[Reborn] OC3D Delidded Ivy Bridge Club

[jamesriley94]

I know I have issues with my cooling atm...

Due to my terrible reservoir, it's near impossible to bleed to loop properly and so the rads are both full of air.

Been folding for the past few days non-stop and I'm seeing temps around 85-90 degrees C.
What's a little interesting though is that 1 core is at 65 max, whilst the others are all 85ish - so I definitely need to redo my thermal paste at some point. Don't see the point in taking my loop apart until I've got another reservoir though.

 

[Genesius]

One of my cores is 8-10 degrees colder still, even though I have liquid pro on it. But delidding got me a 30C drop, if I could get all cores as low as core 1 though it would be a 40C drop in total which is just ridiculous. I'm happ with my temps so the core difference done ant really matter to me.

How come you can't get the air out of your loop? What's wrong with your res?

 

[jamesriley94]

One of my cores is 8-10 degrees colder still, even though I have liquid pro on it. But delidding got me a 30C drop, if I could get all cores as low as core 1 though it would be a 40C drop in total which is just ridiculous. I'm happ with my temps so the core difference done ant really matter to me.

How come you can't get the air out of your loop? What's wrong with your res?

It's a single bay res, and the barbs are mid level for it. But due to the design, I can only fill it to 80%ish, meaning any slight knock on the case sends air into the loop. Any attempt to turn it upside down just sends air into it, and I can't even put the D5 above speed setting 1 because otherwise the water in the res moves too violently, and again sends air into the loop.

It's just a terribly designed res.
I'm gunna get a new one as soon as I can, but I'm already well into my overdraft so trying to put it off for another week until I get money

 

[PapaSmurf6768]

My tools:

Delidding wasn't that hard, what had me most nervous was how close together the PCB and IHS were, I though I'd have a little more wiggle room for the blade. I powered through it and once everything was separated, it didn't look like there were any nicks.

Forgot to take any more pictures, but after that I cleaned up the thermal paste and used my fingernail to scratch off the adhesive. Then I used CLU on the die and the IHS. Everything booted up no problem, was sweating bullets when I pressed the power button. Right now I'm running 4.8GHz at 1.38V 24/7 stable, temps are down 10 - 15C.

 

[nepas]

http://www.youtube.com/watch?v=2eFzNpTOaOw


Beats using a blade:lol:

 

[Genesius]

Nice work! Welcome to the club! What are your temps like at 1.38v?

 

[Genesius]

http://www.youtube.com/watch?v=2eFzNpTOaOw


Beats using a blade:lol:

That's a nice quick way to do it, but I would be worried about scratching the die on the underside of the IHS. Maybe if I practiced a lot before I'd use this way but the thought of using a bit of wood and a hammer on a CPU is a bit too far even for me!

 

[Master&Puppet]

Love this thread

So I did my yearly service this weekend and thought that I might as well set about delidding my 3570K too and make better use of my water cooling.

I used a single-edged razor blade which I held upside down and it worked like a dream separating the IHS quite quickly and easily.

My RASA block has seen a fair bit of use over the last 2 or so years that I've had it so I wanted to treat it to a make over first and thought that the IHS might as well get the same treatment whilst I was at it.

First off I bathed them both in white vinegar and salt for a few mins. The combination forms a solution of Acetic and Hydrocloric acid which attacked the tarnishing.

I then gave them a bit of a rough scrubbing and set about lapping them both.
Before:

After:

I used CoolLabs Liquid Ultra between both the cores/IHS and IHS/Waterblock. I found it very easy to use - dabbing a small drop onto the supplied brush and wiping it evenly across each surface. I put it back together in 3 stages:

1. The processor into the socket.

2. I dabbed the Liquid Ultra onto the cores and a little bit of MX-4 onto the underside edges of the IHS to make it stick before clamping it into the socket (you can see the grey MX-4 leaking out a bit below):

3. And all back together:

After an hour or two of refilling the water cooling and letting it settle and bleed I put back on my OC to see the change. It was running at 1.355 @ 4.6 (Prime95 48hr stable, not that it is an awesome chip anyway). For a temp test I used Intel Burn Test (AVX) on very high, 10 runs.

Previously I'd see a highest core temp of 80-85C (~60-65C delta).
I now see a highest core temp of 63C in 17C ambient (46C delta).

Very happy with that . Next I'll try and see what difference that makes to stability when pushing it a bit further...

 

[FTLN]

Love this thread

So I did my yearly service this weekend and thought that I might as well set about delidding my 3570K too and make better use of my water cooling.

I used a single-edged razor blade which I held upside down and it worked like a dream separating the IHS quite quickly and easily.

My RASA block has seen a fair bit of use over the last 2 or so years that I've had it so I wanted to treat it to a make over first and thought that the IHS might as well get the same treatment whilst I was at it.

First off I bathed them both in white vinegar and salt for a few mins. The combination forms a solution of Acetic and Hydrocloric acid which attacked the tarnishing.

I then gave them a bit of a rough scrubbing and set about lapping them both.
Before:

After:

I used CoolLabs Liquid Ultra between both the cores/IHS and IHS/Waterblock. I found it very easy to use - dabbing a small drop onto the supplied brush and wiping it evenly across each surface. I put it back together in 3 stages:

1. The processor into the socket.

2. I dabbed the Liquid Ultra onto the cores and a little bit of MX-4 onto the underside edges of the IHS to make it stick before clamping it into the socket (you can see the grey MX-4 leaking out a bit below):

3. And all back together:

After an hour or two of refilling the water cooling and letting it settle and bleed I put back on my OC to see the change. It was running at 1.355 @ 4.6 (Prime95 48hr stable, not that it is an awesome chip anyway). For a temp test I used Intel Burn Test (AVX) on very high, 10 runs.

Previously I'd see a highest core temp of 80-85C (~60-65C delta).
I now see a highest core temp of 63C in 17C ambient (46C delta).

Very happy with that . Next I'll try and see what difference that makes to stability when pushing it a bit further...

Great work M&P, just wondering though why you used that MH4 around the edges of your IHS. Im sure if you got rid of it you could easily shave a couple more celsius.
Welcome to The Club

 

[Genesius]

Great work M&P, just wondering though why you used that MH4 around the edges of your IHS. Im sure if you got rid of it you could easily shave a couple more celsius.
Welcome to The Club

Welcome to the club M&P!

Ive seen others use a high heat silicone to stick the IHS to the PCB. That works well and is much easier to remove than the old Intel glue.

Personally I would use silicone to secure the top if its a permanent fixture. Ive used nothing at the moment because I will probably take the top off again and add some more CL Liquid pro later in the year to see if I can get all my cores about the same temp.

Nice lapping! ill have to try that again next time the rig is apart

 

[Master&Puppet]

Thank you

Yea I had the very same thought about using silicone as a long term solution too. I used the MX-4 just as a quick fix, though it will probably turn into a permanent one lol (didn't know if it would work first time and whether I'd have to re-do it straight away). The paste is no where near the cores and I didn't think that there would be much heat being transferred from the PCB!

 

[coffeefoot]

You guys, this is some scary stuff !! I'm not ready to do any cuttin' yet but I thought I'd try that CLU though.

 

[Genesius]

Thank you

Yea I had the very same thought about using silicone as a long term solution too. I used the MX-4 just as a quick fix, though it will probably turn into a permanent one lol (didn't know if it would work first time and whether I'd have to re-do it straight away). The paste is no where near the cores and I didn't think that there would be much heat being transferred from the PCB!

well ive got nothing on mine at the moment, the IHS just sits onto the PCB. I tsill have a core that's 8 degrees cooler than the rest though so I want to play about with it when I get time and see if adding a bit more TIM helps

You guys, this is some scary stuff !! I'm not ready to do any cuttin' yet but I thought I'd try that CLU though.

its not that crazy, I think DICE is more extreme. This is just righting the wrongs that Intel made. Its daunting, and it can go wrong, but if you follow the guides then you SHOULD be fine. It depends on what you chips like. Ig you need to do it to get the OC you want then its the only way.

The die can only transfer so much heat to the IHS so fast and then no matter what regular cooling system you have (water/air) it wont help. If you need to or want to go further than that point then this seems to be the only way.

 

[grassman]

48 hours whats the point of that? What are you using your rig for? Your just killing the cilicone doing that, Of course it's going to start failing eventually.

 

[Genesius]

48 hours whats the point of that? What are you using your rig for? Your just killing the cilicone doing that, Of course it's going to start failing eventually.

Who are you responding to?

 

[grassman]

Last page master, forgot to quote.

 

[FTLN]

Last page master, forgot to quote.

Its all about being stable mate, if your folding or mining bitcoins then it is very important to be 24/48 stable..

 

[Genesius]

I still can't see any mention of what your talking about but I presume it's about bench testing for 48 hours to check stability.

It's not going to kill the chip running it hard for that long. Well not by any noticeable degree. What it will do though is make sure that it is as stable as if it were stock. That is very important to me because if anything happeneds I know that it's not OC related.

I check every OC with IBT P95 for at least 16 hours.

So I know the system will never over heat and never be anything but rock solid.
I dual boot OSX & windows 7. OSX throws all sorts of crap at you at random intervals so its good to know that that random crap isn't being caused by a poor over clock. I'd still do it if I didn't dual boot as I want my OC to be as stable as stock.

I think the pros out way the cons.

 

[Master&Puppet]

48 hours whats the point of that? What are you using your rig for? Your just killing the cilicone doing that, Of course it's going to start failing eventually.

Yea prime for 48 hrs is OTT but I do quite a few builds for people and I like to push my chips harder than I would theirs' to see the effects on stability, life and performance. Usually I do 24hrs which is plenty. They don't have the knowledge or inclination to tweak themselves and therefore I have to rely on a 'fire and forget' level of stability.

To be honest I don't much worry about the strain a day or two of stress testing can do once in a while - in the big picture I think it's pretty negligible.

@Genesis - I said it just under the bottom picture in my post

On another note... I've had a play and the drop in temperatures is doing wonders for my stability. Currently looking at 4.8 running on 1.355v under load, max temp 66C (ambient 17C). More playing to be done yet though...

 

[Scoob]

Hey all,

Just stumbled upon this topic...fascinating stuff! Im shocked what a bodge job the standard TIM under that IHS is, a really amature looking job.

I was wondering, do you think there's any way to forego the IHS entirely and mount the block / heatsink directly on the naked CPU? Surely that would be the best solution with the metal TIM in between? Not sure how you would go about adjusting things to clamp tightly, but not too tightly mind.

Not got an Ivy Bridge myself, but it's still intersting to see people get on and do this stuff & show what a poor job Intel did.

I've lapped several CPU's in the past - back when many enthusiasts hadn't heard about it - made a big difference...well, 5c IS big for us perfectionists lol. Realised earlier that I totally forgot I'd NOT lapped my 2500k! Can't find my lapping kit now - doh!

Anyway, well done to those of you who've gotten some good results from this slightly scary process & best of luck to those planning to try it.

Scoob.