Can this delidding catastrophe be salvaged?

[Chipfiref]

This is a continuation of this http://forums.anandtech.com/showthread.php?t=2334244 where I prepared for and execute delidding.



Here is what I was happy about

 

[Chipfiref]

It was actually pretty easy - with a bit of force the blade went right into the silicon glue and I could feel the IHS coming up as I went around on the corners and then the sides...then I noticed this:

Which with my friends macro lens looks like this:

I have gouged my pcb with the tip of that bade, real good.

Is this salvageable? I think it is, when I look at the traces on the pcb that you can see and there don't seem to be any under that gouge. Maybe a little superglue to cover that gouge?

that white thing is just a fiber I think. If you need more I can get much closer, but not really a better view of the gouge itself.

 

[crashtech]

So does it work, or not? Sealing the gouge is meaningless if the CPU is broken. If the CPU works, a dab of nail lacquer will seal it fine.

 

[Chipfiref]

So does it work, or not? Sealing the gouge is meaningless if the CPU is broken. If the CPU works, a dab of nail lacquer will seal it fine.

True - I want to put on CLU, which is pretty much a one shot deal I think, but for now I can just use some NH-T1 to check it. I will have to try it later tonight - I have a cookout and I am cooking...

 

[Schmide]

Looks way out of the traces. I think you got lucky.

 

[Idontcare]

I nicked mine like that, maybe not as large of a nick but I could see the copper like you can.

Just fill it in with something that will protect the copper from corrosion and your CPU should be fine. I used crazy-glue to fill in the nick on mine.

 

[Ferzerp]

I nicked mine like that, maybe not as large of a nick but I could see the copper like you can.

Just fill it in with something that will protect the copper from corrosion and your CPU should be fine. I used crazy-glue to fill in the nick on mine.

Didn't yours fail rather quickly?

 

[Idontcare]

Didn't yours fail rather quickly?

That was 3770K #1, which I didn't nick (IIRC). But 3770K #2 and 3770K #3 are still alive, both delidded, and one was nicked but patched with super-glue.

 

[aigomorla]

thats a big die... :O

and looking at TIM footprint... thats horrible... i wouldnt let any of my TIM imprints look like that on the outside of the IHS.

 

[spat55]

I reckon you might have gotten lucky, like Schmide said. This is why I don't do this just to save 10c!

 

[oneofusjustin]

should have used that vice

 

[Tsavo]

thats a big die... :O

Look at a Pentium Pro. ^_^

 

[Chipfiref]

should have used that vice

Yeah I could have probably used the vise to cut the silicone glue more carefully :|

Anyway will do the CLU later today!

 

[Chipfiref]

Have applied CLU to the Intel die and a little to the underside of the IHS. I kind of scared me at first because the CLU seemed to have a phobia of the IHS - kind of like beading up. But now there is a film on both the die and the IHS.

It has been a busy day with other activities, but I finally got to clean, seal the gouge (last night).

I had to completely remove my motherboard to install my Noctua bracket so that has been a pain. My old Cosmos case did not have a cutout for the CPU back plate.

Now I am searching for the video I saw of how to position the IHS on the pcb so I don't smeat CLU all over the inside of the IHS experimenting with the correct placement.

 

[hans030390]

I did the same thing when I removed the IHS from my GTX480 (granted, it was on the edge). Worked just fine. You're probably OK, but I can't guarantee it. Good luck!

 

[Chipfiref]

It works, it really works!

I successfully positioned the IHS so that it lined up on the die as it should. My retention mechanism was torquing one side further than the other, so I had to position to compensate.

Then I installed my Noctua using NT-H1 paste. Maybe someday I will try CLU between the IHS and the Noctua.

Temperatures are unbelievable.

Just stressing the CPU with Aida64

Stressing everything, which yielded throttling very quickly
before delidding

I have not seen temps above 57 even playing with the first Prime95 stress test.

 

[AkumaX]

:thumbsup; !!!

 

[Idontcare]

It works, it really works!

Nicely done! :thumbsup:

 

[Chipfiref]

Thanks to all of you for your help, encouragement, warnings, and all the information you have shared, especially Idontcare. I could not have done it without you!

Now I will continue to test and benchmark and build out this machine. And since I see such good results with the delidding and the Noctua NH-U14S, I see myself dabbling in a little overclocking

Also I will attempt to summarize some of the key points for delidding a 4770K this week, including things I have learned not to do.

 

[Ed1]

Thanks to all of you for your help, encouragement, warnings, and all the information you have shared, especially Idontcare. I could not have done it without you!

Now I will continue to test and benchmark and build out this machine. And since I see such good results with the delidding and the Noctua NH-U14S, I see myself dabbling in a little overclocking

Also I will attempt to summarize some of the key points for delidding a 4770K this week, including things I have learned not to do.

So what was the temp before with same load ?

 

[Chipfiref]

So what was the temp before with same load ?

I meant to post that since it is located in my previous thread...

Now, keep in mind that previous observations were with the stock Intel cooler, which seemed to be properly installed and TIMed.

CPU stress via Aida64, before and after delidding and CLU

CPU, FPU, and GPU (Intel 4600) stress via Aida64, before and after delidding and CLU

Well I don't have the before picture for stressing everything (there was trottling and 99C temps after 20-30secs), but here is the after

So I saw 40-50C decrease in temps with Noctua NH-U14S with NT-H1 paste PLUS delidding the 4770K and CLU between the die and the IHS.

I will accept that the Intel stock cooler may have had a pin pop out, but the paste imprint on the IHS did not indicate that. It may have had all 4 pins secure for a while and later while I was monitoring, 1 had popped out. With this temperature delta I am inclined to think that.


Many are reporting 20C delta with delidding. I may have gotten more by using CLU. No lapping, just thorough cleaning of the chip pcb and underside of the IHS, to get an good contact between cpu die and IHS. So that would point to 20C for switching from the Intel stock cooler to the Noctua.

Would anyone give different estimated numbers based on what you see? What should switching from Intel stock to Noctua bring?

 

[aigomorla]

ahaha so intel is gimping overclocking by not properly lidding the IHS.

Man... you guys see why i absolutely love intel lately.

 

[Chipfiref]

ahaha so intel is gimping overclocking by not properly lidding the IHS.

Man... you guys see why i absolutely love intel lately.

Yeah I am wondering if it is more of an unintentional manufacturing issue - it seems that the degree of expansion of the silicone rubber glue, or the misapplication (too much glue applied in one spot) is what is different about different chips. But at the same time, one could think that they were aware of this and there are bins of rejected chips due to the silicone issues as the cost of doing this kind of manufacturing?

Beyond the lidding issues however, the chip has limited ability to be overclocked, from what I am reading. That is, the voltage you have to apply past a certain level of overclocking exponentially increases the heat. Someone correct me if I have this wrong.

 

[crashtech]

One hypothesis is that Intel may have been seeing more early failure from mechanical stress upon the die than from excessive heat, which the chip is designed to avoid via throttling anyway. A large physical gap filled with goop (TIM) will eliminate mechanical stress on the die.

 

[Idontcare]

One hypothesis is that Intel may have been seeing more early failure from mechanical stress upon the die than from excessive heat, which the chip is designed to avoid via throttling anyway. A large physical gap filled with goop (TIM) will eliminate mechanical stress on the die.

And cost. I don't mean cost of the TIM, but the cost of the solder (as in a BoM reduction) as well as the manufacturing cost.

Going to cost you a lot more money to put a CPU through the soldering step than the expense of putting it through a TIM step. No heating/melting/setting of the solder. The time involved, the cost of the facilities, plus QRA on soldered items is a lot more intensive than QRA on a silicon-based TIM pad.

There are lots of reasons, all expense related when it comes right down to it, that might have motivated Intel to transition from solder to TIM for their mainstream desktop SKUs.

But I disagree with the school of thought that would argue to say Intel intentionally made the transition because they wanted to gimp the OC'ing headroom on the mainstream "K" chips.

That may have been an unintended consequence of the decision to save a few more bucks from the BoM and manufacturing expense but I do not believe it was done out of malice or an intent to manipulate the enthusiast to up-sell them into a LGA2011 platform.