3770k IHS Removals - CPU temp dropped from 79C to 71C

[Chukundur]

It sounds like you have poor contact between the IHS and the core because you did not scrape the black silicone/sealant off the perimeter of the chip and IHS..

You assumed that I might have tried several types of paste without being completely removed the black silicone!? The liquid metal probably do, but I have no where to buy now.

[andressergio]

since i use MX4 or MX2 month after i take the block out and paste never dry its no so liquified as new but never what i saw on intel IHS that i really crap looked like cement or poxipol horrible

[babalouj]

You assumed that I might have tried several types of paste without being completely removed the black silicone!? The liquid metal probably do, but I have no where to buy now.

I did not intend to offend. You are just the first case I have heard of the temps slowly getting worse after replacing the stock Intel TIM.

[tscompusa]

dont waste your time with paste inbetween the IHS and die.. go straight to liquid pro/ultra.. it is an easy 10c drop over paste.. order it internationally if you have to.. its worth the performance increase.

[tscompusa]

Nice! Now you are getting almost the same temps as me with my exact same setup (liquid pro/ultra, pk1, and raystorm). I also found that the raystorm does not like direct die contact because of the bow.

Yep. ya I had the ac on in the house when i did that last test and now i have the AC off and i did a stress test and temps are almost the same and its atleast another 3c hotter in the room now .. and i noticed the cores are stabilized more temp wise.. they are more reading similar vs before they were kind of different.

[OldChap]

Those of you folks that have struggled to remove the IHS because there is so little clearance between it and the circuit board..... I wonder if you also experience worse performance than perhaps expected once the paste has been replaced?

This made me wonder if there may be some advantage in "lapping" the edges of the IHS that were such a close fit to relieve any pressure that may be stopping the die and IHS mating perfectly.

I am thinking that once the die and IHS have been cleaned and are ready to go to insert a feeler gauge to measure the gap whilst holding the assembly finger tight. ensuring a certain clearance might allow certain IHS's that needed lapping because of their high edges to drop the edges closer to the die circuit board under the pressure of the mount thus flattening the contact area with the block/heatsink too

[andressergio]

Those of you folks that have struggled to remove the IHS because there is so little clearance between it and the circuit board..... I wonder if you also experience worse performance than perhaps expected once the paste has been replaced?

This made me wonder if there may be some advantage in "lapping" the edges of the IHS that were such a close fit to relieve any pressure that may be stopping the die and IHS mating perfectly.

I am thinking that once the die and IHS have been cleaned and are ready to go to insert a feeler gauge to measure the gap whilst holding the assembly finger tight. ensuring a certain clearance might allow certain IHS's that needed lapping because of their high edges to drop the edges closer to the die circuit board under the pressure of the mount thus flattening the contact area with the block/heatsink too

i dont really get your point oldchap i got 10C less and better peformance now i can memtest my mem for 2 hours seeing barely 60 when before it was 70C or 72C or whatever test, i also got vcore down for my daily 5GHz

currently this are my settings i tighten up on the MIVE P67 the more i could i can go 2600 mem but will have to increase CL cauz its the board limiting

5GHz 1,488 loaded Team X 2400 9-11-11-28 1T tRFC 1,41 stock vdimm

and this is the part i like most when i had the 2600K at 5GHz HT at 1,425 loaded with mem at 2133 9-11-9-27 1T i have same score on windows than now

this is my 3570K score




Uploaded with ImageShack.us

[trance565]

I had a bit of a pain getting the IHS off, but i think it was more of an angle of approach error. I made an ever so slight lift in the blade, and it went thru like butter. got a good 10* drop, it's not that the contact is bad, it's the TIM is bad.

[Chukundur]

You are just the first case I have heard of the temps slowly getting worse after replacing the stock Intel TIM.

I hope that I am the only.

[tscompusa]

i dont really get your point oldchap i got 10C less and better peformance now i can memtest my mem for 2 hours seeing barely 60 when before it was 70C or 72C or whatever test, i also got vcore down for my daily 5GHz

currently this are my settings i tighten up on the MIVE P67 the more i could i can go 2600 mem but will have to increase CL cauz its the board limiting

5GHz 1,488 loaded Team X 2400 9-11-11-28 1T tRFC 1,41 stock vdimm

and this is the part i like most when i had the 2600K at 5GHz HT at 1,425 loaded with mem at 2133 9-11-9-27 1T i have same score on windows than now

this is my 3570K score




Uploaded with ImageShack.us

damn and i thought my 7.5 was good lol. my graphics card was the weak link at 7.5 then the processor a 7.8 then the rest 7.9

what oldchap is trying to say is he thinks that the IHS vary in size.. he thinks that because some are harder to remove then others that they are not putting as much pressure on the cpu die as others.. so he thinks you should measure the clearance with a feeler gauge to find out the exact height the IHS is sitting inbetween the pcb..

not a bad idea, but i would be surprised if they varied enough to matter.. that would be crappy manufacturing by intel.. then again this cheap TIM is not cool either (literally LOL), so who knows!

[Conquistador SW]

Those of you folks that have struggled to remove the IHS because there is so little clearance between it and the circuit board..... I wonder if you also experience worse performance than perhaps expected once the paste has been replaced?

This made me wonder if there may be some advantage in "lapping" the edges of the IHS that were such a close fit to relieve any pressure that may be stopping the die and IHS mating perfectly.

I am thinking that once the die and IHS have been cleaned and are ready to go to insert a feeler gauge to measure the gap whilst holding the assembly finger tight. ensuring a certain clearance might allow certain IHS's that needed lapping because of their high edges to drop the edges closer to the die circuit board under the pressure of the mount thus flattening the contact area with the block/heatsink too

I was one of those people complaining about how difficult it was to get anything between the IHS and PCB.

I did nothing except scrape of that black glue and this was the result using liquid pro and mx4:
(supreme HF/ PA120.3 / scythe GT 1000rpm / 18C ambient)



So for me lapping of the edges was not needed.

[Nico]

Ended up Direct Die cooling with Swiftech Apogee HD block and MX-4, about 10c better and that was with going from a 360 rad back to a 240 for case compatibility as well, so probably better by 12-15c.

Taking the socket clampdown mechanism and backing plate out made the motherboard flex a bit more, but straightened up a bit when fitting into the case Just added some plastic washers to the thickness of the IHS and some foam pads for some mounting protection.

Working on 5ghz staibilty at 1.45v 3770k, HT off as thats just way to hot with it on cores ranging from 65-76 running OCCT core0 is allways about 10c cooler, maybe the position or surround modules, not sure if uneven pressure is like to cause that, but possible I guess

[ToxicBug]

Hey guys,

Is there a consensus about which TIM works for putting between the die and the IHS? I put Noctua NT-H1 on both the die and the IHS and my temps dropped by 10C at first, but a day later they increased by 10C over stock!

The NT-H1 worked very well between the IHS and the HSF, so I'm thinking of putting the Arctic Silver 5 between the die and the IHS for now. Meanwhile, I'll order Liquid Ultra, I just don't want to wait a week until I can use my PC again

Also, I might be able to get some Shin-Etsu, is this a good option?

Thanks!

[theseeker]

Shin Etsu is very good.

[andressergio]

damn and i thought my 7.5 was good lol. my graphics card was the weak link at 7.5 then the processor a 7.8 then the rest 7.9

what oldchap is trying to say is he thinks that the IHS vary in size.. he thinks that because some are harder to remove then others that they are not putting as much pressure on the cpu die as others.. so he thinks you should measure the clearance with a feeler gauge to find out the exact height the IHS is sitting inbetween the pcb..

not a bad idea, but i would be surprised if they varied enough to matter.. that would be crappy manufacturing by intel.. then again this cheap TIM is not cool either (literally LOL), so who knows!

lol yes it scores real nice

i took 3 and all where th same exactly and so easy, maybe it was just luck, and i also think that is wise keeping the IHS over and dont go die alone

cheers !
Sergio

[st0ned]

I would, by my experience get a liquid metal based TIM.

[ToxicBug]

I would, by my experience get a liquid metal based TIM.

Any other brands to suggest other than Coollaboratory?

[andressergio]

Any other brands to suggest other than Coollaboratory?

i did on 3 chips used MX2 and all dropped 9 to 10C, i think they have so so bad contact from factory that using a normal good paste will do the job

[ToxicBug]

i did on 3 chips used MX2 and all dropped 9 to 10C, i think they have so so bad contact from factory that using a normal good paste will do the job

I'm assuming its performance is still good? I was expecting the same thing from the NT-H1, it really disappointed me :/

[st0ned]

Any other brands to suggest other than Coollaboratory?

Yes, I'm using the Phobya LM. I tryed PK-1 1st and it lost performance in 1 or 2 days, then I used a good chunk of PK-1 and it held the same performance for a week when I end up replacing it. Problem is even though I had good performance core-wise I had problems on the memory controller ( It wouldn't do 2400 Mhz ) and once removed you could clearly see the imprints from where the cores, memory controller, etc stood on the die . I'll try to see If I had any photos of this.

If I had to guess I think points of big heat influx such as cores and memory controller being where most of the heat is on the die will generate a big amount of heat / area thus drying normal TIM much faster then what it was supposed. I did a final mount using the LM TIM and I didn't have a problem ever since.

I just wanna state this is just my opinion based on the experience I had, I'm sure others might have had it different

[andressergio]

I'm assuming its performance is still good? I was expecting the same thing from the NT-H1, it really disappointed me :/

ya of course

http://skinneelabs.com/2011-thermal-...-comparison/2/

enjoy !

[andressergio]

Yes, I'm using the Phobya LM. I tryed PK-1 1st and it lost performance in 1 or 2 days, then I used a good chunk of PK-1 and it held the same performance for a week when I end up replacing it. Problem is even though I had good performance core-wise I had problems on the memory controller ( It wouldn't do 2400 Mhz ) and once removed you could clearly see the imprints from where the cores, memory controller, etc stood on the die . I'll try to see If I had any photos of this.

If I had to guess I think points of big heat influx such as cores and memory controller being where most of the heat is on the die will generate a big amount of heat / area thus drying normal TIM much faster then what it was supposed. I did a final mount using the LM TIM and I didn't have a problem ever since.

I just wanna state this is just my opinion based on the experience I had, I'm sure others might have had it different

are u sure that this aint cause of using direct die mate ? i been using it since 3 weeks now and not a single degree less just stays exact the same 9 to 10C less never went up again and im firing up prime 27.7 avx at 1,485 vcore

[tdbone1]

i have the P8Z77-I DELUXE and I5 3570K on the way.
it will be here in a few days so i been reading and reading

yea these chips run hot!
yea if you do this mod they will run cooler!

but the real question is?
void the inel warranty for an extra 100 MHz gain?
try re-selling it on ebay like that?

ok now here is a legit Q

a stock IB cpu IHS and stock TIM definately runs about 10C hotter then SB cpus but that does NOT mean it makes your case hotter on the inside.

ex:
1 x 100 watt bulb = 1 x 100F
3 x 50 watt bulb = 3 x 50F but since you have more it will make your case temps hotter then the 1 x 100F bulb

the surface area of the IB is smaller then SB so when IB = 75C vs SB 65C the SB could make your case hotter

so im asking what is the real purpose of this mod?
is it really just for the 100MHz gain and to void warranty and have trouble selling on ebay later?

[Church]

tdbone1: Purpose? Probably to get temps and max overclock closer to that of SB, which were gimped by worse stock TIM between core & IHS. According to many that is more of a reason for worse temps/overclock (at least without throttling and with not extreme cooling solutions) then that of smaller core/more concentrated heat flow/more current leakage when overvolted & overclocked a lot. I'm guessing frustration comes mostly from high expectations from IB before it's launch. If it were at least as good of overclocker as SB without things like improving TIM, no one would bother. But with state of things being as is people wish to "fix" "intel wrong decisions" with this, relatively simple mod. As for warranty .. very few from overclockers do care .. and from what i've heard, few people online even sell pre-moded IB cpu-s with replaced TIM for some price premium

[tdbone1]

tdbone1: Purpose? Probably to get temps and max overclock closer to that of SB, which were gimped by worse stock TIM between core & IHS. According to many that is more of a reason for worse temps/overclock (at least without throttling and with not extreme cooling solutions) then that of smaller core/more concentrated heat flow/more current leakage when overvolted & overclocked a lot. I'm guessing frustration comes mostly from high expectations from IB before it's launch. If it were at least as good of overclocker as SB without things like improving TIM, no one would bother. But with state of things being as is people wish to "fix" "intel wrong decisions" with this, relatively simple mod. As for warranty .. very few from overclockers do care .. and from what i've heard, few people online even sell pre-moded IB cpu-s with replaced TIM for some price premium

max temps on IB go upto 105C so running at 75-80C is nothing
my video cards i have had in the past would get upto 90C (i had 5870`s in crossfire) with not even a 1/4" space between them and everyone runs those like that still
gtx 480s ran in 90`s too
this is not uncommon to run that hot (as long as its designed for it)

to my understanding the IB is designed to run that hot

so my Q again is:
does it make the case temps higher running at 75-80C then SB running at 65-70C?
see i dont think so because the SB had more of a surface area (meaning it puts out a larger heat pattern)

ok so why do this mod?
it only gives 100MHz extra to void the warranty, risk damaging a nice cpu and resell value DROPS on ebay im sure.

im really thinking this mod is pointless if it dont bennifit anything?

let me know where im wrong at.