3770K De-lidded and lapped With Liquid Pro - Awesome results.

[Stewart@MSD]

Hi Guys,
Took the plunge and de-lidded my 3770K as even with my huge loop, 90c at 1.28v and 4.7ghz was stopping me testing any further and I really want some more clocks. Also, my core temps were very uneven which has always niggled me.

Plus - I just wanted to do it for the hell of it, so off we go...

Quick test beforehand using ten runs of IBT on normal yield the following results.

• Idle Core 1 = 31c
• Idle Core 2 = 26c
• Idle Core 3 = 29c
• Idle Core 4 = 23c

• Max Core 1 = 76c
• Max Core 2 = 88c
• Max Core 3 = 83c
• Max Core 4 = 79c

After the work was completed, I performed the same test, and yielded the following results.

• Idle Core 1 = 27c (-4c)
• Idle Core 2 = 24c (-2c)
• Idle Core 3 = 26c (-3c)
• Idle Core 4 = 22c (-1c)

• Max Core 1 = 53c (-23c)
• Max Core 2 = 56c (-32c)
• Max Core 3 = 58c (-25c)
• Max Core 4 = 54c (-25c)

An incredible max improvement of 32 Degrees C!
And importantly, the cores are now all within 5c of each other, instead of 12c as they were previous, which I expect is a result of the lapping.

Just to be sure it wasnt a fluke, I put ona quick double run on Max settings. results are almost identical.

Pleased?
Oh yes! Now for some more volts and clocks!

Pictures to follow. just uploading.

 

[Stewart@MSD]

The bits n bobs...


She's under the knife...


No turning back now...




IHS was definately concave... that pleased me.








Pop her back in with some Liquid pro...




And that is pretty much that. All I can say is TAKE YOUR TIME.
Rush this and you will slip and wreck her... but take your time and its very simple to do.

 

[EarthDog]

First, awesome results!

Second, Im a bit confused. You said you delidded, but then say lapped? So you put the IHS back on with different TIM between?

(sorry I cant see the pics)... but reading the text, that seems like a yes. See line 1.

 

[Stewart@MSD]

First, awesome results!

Second, Im a bit confused. You said you delidded, but then say lapped? So you put the IHS back on with different TIM between?

(sorry I cant see the pics)... but reading the text, that seems like a yes. See line 1.

Thanks.
Yeah, stripped her down, lapped the IHS flat then put Liquid pro between the IHS and the Die, then Liquid pro between the Die and Waterblock.

 

[Mr.Guvernment]

thats some dam good results!

 

[soulwatch5]

Just to get things strait lapping is pretty much sanding the IHS for a more even and smooth surface correct?

Also another question I had about de-lidding the cpu is if there is anything that is used as a adhesive so the IHS stays on the die if you later have to remove the cpu?

 

[Stewart@MSD]

Just to get things strait lapping is pretty much sanding the IHS for a more even and smooth surface correct?

Yes mate it is, you just lap it flat with wet n dry paper.
The images above where you can see copper coming through on the outer show that the outer edges were raised, then i had a dip in the middle, RIGHT where the cores are!

Also another question I had about de-lidding the cpu is if there is anything that is used as a adhesive so the IHS stays on the die if you later have to remove the cpu?

You can choose to stick it back down mate, but I havent bothered as at some point I am going to try liquid ultra, just to see what difference it makes.

 

[Robert17]

What method did you use to get the OEM goo off the die?

 

[bing]

Not sure if this is applicable, if the cpu bracket can be easily detached, a direct cpu die cooling without the ihs probably could yield another better temp.

In order to protect the die from cracking from the pressure of heatsink base and also spread the mounting pressure evenly through out the chip green substrate, use this technique -> http://www.overclockers.com/forums/showpost.php?p=4838732&postcount=11

Since you've de-lidded it, I guess this worth a try, just a wild suggestion.

 

[Whiskey11]

My results were pretty much the same as yours with Liquid Ultra.

I went from 94ºC under load (1.385V @ 4.7 GHz) to 66ºC max. I'm now running at 4.8GHz with 1.450V and temps are barely into the 70's.

 

[larrymoencurly]

Also another question I had about de-lidding the cpu is if there is anything that is used as a adhesive so the IHS stays on the die if you later have to remove the cpu?

Regular silicone rubber sealant.

 

[Fever]

Great post! I'll be doing the same when my liquid ultra arrives. I'll try to remember to post my results!

 

[I.M.O.G.]

Impressive results!

That article I wrote so long ago was a lot of educated guesswork... Reassuring seeing all the people who have taken that idea and gotten results off it.

 

[Fever]

What would be a good way to re-seal the heat spreader? Would 4 dots of crazy glue work at each corner?

 

[voodoo do'er]

INB4 liquid pro eats the part under the IHS
that stuff can't be used with anything other then gold, nickel and copper. it will eat the cpu with time.

that being said, that stuff is amazing.

 

[Woomack]

If I'm right then it can be used with copper, gold and nickel but can't with some other metals and for sure not with aluminum. There is some reaction with IHS or maybe on the surface of it but if you put IHS and won't take it off then all will be fine. If you wish to reapply LCP after longer time then better use some sandpaper to make IHS clean/flat.

@Stewart@MSD , nice results you have there
Looks like it can vary a lot depends from cpu. In most cases it's about ~10ºC difference, sometimes ~20ºC but for the 1st time I see ~30ºC. Something was clearly wrong with IHS mounting.
On stock TIM there is 12ºC difference between cores so it was not so good. After de-lidding it's about 5ºC which is about what it should be at the beginning.

 

[Pierre3400]

I have to do this now!

I have questions. Did you "glue" the IHS back on?

This Liquid pro and liquid ultra, what is what, and does it have to be this when doing this?

 

[Woomack]

Both are the same. A bit different package and ultra has some liquid to remove it + something more. I used Phobya version which comes in exactly the same package as LCP but was ~30% cheaper.
All of these TIMs = Gallium.

 

[m715]

Nicely done, impressive temperature results
Don't think I have the nerve to do that yet...but your results certainly make it tempting!

 

[Stewart@MSD]

Thanks guys. I didnt bother re securing the IHS. After 48 hrs it should be welded in place anyway. And its not like it moves about.

After some tests and benching, my best overall settings with 100% stability are 45x103 for 4640mhz @ 1.27v with ram at 2200 with 1.55v.
I have also brought PLL voltage down to 1.7 for less heat.

So I have done some IBT tests set to max, with extreme ticked for ten runs using all the ram.

Results are very impressive. A max of 54C.

But perhaps even more impressive are the results of my same tests PASSIVE with no radiator cooling at all.
(All cooling fans switched off except the 2x silent case fans to keep the Vregs cool.)

Still a max of 67C. Well happy.