well.

If you sill have a dead card on you, i would just comb the pcb with a magnifying glass and see what is going on. There is most likely a common denominator between these problems that is being overlooked.

It is possible your mb is eating them. Have the cards all been in the same pcie slots?

(1) I am using Prolimatech PK-1 Nano Aluminum Thermal Compound. I also applied a tiny bit of TIM to the tops of the phase regulators as the waterblock directions sugguest so it is possible that some of that could have come in contact with the PCB, but I have done this with my three perfect waterblocked lightnings too and I use very small amounts, so I don't know if thats it. I mean what are the chances that this happened with three cards and not my other working three?

(2) None of the defective cards have any cracks or breaks. I haven't inspected with a magnafying glass, but there is nothing noticeable to the naked eye.

Its not the motherboard. As soon as I pull the dead card out, I can put any one of my three working lightnings in the same pcie slot and they work fine. I have tried all four slots and all power rails of my PSU so its not that....unfortunately.

The only thing all of my FOUR dead cards have in common is that they all used the SAME waterblock & MSI factory backplates. I have scoured this waterblock from head to toe but it does not look any different than my other three. I dunno, perhaps its possible that something on this block is causing the cards to immediately short out. Its the only common denominator as I have tried cards from different retailers so I don't believe its a bad batch of cards.

My leading theory right now is that waterbock #4 is either cursed or haunted
Lightning #9 arrives on Friday, I am gonna install the waterblock without any thermal compund on the phase regulators and I am not going to install the backplate either. If card #9 dies then I am throwing everything over my 5th floor balcony.

PS
I am not the only one that has had this problem. CurlyP from Overclock.net had the exact same problem with one of his waterblocked 7970 lightnings:
http://www.overclock.net/t/1295029/b...0-water-cooled

[jomama22]

Hm this is strange. I don't think it is the back plate honestly. If it was, I feel like there would be many problems even without the water block.

One thing about the water block installation that I noticed could be missed (as I made this mistake myself before realizing) the two different screw lengths included with the block could be mixed up. The screws around the gpu itself need to be the shorter of the 2 while the back plate issues the larger. I could see potential problems if using the larger screws around the gpu. It also says if not using the back plate to use the small screws all the way around the card. Could easily be missed if just grazing through the install.

I also also noticed with the Thermal pads that the long thicker one (the one that goes over the vrms) isn't quite long enough to cover the whole strip of vrms. If you stretched it it would just eventually unstrectch. I ended up using a bit of the shorter thick pad to cover the last 2 vrms in that row. Also, the vrms for the mem (the 2 on the right and 1 on the left) can be easily overlooked.

If the block were to touch any of these you would definitely short the board.

I really don't know what else to guess. Have you been using the same pads over and over? Because they seem to wear out quite quickly.

[Callsign_Vega]

The only thing I can think of is if the wrong standout(s) are on the block or aren't fully seated or the wrong length screws are used. This may cause some tweaking and cracking of the PCB/chip and it's connections. Always make sure the stand-offs are fully seated with a pair of pliers and and they are all the same height. Whenever I am done with a WB install I always take a high powered flashlight and give it a once over around the entire edge where the WB/PCB come together. I too use PK-1 on all of my installs, all over the VRM area etc and my four cards work great. It is non-conductive.

[l88bastard]

Hm this is strange. I don't think it is the back plate honestly. If it was, I feel like there would be many problems even without the water block.

One thing about the water block installation that I noticed could be missed (as I made this mistake myself before realizing) the two different screw lengths included with the block could be mixed up. The screws around the gpu itself need to be the shorter of the 2 while the back plate issues the larger. I could see potential problems if using the larger screws around the gpu. It also says if not using the back plate to use the small screws all the way around the card. Could easily be missed if just grazing through the install.

I also also noticed with the Thermal pads that the long thicker one (the one that goes over the vrms) isn't quite long enough to cover the whole strip of vrms. If you stretched it it would just eventually unstrectch. I ended up using a bit of the shorter thick pad to cover the last 2 vrms in that row. Also, the vrms for the mem (the 2 on the right and 1 on the left) can be easily overlooked.

If the block were to touch any of these you would definitely short the board.

I really don't know what else to guess. Have you been using the same pads over and over? Because they seem to wear out quite quickly.

Oh yes I am well aware of the different screw lengths and used the proper lengths for the cpu and block. Card #3 which lasted for a few days and died got recycled thermal pads from card #2 which never worked, when I did card #3s autopsy I discovered that two of the thermal pads never contacted the waterblock which may explaine its death. However, I bought brand new thermal pads for card #4 & #5 and used the proper thickness & lengths for the proper areas. Cards #2, #4 & #5 were dead immediately after the WB install so they didn't even have time to have thermal issues.

This is not my first go round with cooler installs. I did two accelero cooler installs with two 6990s and never had a problem with those and three of my 7970 lightnings work perfectly with the EK waterblocks. This is beyond crazy to me...I am about to buy the port block for the water bridge and just go Trifire as that seems a better solution than throwing my rig over the balcony.

The only thing I can think of is if the wrong standout(s) are on the block or aren't fully seated or the wrong length screws are used. This may cause some tweaking and cracking of the PCB/chip and it's connections. Always make sure the stand-offs are fully seated with a pair of pliers and and they are all the same height. Whenever I am done with a WB install I always take a high powered flashlight and give it a once over around the entire edge where the WB/PCB come together. I too use PK-1 on all of my installs, all over the VRM area etc and my four cards work great. It is non-conductive.

Ya, when I pulled the waterblock off of dead gpu 4 & 5, all of the pads & thermal compound had made perfect contact with the proper areas. I am gonna go over this gpu #9 install like a hawk. I feel like Tim "The Toolman" Taylor of gpus right now.

[jomama22]

I didn't mean to come off as though you didn't know or have experience, was just trying to cover all the bases. I just know we can all overlook things from time to time.

[Callsign_Vega]

You wanna video record the application of the WB on the next card so it can be dissected if it dies? Just think, you could be a star!

[l88bastard]

I didn't mean to come off as though you didn't know or have experience, was just trying to cover all the bases. I just know we can all overlook things from time to time.

Oh I know you were not implying that jomama, your advice has been greatly appreciated

But after 4 dead cards i certainly am questioning my own abilities and sanity

You wanna video record the application of the WB on the next card so it can be dissected if it dies? Just think, you could be a star!

Ya, I ordered an EK blank plug for the waterbridge just in case it doesnt work out I am gonna give up and just go trifire. However, according to murphys law, everything will work out perfectly if I video record the procedure and spend the money on a EK blank, so #9 should be a winnar!

[jimba86]

and im thinking of getting one these cards..this isnt good news.

[Callsign_Vega]

The cards are fine to purchase.

[l88bastard]

GREAT SUCCESS!!!

I got the replacement card today...one day ahead of schedule. I performed the operation doing everything I did with dead cards 4 & 5 SANS THE THERMAL PASTE ON PHASE REGULATORS. The fourth card is now working perfectly and its temps are inline with my other three cards which have thermal paste applied to their phase regulators.

Now I can't say for certain if thermal paste on the phase regulators was causing the problem but the argument looks very compelling. For clarification, I did not use much thermal paste on the dead cards phase regulators at all...just teeny, tiiiny pindot dabs from a toothpick and I did not see any thermal paste "spill" from the phase regulators on the PCBs of the dead cards....so who knows...who cares....all I know is I am FINALLY up and running again *KNOCK ON WOOD*

[Jonik]

[pgi947]

Any of you guys found a method of pushing over 1.35v with these yet? I think 1.4v would be nice for a few bench runs

[Gunslinger]

This has been the rig build from hell cause I cant get 4 lightnings to work with these blocks properly.

I have three lightnings watercooled with these blocks and they work perfectly, but getting a fourth card working for quadfire has been a nightmare. So far in total I have gone through EIGHT yes EIGHT MSI lightnings and of the EIGHT only THREE work perfectly.

The first defective lightning had runaway thermal temp issues with its stock cooler so I returned it. I did not try this card with a waterblock.

The second defective lightning didn't work at all, I did not test it before installing the waterblock so I dont know if it was DOA beforehand.

The third defective lightning worked fine with the waterblock installed for 3 days and then died

The fourth defective lightning worked perfectly with the stock fan cooler before I installed the waterblock, but was dead immediately after I installed the water block.

The fifth defective lightning worked perfectly with the stock fan cooler before I installed the waterblock, but was dead immediately after I installed the water block.

Since installing the waterblock killed defective lighting 4 & 5 I am going to go out on a limb and say the same happened to defective lightning 2. When I say dead I mean paper weight dead. The led & reactor lights do not light up and the system does not recognize them at all. I have tried all four pcie slots, tried different power rails from the PSU, etc, etc...but nothing, they are dead.

So what is happening? How did installing the waterblock kill at least three of my lightnings? Installing these waterblocks is a very easy procedure and I have successfully produced three other perfect watercooled lightnings, so I don't understand what is going wrong. I have done some research and seen that people had problems with EK waterblocks and EK backplates, but I am using the stock lightning backplates. I have seen some speculation about too much thermal paste, but I applied perfect amounts of thermal paste that did not go over the gpu. I have seen speculation about overtightning screws but I did not tighten screws once they stuck even though I could have gone another half turn or so. I have seen speculation about screws shorting out GPUs but I have used the EK plastic washers and screws.

I don't get it, this has been a total nightmare and I am ready to give up, cut my losses and sell everything.

Eerily similar to what I'm going through.

Bought 4x cards and 4x blocks, all the cards worked on air just fine. After installing the blocks, the bottom card was graveyard dead, exactly like yours.
I RMA'd that card, one month later when I finally received the replacement (which ran fine on air) I installed the block again and had the same scenario of a dead card.

Which brings me up to today. Instead of waiting for another month on RMA (both card and block), I just bought a new card and new block. New card runs fine on air, installed the block today, and have a 3rd dead card. I'm totally pissed, and not sure what can be happening. Installation of this block is quite simple and straight forward. The thermal pads all look to be making good contact, and I know I'm not overtightening the screws.

[l0ud_sil3nc3]

^this sounds like a total nightmare.

[Gunslinger]

^this sounds like a total nightmare.

So much so, I'm seriously considering offering the 2x cards I'm RMA'ing for a single working watercooled card.

I've got a 5.77GHz 3930K sitting here waiting to bench 4x CF and I can't install a block without killing the card...

[Elmy]

I also had the exact same problem. I only got 2 MSI 7970 Lightnings though and I installed EK's waterblock on both of them and one would not boot up at all. Same thing graveyard dead. IDK what is happening to these cards when you install the waterblock. Very strange. Kinda sucks because that was my favorite one that died. Could do 1325/1700 no problem.

I sent my card out 2 weeks ago... Just sitting here to waiting to play Crysis 3 on my 3 Asus VG248QE 144hz 1ms monitors :-/

I need to keep my FPS above 144 so I might be adding a 3rd 7970 Lightning here soon.

[Splave]

another EK WB lightning death here yesterday, gunny sent me his block to try on my card and mount it up....dead card wtf EK? this isnt rocket science!

[Gunslinger]

double post :/

[Gunslinger]

That was a brand new block I want to add, which was 1/1 in install/death rate for me and 1/1 for Splave. I RMA'd the block that was 2 for 2 in install/deaths. So in summary the last 2 blocks are a combined 4/4 install/deaths. WTF is going on.....

[chispy]

So sorry to hear so many Lightnings die with EK blocks definitely something wrong with those waterblocks I have heard so many people having the same issue all over the net , cards die after installing this EK waterblock , you guys might want to contact EK and let them know of this. Have any of you guys tried the Swiftech MCW-82 7900 edition ? , im using it and works very well on my Lightning , great flow and amazing low temperatures even when pushing lots of volts , guys give them a try it's a great waterblock , also is universal so you can re-use it on any card you like after your doen with the lightning ,I don't like full cover water blocks as it can only be used on one spesifically card and that's it. Good Luck guys and I hope the cards stop dying.

Regards: Chispy

[Splave]

sh1t at this point a universal block and leaving the stock vrm/mem sink and back plate is better than death! :X

[chispy]

sh1t at this point a universal block and leaving the stock vrm/mem sink and back plate is better than death! :X

Agreed that's exactly what I have done , just make sure if you push lots of volts get some good air flow on the stock vrm/mem heatsink and you will be fine