Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz

Page 32 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
19,427
1,764
126
@aigomorla, Please clarify and expand on, either in this thread or elsewhere, what exactly you mean by disaster, and how exactly you're killing your cpu's. If temperatures are within acceptable limits (say < 70C), while under extreme loads, there must be other factors involved.
dma has explained it exactly.

He meant that you're flexing the die with a curved surface. Its like glass, flex it a little and it breaks.
yes so please make sure your using a flat surface when mounting a bare die.

I believe IDC did sand his block down flat.
In AIO's its safe, because they dont have as thin of a surface as per say EK and god forbid u do it on a koolance because u get rid of its beautiful nickle plating.
 

GPz1100

Senior member
Jun 10, 2001
354
3
81
@Dma, thanks for the clarification. I suppose it depends on how bad the bow is, and how much you torque it down. Running the raystorm block here on a delidded 3770k. I torqued it down to what felt right. Hard to describe verbally :). That is, there was still room to make it tighter, the springs were not fully compressed.

Time will tell. If I get 3-4 years out of it, i'll be satisfied.

@IDC, perhaps instead of using an pound/inch torque wrench would be a more accurate way of assuring consistent pressures. Although, i've never used a torque wrench on a oil drain plug. Have yet to strip a single one after 20 yrs of doing my own oil changes on a number of different cars.
 

C.C.

Member
Aug 21, 2012
28
0
0
dma has explained it exactly.



yes so please make sure your using a flat surface when mounting a bare die.

I believe IDC did sand his block down flat.
In AIO's its safe, because they dont have as thin of a surface as per say EK and god forbid u do it on a koolance because u get rid of its beautiful nickle plating.
I believe most of this is sheer non sense..How do I know? I have lapped EVERY Heastsink, and waterblock I have ever owned. I have never had any problems "killing" a cpu, including running them on bare die cpu's from the days of the AMD T-birds, to using a mobile Core2Duo (which has no IHS) in a custom desktop board, to my current de-lidded 3770K..

IDT, nice to see you opted to use the thin metal washer method I came up with somewhere in this thread lol..It has worked wonderfully for me, as long as you make sure the number of washers is equal!
 

bononos

Diamond Member
Aug 21, 2011
3,767
86
91
I believe most of this is sheer non sense..How do I know? I have lapped EVERY Heastsink, and waterblock I have ever owned. I have never had any problems "killing" a cpu, including running them on bare die cpu's from the days of the AMD T-birds, to using a mobile Core2Duo (which has no IHS) in a custom desktop board, to my current de-lidded 3770K..
......
The AMD t-birds had an additional hard resin covering the bare die to give it more strength to resist cracking during heatsink installation. Not sure about the mobile C2D. I think it makes sense being careful about messing about with the ivy's.
 

C.C.

Member
Aug 21, 2012
28
0
0
The AMD t-birds had an additional hard resin covering the bare die to give it more strength to resist cracking during heatsink installation. Not sure about the mobile C2D. I think it makes sense being careful about messing about with the ivy's.
His post, and mine, were reguarding his claim of many killed cpus coming from using heatsinks/waterblocks that aren't flat etc. I simply said that I have never killed a cpu, including quite a few bare die mounts, so he is clearly doing something wrong.
 

SonDa5

Junior Member
Dec 1, 2012
6
0
0
New here.


Thanks alot IDC.

This thread helped me alot in my delidding experiments with a 3570k.


Seems clear to me that delidding IB helps lower temps and thermal dissipation performance when proper mount and effective TIM are used.

I am running a bare die mount with DT Sniper water block.

Alot of time and planning went into applying my delidded cooling mods.


I did contact paper pressure tests with the DT Sniper with the IHS on and with the ISH off against the bare die.

With the IHS on with DT Sniper block




With the IHS off against bare die of 3570k with DT Sniper water block





Doing the contact paper pressure test allowed me to see how my block was making contact. By knowing this I knew how to best orientate block and what parts of the block made good contact. This also gave me the idea of adding high quality Fujipoly Extreme thermal pad around die for support and to help cooling.



No IHS. Direct contact between die and DT Sniper water block.

Mounted



using a EK TRUE 115X back plate for stability.



With CL Liquid Pro TIM and bare die against block not alot mount pressure is needed. Just a few turns on the bolts for the water block mount and it is good to go.

Well planned out delidding with proper mount on bare die leads to great results.
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
Hey guys, I am new to the forums and read every single post in this thread. Awsome information from the experts. I am going to dive right in and delid my 3770k and use liquid pro inbetween die and ihs. then use prolimatech between ihs and waterblock. I've never lapped anything before. Doesn't look to hard based on videos. Is it worth lapping an h100? How do I do it? Is it the same process as lapping ihs? So much misinformation on the net regarding lapping I thought I'd come straight to the pros.
 

hokies83

Senior member
Oct 3, 2010
837
2
76
Hey guys, I am new to the forums and read every single post in this thread. Awsome information from the experts. I am going to dive right in and delid my 3770k and use liquid pro inbetween die and ihs. then use prolimatech between ihs and waterblock. I've never lapped anything before. Doesn't look to hard based on videos. Is it worth lapping an h100? How do I do it? Is it the same process as lapping ihs? So much misinformation on the net regarding lapping I thought I'd come straight to the pros.

It is not worth lapping an h-100 unless the surface is no longer pure.

As far as lapping the IHS once u do that u have no chance for a warranty.

so pick ur direction watch videos and take your time.. And most of all good luck.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
Hey guys, I am new to the forums and read every single post in this thread. Awsome information from the experts. I am going to dive right in and delid my 3770k and use liquid pro inbetween die and ihs. then use prolimatech between ihs and waterblock. I've never lapped anything before. Doesn't look to hard based on videos. Is it worth lapping an h100? How do I do it? Is it the same process as lapping ihs? So much misinformation on the net regarding lapping I thought I'd come straight to the pros.
I lapped my H100, there is a thread on it in the Cases and Cooling subforum.

I'm sure no two H100 waterblocks are the same, so YMMV.

In my case, my H100 was distinctly concave (formed a bowl, the the edges being higher than the center) which meant the TIM was pooling in the center to bridge the gap.

Lapping my H100 reduced 5GHz temperatures on 2600k by some 5°C IIRC.

Initial: (I added blue lines with a sharpie pen)



After a few passes on 220 grit, the marker lines remaining in the center is proof that the center was concave.



FWIW, here was what I had to say about my H100 at the time of lapping:

...anecdotal comments -> I've lapped a fair amount of CPU's and HSF's in my time, this H100 was by far the most twisted, warped, pain in the freaking ass surface I have ever tried to lap.

Sure it looked flat, but this thing must have been taken out of a heat treatment furnace and dropped straight into a bucket of ice water to be as concave and contorted as it was.
Yours may be in much better starting condition than mine, the above was from an H100 I bought nearly 18 months ago.
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
Thanks for the reply IDC. What's the best way to lap an ihs on sandpaper. Some people go up and down, others circular, and some clockwise rotating 45 degrees. Also is using oil and polish really necessary? Finally, could I sand the ihs after I delid, that way I wouldn't have to worry about bending pins?
 

Yuriman

Diamond Member
Jun 25, 2004
5,530
141
106
I'm not sure what kind of polish you have in mind, but you definitely do *not* want to coat the base with anything.

I tend to use a figure-8 pattern when lapping.

There are no pins on Ivy Bridge btw.
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
I'm not sure what kind of polish you have in mind, but you definitely do *not* want to coat the base with anything.

I tend to use a figure-8 pattern when lapping.

There are no pins on Ivy Bridge btw.
People tend to use high grade metal polish?
 

BonzaiDuck

Lifer
Jun 30, 2004
15,233
1,168
126
People tend to use high grade metal polish?
I've been "lapping" the processor IHS since early 2007. A lot of people start with 320 or 400 grit wet-or-dri sandpaper, then work up to 600 followed by 1000-grit.

Some have also noted that leaving fine scratches is "theoretically" non-optimal, but at that point, I'm not sure it makes much difference. Besides, if you finish with 400-grit and keep the paper from loading up with copper particles, enough wear on the sandpaper will make it behave like a finer grit.

Someone may try and disprove this, but it seemed more important for the thermal advantages to get a perfectly flat surface and remove the nickel plating.
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
Thanks for the reply IDC. What's the best way to lap an ihs on sandpaper. Some people go up and down, others circular, and some clockwise rotating 45 degrees. Also is using oil and polish really necessary? Finally, could I sand the ihs after I delid, that way I wouldn't have to worry about bending pins?
ha ha, you asked the right person here ;) This is one place where my college education happen to be perfectly applicable to what we are doing as hobbyists in one-to-one fashion.

My bachelors degree is in Materials Science and Engineering (MSE), and one of the things you get to do for a year while pursuing a BS in MSE is you get to take metals classes that include lab classes - think chemistry class only with smelting liquid metals and so on using mini-furnaces and the like.

Well one full semester of these metals labs is dedicated to learning the proper technique for preparing metallographic specimens for magnified studies (taking pictures of grain structure under a microscope).

We had to learn all about the proper way to sand, lap, and polish metal to remove scratches (deformed material, undesired metallographic structures), worry about inclusions, pitting, staining, discoloration, parasitic oxidation, etc etc.

So when it came time to apply that knowledge to the growing hobby of lapping my CPU's IHS or my HSF it was second-nature for me.

Now you ask - how do you it then? Well for 220 grit and 400 grit I very lightly wet the sandpaper. For higher grits I use dry sanding only. No oil, ever. You want to use water because it is miscible with IPA and you want to use IPA to rinse and then dry your CPU/IHS between successive grits of sand paper. Oil cannot be fully/easily removed from the surfaces and it will entrain nodules of grit from the prior sandpaper and carry it over to the next higher grit paper, causing inadvertent scratches.

You can get away without using water on the 220 grit and 400 grit if you like, I only use it to prolong the life of the sandpaper itself. The water is not there to help the lapping of the IHS or HSF.

I intentionally do not use water above 400 grit because the surface cohesion of the water actually makes the surface of your IHS or HSF stick to the paper and you will notice it starts to stick and then skip off the paper. This is actually causing more work for you because the skipping creates even deeper scratches that must then be removed at higher grits with added lapping time.

As for motion - 90°C rotations, nothing more, nothing less. It is intentionally done this way so that the scratches created from the prior lapping efforts are being cross-cut and removed at the optimal rate across the entirety of the surface. Doing figure eights, 45-degree rotations, etc is actually not uniform and not optimal (that is an educated statement, not just a guess or an opinion). So if you are going to put in all the time and effort anyways, may as well maximize what you get from the effort.

90-degree rotation every 10-15 swipes (push away, pull back = one swipe).

That said - there is a time and a place for doing 45° rotations and figure-eights and that is when you are starting out on a new grit and you want to see how uniformly you managed to lap the surface with the prior grit (check to see if you off-balanced the specimen by applying too much downpressure on one side or the other).

In that case a very brief figure-8 (maybe 2 or 3 loops) or a few 45-degree swipes followed by visual inspection of the uniformity of your newly minted scratch pattern will tell you if you left the previous grit too soon and perhaps you should go back to it again. If you leave a grit to soon, trying to rush through and save time, the problem is you find yourself spending forever and day on the next grit trying to remove scratches that should have been removed with the prior grit.

For example, maybe you created a particularly deep scratch from a nodule on the 220 grit. Normally you'd remove it with the 400 grit before moving on to the 800 grit, but if you rush through the 400 grit then when you get to the 800 grit paper you will spend a lot of time removing the same 220 grit-induced scratch because the 800 grit paper makes for much slower removal rates.

If you read through my NH-D14 vs H100 thread in cases and cooling subforum, or the lapped my 3770k thread in this subforum you'll get a good idea of what to do.

That said, while I may have been trained on doing this stuff I can also tell you that doing it the other ways is not going to necessarily be a bad thing either. Worst case is you use oil and you lapping efforts are undermined by scratches that never go away, or you do figure-8 type polishing and you probably spend just a bit more time lapping versus the time you would spend if you did 90-degree rotations.

In other words, there is no wrong way, but there an optimal way when it comes to how much time it is going to take you to achieve a given result.

The one thing you don't want to do is you do not want to use a buffing compound at the very end just to get a mirror-like polish. The buffing compound will actually create a chemical oxide on the surface that is thicker/deeper than that created by the native oxide, and the thicker chemical oxide will have inferior thermal conductivity (running counter to the very reason you are lapping your CPU in the first place).

If you want a nice mirror-like polish then you should use 3000 grit (a dry grit available in a pad format at auto-parts stores) as your final polish step. (the following images are all from various CPUs and HSFs that I lapped to 3000grit)











 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
Thanks IDC that was very helpful! I will delid my processor as soon as my liquid pro come in. Will post the results here.
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
IDC I remember you stating you scratched the underside of your IHS. Would multiple scratches or cuts effect temps? Or is the main concern with the top of ihs? I successfully delidded but scratched up the underside with my razor trying to get the black goop off
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
IDC I remember you stating you scratched the underside of your IHS. Would multiple scratches or cuts effect temps? Or is the main concern with the top of ihs? I successfully delidded but scratched up the underside with my razor trying to get the black goop off
The scratch is not likely to be a problem unless it is extremely deep.

Are you planning on delidding for the purpose of replacing the CPU TIM and then you'll put the IHS back onto the cpu? Or are you delidding with the intent of mounting your cooler straight to the die?

If you plan on relidding with the IHS then you may want to upload some photos of your IHS so we can see if the scratches are just superficial or if they will be expected to be problematic.

You can polish the scratches out on the underside of the IHS but it is a real tedious PITA because you are working inside a cavity. We'll only worry about that if you need to actually cross that bridge though.
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
I'm an idiot. I delidded perfectly but screwed up the easiest part. Lol. As you can see from the photo I got a little over zealous around the perimeter. Will this be a problem? The center is untouched though... Recommendations? Will be reattaching the ihs. If I need to ill figure out how to go straight block... I'm a noob at this stuff :)

 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
I'm an idiot. I delidded perfectly but screwed up the easiest part. Lol. As you can see from the photo I got a little over zealous around the perimeter. Will this be a problem? The center is untouched though... Recommendations? Will be reattaching the ihs. If I need to ill figure out how to go straight block... I'm a noob at this stuff :)

Oh that looks great, you got nothing to worry about.

Proceed as planned
:thumbsup:
 

Idontcare

Elite Member
Oct 10, 1999
21,118
57
81
IDT, nice to see you opted to use the thin metal washer method I came up with somewhere in this thread lol..It has worked wonderfully for me, as long as you make sure the number of washers is equal!
:thumbsup: Yeah man, you were on the money there. Sorry for not giving you proper recognition for your idea, that was my bad as I try and make a point of giving credit where credit is due. For some reason I couldn't remember who exactly had recommended it, and I searched around for a while with no luck, so I concluded I must have been remembering it wrong and figured someone would remind me eventually as to who's idea it had been.

So, thank you good sir, twas a fine suggestion and it worked out grand as planned
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
Successfully delidded. Put liquid pro on the die. Using Prolimatech PK3 between ihs and water block. using h100i in a corsair 650D. 26c ambient temp. I opted not to sand. Getting 29-30c cpu idle. I'm thinking I need to reapply the thermal between ihs and block. Cores 1&3 are idling around 37c but cores 2&4 are idling 27c. 10c difference. Overclocked to 4635 vcore 1.35. Onload temps steady out at 68c. Up to 4800 temp cores 2&4 77 and 1&3 85. Maybe I need to reapply or sand. I can live with these temps but not the vcore. Vcore at 1.35 at 4635 seems a tinny bit higher than I'd like. Any suggestions on lower vcore?
 

Joshdm2001

Junior Member
Dec 1, 2012
8
0
0
Running prime95 will post results. Actually my temps at 4.8 are a lot lower than I thought. Will post screen shots.
 

GPz1100

Senior member
Jun 10, 2001
354
3
81
tw33k, What were you running to load the cpu with? For apple/apples comparison, you should stick to what's been used by others in this this thread - prime95 and ibt.
 

Dadofamunky

Platinum Member
Jan 4, 2005
2,184
0
0
Man, those lapped CPU/HS mirror images above are outstanding. I think IDC has a new profession waiting for him when he retires - lap people's chips for a living!
 

ASK THE COMMUNITY