Lapping the 2.4C IHS

[Liquid3D]

I decided to buy a pack of fine assorted wet/dry automotive grit papaer, and went to work. I lapped the IHS down to the copper, cleaning with rubbing alcohol frequently, and it looks so pretty. I also lapped the SLK900U, which is done to a mirror finish, but after awhile, the silver particles get in the micro-groves. Remember copper is actually a porus material. I lapped it down, and it looks even better then when I got it.

Perhaps the most surprising thing I found was the uneven surface of the Intel P4. The corners had significantly more material on them. I used a L-ruler to ensure the surface was flat and then wrapped the paper around an perfectly flat edge and went to work. Obviously this is critical, because you want a flat surface. But as I said it took quite a while for the corners to get down. This is bad, because with the outside edges contacting the Heatsink, there forms a pocket where hot air would become trapped. No good.

Prooof is in the sanding. Yesterday in the same Ambient temp (my air conditoned room) my overclocking temp was 45C to 49C at 3.6GHz (300FSB) today same ambeint temp, and OC conditions, and the temp's are now 38C, and haven't gone above 39C. That's 6C to 11C temp drop! Next I'll be removing the IHS all together. Intel uses thermal paste which is not of the highest quality, and of course they don't it spread thin and even as we would, but simply shoot a dollup of it on the core, and epoxy down the IHS. The hole is for gases to escape or the thing wouldn't adhere correctly. The latter is speculation. But they do use way too much paste, and as everyone know's this can become as much as a deficit, as a dissipation method. I shopuld have took pic's but i was too lazy. I'll take-em tomorrow, and add them to this post. The cip looks much better lapped. ;)

[Major]

Good results ! I haven't tried lapping my P4's yet but I guess I should ! lol

[Liquid3D]

It's so easy, it's easy. When I first saw the copper i got nervous, until I remembered there's no way you could just lap through to the core. of course you'd see the thermal paste before you hit the core, but stranger things. I'm so silly. Anyway they must coat the IHS just so they can laser in the codes? I guess you wouldn't see it in copper, but i wonder if there's another reason it's coated? I think it's a coat of aluminum powder they put on it? Next I'm removing the IHS, I was going to do it, but decided to lap first. I wonder after removing the IHS if the SLK900U would be too heavy to place on there? Eeven with the more secure mounting system, which is much safer then a retention clip method, there may be shifting weight during installation which could crack the core.

I was concerned about static electricity, when lapping but cleaning with the alcohol frequently seemed to ameliorate any static build-up sinec the chip not only runs cooler, but runs period. Of course not having a wool carpet (or any) and not wearing wool is advisable as well. LOL

[Teus]

same here, silver paste got into the grooves of my SLK800. I'll give it a lap when I change some hardware. did you use your hands to sand it, or a sanding block?

[Creative]

Here are some pics of the 2.8 I had that the IHS came off when taking off the vapo evap head;

http://users.bigpond.net.au/creat/2.8/116-1625_IMG.JPG

http://users.bigpond.net.au/creat/2.8/116-1627_IMG.JPG

http://users.bigpond.net.au/creat/2.8/116-1628_IMG.JPG

Unfortunantly the chip is dead due to the IC7 and vapo not mounting properly and not getting contact. I know it was the IC7 cos when I removed the evap, there was no contact with the AS paste on the core I had put there......I slapped it in the IS7 to see if still alive, and although it got beautifull contact.....the PC would not even post....ala dead cpu.....twas a shame cos this 2.8 was benching pifast and superpi at 4010mhz :(

Since this accident, I am a little hesitant to go ahead and freely remove the IHS by choice in fear of killing another cpu.....I wanted to try it with the 2.4 I have here that benches at 300fsb and maybe get some more fsb, but I dont want to risk such a good cpu.....

Cheers
:)

[Peen]

Liquid3d , what was the finest grit sandpaper you used?i was thinking of lapping it first with 600grit then to 2000 grit that i have laying around. with that much of a drop in temps, im going to do what you did tommorow!

[Peen]

And is there anyway you can post a pic of it after you lapped it?
Sorry about your CPU creative, that would piss me off :mad:

And what can i polish it with to make it look like a mirror?

[STEvil]

hmm... shoudl we be lapping the core of the P4 like lapping the old Thunderbird cores too, now? hehe. 8-)

[Liquid3D]

Originally posted by Peen
Liquid3d , what was the finest grit sandpaper you used?i was thinking of lapping it first with 600grit then to 2000 grit that i have laying around. with that much of a drop in temps, im going to do what you did tommorow! I started with 400, and went to 600. Do you really want to see pic's? It's a pain in the butt, but i could throw in the other 2.4C and take pic's for ya. Maybe I should just lap the other one I have as well? The second one has the warranty, as it's a boxed version, alhtough it doesn't seem to OC as stable? It'll do the same speeds (3696MHz), same voltage, (1.550V) but for some reason I got a lot more memory errors at high FSB when benchmarking it?

Whe you lap the CPU, be really careful how you hold it. Dont do it on the motherboard, because the copper powder might get into some fine circuitry cause a short. I held it really carefull as not to bend the pins, and sanded away cleaning with alcohol every few seconds or so.

By the way the Epox 4PCA3+ doen't have the 3:2 memory divider? Anyone know why? They don't have any BIOS updates either, except one which addresses one issue; crackling sound in the sound system?

[La1kr0diZ]

BTW, intel uses shin etsu thermal grease and thats considered to be the best one around. I`ve used several engeneering samples without heatspreaders - they get more hot tnen the ones with HS instaled - AFAIK its better to leave the hs on da chip.

[CCW]

Originally posted by Liquid3D


Whe you lap the CPU, be really careful how you hold it. Dont do it on the motherboard, because the copper powder might get into some fine circuitry cause a short. I held it really carefull as not to bend the pins, and sanded away cleaning with alcohol every few seconds or so.


When people lap Northbridge chipsets they have to do it on the board because it wont come off, they usally tape a bag around it over the board, you could do that if you want to leave the chip.

IMO, theres no point, doesnt take 2 seconds to take a chip out but just means no chance of damaging chip while holding it I spose.

Craig

[Peen]

nah, i did a search on google and saw some pic of ppl doing it, ima take mine out and go to 2000grit

[Peen]

im scared of getting copper in the little vent hole!?!

[Peen]

ok i lapped mine and got a temp drop of about 5-7c under load

[Liquid3D]

COOL isn't it! I mean literally. And so dam easy. Doesn't it look better. Of course say goodbye to the warranty, unless you remove the IHS first send it back and tell them the IHS fell off somewhere, hehe. :eek:

[Peen]

it look kinda cool from far away cause i didnt polish mine, but round copper section in the middle and a lil copper on the outside

I used my sisters makeup mirror , wet the back of the sand paper and stuck it on their for a flat surface and so the sand paper doesnt slip and slide. First i used 600 grit and got it down to the copper , then used the 2000 grit to perfect it. Right now im kinda upset cause it seems to be running about the same temps :stick:

[STEvil]

Did you go down far enough that the whole IHS had some kind of surface area taken off?

I should scan my old Celeron 366... but its at home right now 8-/

[Major]

OMG I beleive, I beleive !! can I have an Amen !!!! :eek:

Ok just go through lapping both the 2.4c and 3.0c, havn't had time to do anything other than set both rigs back up (actually 3 rigs) but my temps are unreal !!!! I can't beleive the difference it made ! just now starting to OC the P4C800 and 3.0c in promie. I'll be back ! :D

[QuadDamage]

i did that too 2 days ago, but my temps went up... so i decided to reseat the evap and found the heatspreader moving around:D i was like wth? well, got rid of the heatspreader and enjoying direct block to core coolng:)

[Peen]

well on mine (and most others i bet) the middle is higher then the rest , cause it was the first part of the IHS that started to show the copper, i just went to when the copper area in the middle was about 2-3 inches

[Major]

Both the 2.4c and 3.0c that I have where lower in the center than the edges. On the 2.4c it was quite a bit lower. The 3.0c wasn't that far off from being flat.

[Peen]

My cpu is still reaching nearly 70c according to the winbond hardware doc on my IS7 :mad: . didnt really help , geuss i need better cooling

[Major]

Peen, one thing that you have to be careful of is using too much thermal compound especially when the two surfaces are free from pits and are truely flat. It should be very, very thin almost to the point of being able to see through it. Just a light hazing. You might already be appling it corectly, just wanted to point this out as a posible cause for your temps.

[Peen]

yeah, mines so thin that i could barely get it all spread out across the IHS , im guessing its cause the heatsink has very little actual HEATSINK lol

Edit: Major are you using AS3?

[Peen]

wheres the cheapest place i can get a SLK900u?
ok i figured out a better way to lap the IHS maybe, using 600 grit wet sandpaper and 2000grit wet sand paper.

1: Get a mirror and place on floor
2: Get the 600grit sand paper wet both front and back
3: Put 600 grit sand paper on mirror and water will help it not move
4: Slide you P4 across in a figure 8 until you see what looks like mud on the sandpaper (Actually copper)
5: Take sand paper and give it a good washing
6: Put sandpaper back on mirror and keep doing all this till ALL the IHS apears to look copper, and then do some figure 8's around the paper some more to make sure its PERFECTLY flat
7:Clean off CPU with alchohol
8: Now do this again with the 2000grit for quite some time to make sure all the groves from the 600 grit is gone, AND make sure that you clean the sandpaper regularly at least with water

I wouldnt worry TOO much about static electricity killing your CPU as i was doing this on shaggy carpet and no obeying any of the rules.

So basically the first time i did it, i didnt do a very good job and second time is doing about 2c better with a full load of about 64c (Which still sucks!)

Anyone who wants to OC their P4 more id say you HAVE to do this!

[Liquid3D]

I know this is an old thread but i never poated ta pic becauise i was too lazy to take the CPU out. Here's some Pic's I've taken. I'
m really impressed with this. One reason i've decided to take processor out is to remove the IHS, First I'm comparing it after lapping it almost through the IHS.

FYI I began with 320 ended with 600 Automotive Sandpaper. Taking a cloth, dabbed with alcohol and clean the IHS surface. When you get to the 600 paper, do the same to the paper as well preventing any possible friction related static build-up.

Holding the chip so it's not da,aged isn't that dificult. The pins on the Intel are signifcantly stronger then AMD processor socket pins. Simply place two fingers uder the pins, and your thunb on one half of the IHS, and then switch to the other side. Notice the reflection of my movie poster on the wall ten feet away in the surface reflection. Smooth as as a silk-worm's backside. You can also make out my poor Prometeia waiting to be RMA'd (sigh).

[Major]

I happen to have a dead P4C800 that I stuck my processors in to lap them ! hehe Made it real easy. Lapping made a significant improvement in my temps, but regardless of temps my processors are both topped out. I've had them in 4 different CW mb's now and they top out at the same OC in all the mb's with the exception of the 2.4c in the MSI Neo which didn't do so well.

I started to by a 2.8c today while I was in Atlanta, but it was a Costa and really want an OEM Malay this time.

I hope removing the IHS works out well, I'd like to try that myself atleast on the 2.4c I have.

[LikwidKool]

so using the sand paper wet is ok? wouldn't the water seep through the hole and get to the core? Just want to make sure I do this right when I have some time!

[Peen]

Yeah i used my sandpaper wet , and it worked fine

[Liquid3D]

Originally posted by LikwidKool
so using the sand paper wet is ok? wouldn't the water seep through the hole and get to the core? Just want to make sure I do this right when I have some time!

NO DO NOT use liquid of any kind!!! I said "DAB" the cloth in alocohol, meaning moisten it. And I only use it on the paper at the 600 grit. Wiping the paper with a alcohol moistened cloth should only bne done at the 600 level, because wiping any larger paper would just be akin to cleaning the grains off it. Periodically you can wipe the surface of the processor with an alcohol mopistened cloth, just to see more clearly where your at in the process.

Here's a picture which focuses on a reflection in the heat-sink's surface.

[DoGMaN]

Well damn it, I know what I am doing this afternoon :(

[DoGMaN]

Wow!! Mine is really bad. I'll post some pics later when I am done.

[shray]

liquid3d.. (little OT) Whats that thing next to the computer in your last pic?

[DoGMaN]

Here is the 'before'. I had used a little 600 grit and wanted to show just how bad my IHS was.
http://www.thecrucible.ca/images/unlapped.jpg

Here is the after

http://www.thecrucible.ca/images/lapped.jpg
http://www.thecrucible.ca/images/lapped2.jpg

I'll post some temps in a few.

[Peen]

dogman , yours is complete oppostire of what mine was, the core part was higher then the edges, but yours looks great now, temps improve alot?

[DoGMaN]

I had to lap my Alpha again so I don't have any temps yet. I just fired it up. I'll burn in the PCM+ and we'll see how she does.

[DoGMaN]

Weird as this sounds, it seems to have made my temps go up. :( I'm going to try and reseat the alpha again see if that helps.

[Peen]

i had the same problem , actually my temps staid the same as what they were. Then i tried to do it again and it helped. Now im kinda pissed cause my CPU is running 5c hotter! AND its perfectly flat. I dont think my crappy Zalman 5700u can handle the heat from the cpu

[Spawn]

wow looks good dogman...think ill have to do mine soon...hopefully do it on tues as ill be taking my pc apart to fit in my antec truecontrol 550 watt monster...lol

just a few q's tho...has lapping helped u guys get more of an oc out of ur cpu??...

also for the cpu wot sandpaper do i need???...and for the slk900u heatsink which sandpaper do i need??...and when u say alcohol..is that the white spirit sort of stuff u get from a pharmacists???......jd's wouldnt do the jobv would it???...:p: ...also when ur lapping whats the best method to do it...a straight up and down motion or a figure 8 motion with the sandpaper???

thanks for any info u can give on the above...:)

[Liquid3D]

Here's my new temps after applying the Nanotherm PCM+, and lapping down a wee bit more. The stuff is very strang. It's like there's "blue" solids, in a clear lquid base. Directions instruct to use just a few drops, and spread "very thin coat". But it's so hard to tell what the Hell your doing. You would need almost ten drops spreading even the thinnest layer, if covering the entire P4 IHS with the blue. I swera to God, I think what they've done, is taken their Nanother Blue II (which is the worst in their line0 and added a base to dilute it? STill the temops are fabo for 3.0GHz on the 2.4C. Here's another issue. My PCPower&Cooling is under volting the Vcore on five mobo's now. The vcore is t 1.550 to compensate, and you can see it keeps dropping to 1.488V. I'd say it's the software but on five mobos? And I always check the BIOS as well as the software in Windows. I wrote them, they said if it's VCpore it's the motherboard. I said "On five different boards?"

[Nanotherm]

Originally posted by Liquid3D
Here's my new temps after applying the Nanotherm PCM+, and lapping down a wee bit more. The stuff is very strang. It's like there's "blue" solids, in a clear lquid base. Directions instruct to use just a few drops, and spread "very thin coat". But it's so hard to tell what the Hell your doing. You would need almost ten drops spreading even the thinnest layer, if covering the entire P4 IHS with the blue. I swera to God, I think what they've done, is taken their Nanother Blue II (which is the worst in their line0 and added a base to dilute it? STill the temops are fabo for 3.0GHz on the 2.4C. Here's another issue. My PCPower&Cooling is under volting the Vcore on five mobo's now. The vcore is t 1.550 to compensate, and you can see it keeps dropping to 1.488V. I'd say it's the software but on five mobos? And I always check the BIOS as well as the software in Windows. I wrote them, they said if it's VCpore it's the motherboard. I said "On five different boards?"

The directions on the label say to "Apply small drop of PCM+ and spread very thin even layer across area of direct contact on surfaces of both CPU and cooling device." I know some people are having a hard time grasping the concept of a liquid TIM, but it's really quite simple. Do what the directions say and you're all set.

One small drop provides adequate coverage (in a very layer) of up to 3 sq. in. of surface area. In other words, a little drop goes a long way. All you want to do is apply a micro-thin coating - not frost a cake. If you used 10 drops to achieve what you thought was a proper application, then you wasted a lot of PCM+ unnecessarily.

And, no, it doesn't have to be a perfectly thin layer - it just has to be a relatively thin, complete and even coating across both mating surfaces. Of course, that doesn't mean the whole HSF or Waterblock - it means the top of the CPU die and the area on the HSF that actually touches the CPU die. I know you know that already, but others don't, hence my explanation.

Since the PCM+ is a liquid and not a viscous paste or "plumber's putty," the PCM+ will easily spread itself evenly throughout the thermal junction under the clamp pressure of the HSF or Waterblock. It a No-Brainer.

The problem is that there are myriads of people out there who are so used to the thick, viscous "thermal grease" TIMs, that it's really hard to grasp the concept of a thin, liquid TIM. Here's the key - don't think really hard about it. Put a drop on both surfaces, rub it around a bit and attach the HSF or Waterblock to the CPU. That's it! Nothing to it!!

Liquid3D, believe me, I'm not trying to embarass you or make you feel foolish in any way. This is a prevalent misconception that I come across on Forums all the time. And these misconceived notions are brought about by people's experiences using other thermal compounds. Consequently, you have no other frame of reference to work from, so you go with what you know.

Well, the PCM+ is completely different than any other thermal compound out there. It is not a thermal grease at all. And, NO, it is not Nanotherm Blue II - in any way shape or form - diluted, watered-down or otherwise. It's a unique "Thermally Conductive Phase Change Material" just like it says on the bottle. The only thing it has in common with the Blue II is the fact that it is blue, too - and a nice shade of blue, too, wouldn't you say? ;)

Let me summarize this post by stating that I purposely designed the PCM+ in a liquid form to simplify the TIM application process - and make a TIM that would be almost literally impossible to misapply. So, the moral of this story is - Keep It Simple - and you'll get great results. :D

[Liquid3D]

Your not "embarrassing" me at all. One thing I do know, it's that I spread thermal pastes much thinner then the average person. I always try to use the absolute thinnest layer, but of course full coverage. In fact my perfectionism where this subject's concerned, has me constantly explaining that no paste should be left to "hang" over the sides of the "core." Albeit only a few mm in height, any material will prevent air-flow from moving around the processor core's sides.

I realize now I used way too much paste, and I'd be willing to bet, if I'd used two drops instead of 6 (I said ten which was not accurate) my temps would be even lower. So i'm going to do it again this evening, (if here's any left) after i finish some writing. Then I'm writng an article on this entire subject, and belive me I'll be prasing the cooling prowess, and intelligent engineering behind your PCM+. If you read this (http://www.xtremesystems.org/forums/showthread.php?postid=198881#post198881) post you'll see I re-thought my comments made in the reply above. Below is an example of my observations, there's two materials which aren't mixing 100%. :) If you notice in the Pic. there's area's where the blue is of a lower viscocity, and I couldn't prevent the refletion, but where it's looks "pasty white" it's actually more clear and more "oily." Now I realize where not talking Jiffy peanut butter hear so "spreadibility" isn't so important. So long as the surface is covered (or lets say moistened) it's fine, is that correct? Because if one were to attempt to ensure an all blue coverage, they'd end up like me, with little, to no PCM+ left, which now I'm psyched about.

[Major]

I got some samples today in the mail and will be trying out the Blue II tomorrow.

One thing I would suggest is using some thin plastic (I use clingwrap) streched over the end of your index finger and just dab it around sorta like mop painting on the IHS, you can use a clean peice of plasic wrap to thin out the aplication. Just a suggestion. ;)

I picked up a OEM Mayla 2.8c today from a local Computer distributor that I will give a wirl tomorrow in the promie :D

[uclajd]

Originally posted by Liquid3D
Here's my new temps after applying the Nanotherm PCM+, and lapping down a wee bit more.

Liquid,

Is it possible that some people's temps go up because the CPU is now smaller, leaving a small gap between HSFs/waterblocks that have standoffs? My waterblock (Asetek) comes with standoffs that are meant precisely for the standard height of the P4. Sand off the IHS, and it's thinner, and the standoffs are now too tall. This would be a concern for me, unless I "lapped" the standoffs too. ;)

Your thoughts on this?

[STEvil]

Dont bother with standoffs for the P4? 8-)

Heres something else you guys could be missing.

Surface area.

The contact area between a bare AMD die and a CPU heatsink is much smaller, allowing for much more of the TIM to squeeze out easier.

The contact area between a P4 IHS and a heatsink is HUGE. The prssure required to press the TIM out properly from the junction between them is nearly insane when you think about it... so what Nanotherm says is completely true in the ammount that should be applied to a P4, but not necessarily an AMD.. although 6 would definately be a waste on an AMD.. heh.

Anyways, i'm looking forward to getting the stuff I was sent and comparing it to "Product-X" asap.. and no its not the Anti-Seize compound I was playing with earlier.. It slaughters that by over 5 deg c so far (and DOES NOT DRY OUT EVER!) and sits right in front of anyone's nose that lives near an agricultural area... lol 8-)

[Liquid3D]

Originally posted by uclajd
Liquid,

Is it possible that some people's temps go up because the CPU is now smaller, leaving a small gap between HSFs/waterblocks that have standoffs? My waterblock (Asetek) comes with standoffs that are meant precisely for the standard height of the P4. Sand off the IHS, and it's thinner, and the standoffs are now too tall. This would be a concern for me, unless I "lapped" the standoffs too. ;)

Your thoughts on this? Very astute observation, thnak you for brinign this up Tell you why this is troubling me, besides the prima Facie. When i was lapping the IHS it occurred to me, that I needed not only a perfectly flat object with which to wrap the sand-paper around, but the CPU had to be on a flat surface, such that the surface remained even. Same for the adjoining SLK900U heatsink. Call me crazy, but here's waht i've been thinking (albeit coincidence or not). Since lapping the IHS, i've been experiencing a consistent Bandwidth efficiency of 72%. I've benched three different memroy's, and on both channles. Using Corsair XMS3500, I'd though I'd try and attain some of the LL results people are getting using Kingston Hyper-X3500, in the hopes they used simliar IC's. At 2-5-2-2, with Intel Fast CS Enabled, and 240FSB 1:1 I came up with 5200MB/s. Then I tried the OCZ GolD4000, and Corsair Twin-X4000, and each time the "inefficency" was 72%. Lets say I lapped just slightly below the stand-off's height. Now there's less contact pressure under a specific area on the IHS. Because I used a perfectly flat instrument, but had no level to ensure I was lapping uniformly, lets say 30% of total surface area has 40% more material removed. Now let's pressume specific Core architecture responsible for memory related functions, reside's under said area, or akin to it. As a result of the uniform stand-off height, but the one area experiences reduced contact. The core architecture residing directly below said area, now experience's higher temps. Since the on-board diode can only measure that minute area directly beneath the CPU, there would be no way to determine, if there's inconsistent temps along the IHS surface. Ergo, localized temperatures may be high enough to adversly affect specifc processor functions. And if it were the area above the L1 cache, is it not feasable a bottleneck, or other thermal overload related anomaly may occur?

[eva2000]

The problem is that there are myriads of people out there who are so used to the thick, viscous "thermal grease" TIMs, that it's really hard to grasp the concept of a thin, liquid TIM. Here's the key - don't think really hard about it. Put a drop on both surfaces, rub it around a bit and attach the HSF or Waterblock to the CPU. That's it! Nothing to it!!

i've been using PCM+ on all my cpus and still get into this old habit of use to seeing how thick a layer of AS3 is compared to PCM+ :o

[Peen]

I was also thinking about the fact that the cpu is now shorter, but i tested how the presssure affected the cooling, and it wasnt a whole lot really. Remeber, my cooling isnt great with a zalman 5700cu but this is what i got. Full load temp of 59c with it all Snapped on the right way. Then i tried it with out snapping it down so it was lose. Went up to 62c full load , so im not worried to much about the whole pressure thing

[Nanotherm]

OOh, don't do that... Latch that thing back on. Pressure does make a difference and you don't want to create an air gap in the thermal junction between the HSF and CPU. You think the performance is poor now - just wait...

[Mr. Thompson]

You may want to read this. (http://www.overclockers.com/articles706/index10.asp) The thoughts Bill has outlined have been confirmed by Intel. ;)

[Liquid3D]

Thank you Mr Thompsom for the Overclockers.com link. The article for me, transcends prima facie dicussion of "flatness" under load. I beleive there's a more important, underlying lesson here. As Enthusiasts, much of what we do, is experimental. Therefore it should meet a testable criteria. There are different types of truth. A specific claim is either an empirical generalization, for which there is wide scale conformaton. Or a "conceptual truth" derivable from our concept of invention. The latter is a more abstract, creative process, the former much more important for our purposes.

My post above is hypothesis, based on certain knowledge of CPU architecture, however; it eschews any emprical confrimation. This does not necessarily detract from it's possibility, however; lack of any testable criteria, does detract from it's probablity. It's based upon Inductive Logic. Where a universal conclusion, is derived from a specific fact, or observation. The problem with the Inductive argument (modes ponens)is that it suffers from what is know, as an, "Inductive Leap." In other words, at some point we must "presume" based upon a tenative hypothesis, which itself is based upon, one or more observations. It's more creative, then empirical.

Albeit Deductive (Modes Tollens), Inductive (Modes Ponens) or Catagorical, as Enthusiats, we should be employing logic in our testing, of hardware. Only in this manner can we have a conversation with a common frame of reference. Otherwise were simply talking from the side of our necks.

If I'd taken my IHS, and merely lapped it with my finger, then I'd have no idea, if the sutrface was uniform or not. I actually disgree, with Intel's ad hoc remedy for the variences among heat spreaders. Not becauise it isn't the best remedy, but i wonder if it perpetuates complacency on the part of Heat Sink manufactuers, not to be overly concerned with flatness? I can see why Intel has done what they have, done, and had mr Thompson not provided us with the link, I wouldn't have known "all" P4 IHs were designed in this manner. Just about every person responding to this thread, seemed to experience the same phenomenon while lapping. They found the center wore away more rapidly the the sides. Yet on my ffirst 2.4C, the opposite occured. I'm too tired to use an example of Syllogistic logic here, but why don't you try? For example;
Premise-1: When lapping a concave IHS, the center will be exposed first
Premise-2: The P4 has a concave IHS.
Conclusion: Lapping a P4, results in center exposure first.

[nikhsub1]

I just removed my heatspreader altogether today and here is why... Each and every time I would take off my water block, the CPU would get stuck the the water block so strongly that it would yank the CPU right out of the socket, almost like the thermal paste created suction. The only way to get the CPU off the block was to slide it all the way to the edge of the block. Today, when i took my block off, it yanked the CPU out but this time, bent several pins. Now I was pissed. So, I got a straight edge razor blade and a pair of plyers and slide the blade under the spreader at each corner. The thing came off like butter. I cleaned it up. put some paste on the die, fixed the pins and put the block on. I was prepared for a dead CPU, either from the pins or taking the spreader off. Sure enough, it booted up no problem. My temps are now 4C lower. A good result I think. I also lapped the heatspreader down to the copper weeks ago, but saw no improvement over doing so.

The reason why I may be seeing such a big drop in temps without the spreader is because of the WB im using, the Cascade which is VERY focused.

Anyone know if the P4 die is more fragile than an AMD die?

[Liquid3D]

I belive the P4 core is much more so then the AMD, as it was designed differently. AMD obviously designed their core, with the intention of having the heatsink pressure directly on it. Not only that, but any designer worth their weight in solder, is going to design some extra duribility, to compensate for operator error.

I have the same exact problem everytime I remove the P4. It depemds upon which thermal transfer material you use, and how much is applioed. My ABit IS7-E, and IC7-G (for example) couldn't hold the CPU down while I removed the hetsink, the Epox 4PCA3+, has a more secure locking mechanism, and the CPU rem,ains frimly in plce when i remove the HS. Glad you brought this up. It was really bothering me, until i tried another board. I was concerned, in the locked position, pulling off the HS, may even pull pins out, and damage the CPU. I also had to slide it several times.

[nikhsub1]

Originally posted by Liquid3D
...Glad you brought this up. It was really bothering me, until i tried another board. I was concerned, in the locked position, pulling off the HS, may even pull pins out, and damage the CPU. I also had to slide it several times.
Yeah well I figured if i left the Heatspreader on, i would surely kill the cpu by bending/breaking pins off the bottom and that would suck. I guess I'll take my chances without the spreader, but I do like the temp drop! Gone with the CPU condome!

[Liquid3D]

Let me say it now, because someone will. "Your not praciting safe CPU-sex (overclocking) and you never know what you could catch (overheating)." :D

[nikhsub1]

With the recent heatwave in Los Angeles (95-100F) I have been forced to bring down the OC to 3450. Now with the heatspreader removed, my temps are rediculous! And this is an IC7 which people claim reads high. Room temp is about 29C.

http://www.netbetty.com/H20/3600temps.gif

[STEvil]

Lol liquid 8-)

[DoGMaN]

Liquid3D you can see from the pic I posted that my IHS was not high in the center either. Rather much higher on the outer edges. That particular pic is of a 3.0C not a 2.4C but that shouldn't make any difference.

[Peen]

nikhsub1 can u explain a little bit on how you u did it? some ppl say its tff getting it off, and i would like to take OFF my heatspread and stick some good thermal grease in there

[Major]

IMO it might make it easier to pop off the IHS if you put the chip in the freezer for a while. (It works with other thermal adhesives) Now that I have an extra chip I'll probably take the hs off the 2.8c and see what difference it makes too. ;)

[nikhsub1]

Originally posted by Peen
nikhsub1 can u explain a little bit on how you u did it? some ppl say its tff getting it off, and i would like to take OFF my heatspread and stick some good thermal grease in there
It is pretty easy, I dont thing the freezer is a good idea, that will make the epoxy brittle, you want it soft. I used a regular razor blade, a brand new one so it was sharp. I used a pliers to hold the blade. I then slid the blade under each corner of the spreader and gently made sure that i cut all the epoxy. Just take your time, and use a new blade and it is a sinch. Once you cut the bond of the epoxy, the spreader will fall right off. You will freak out on how light the cpu is without the spreader.

[Liquid3D]

nikhsub1 how high have you had that 2.6C, i'm thinking of adding a 2.6C to my collection, but was told as the chips increase in speed, they don't OC as well. Of course their all the same core .13-micron die, architecture, same Masks (templates), Photolithographic process (248nm), wafer's (200mm, and 300mm?) and only bined by multiplier, and vcore. Obviously Intel hit a home-run with the Northeood-D as they'll overclock at default vcore to 3.6GHz and above, and minme OC'd to 3.6 at defualt, and only ran at 36C. Just imagine if the NB and/or memory weren't holding me back? That's one advantage of the higher multipler, CPU clock speed, lower multi bandwidth. Their all most likely capable 3.6GHz, the only thing holding back my 2.4c fom 4GHz is it's multiplier is too low. The 875NB-MCH just can't clock over 300FSB very well without customizing something. My Abit would clock better then my Epox, had it more then 2.8VDIMM, and perhaps my Epox would clock better then my Abit had it higher then 1.6Vcore, and active cooling (although I've yet to feel my NB-HSF even warm).

dogman yes, it's very strange that article discussed Intel's designing the IHS concave to compensate for any variances among Heat Sinks, but I find this may not be entirely accurate as both of us had IHS's which were higher at the edges. Of the two I have, one is higher in the center, and the other the edges? That's not consistent in any respect? I'm taking off the IHS like I told Major two weeks ago, I procrastinated because I'd had so much succcess, but it's time to find some thread.

The safest way I've heard of removing the IHS, is to take a piece of thread, and saw through the epoxy. It's time consuming and strong thread is required, but it doesn't get much safer.

[uclajd]

Originally posted by nikhsub1
I just removed my heatspreader altogether today and here is why... Each and every time I would take off my water block, the CPU would get stuck the the water block so strongly that it would yank the CPU right out of the socket, almost like the thermal paste created suction. The only way to get the CPU off the block was to slide it all the way to the edge of the block. Today, when i took my block off, it yanked the CPU out but this time, bent several pins. Now I was pissed. Actually, pissed is understating it. I had the same phenomenon with my Volcano 7+ and Shin Etsu - I ripped the processor right out, without benefit of the lever. I don't care how well that $#!t cools, never again do I use Shin Etsu.

I think it killed my socket/board, because it was dead when I put my watercooler on. I swapped the 2.4c for a 2.53b, still dead, not even a beep. I dropped the 2.53, RAM, HDD, vid card into my old IT7, and it worked fine.

So I RMA'd the less-than-a-month-old P4P800DX, which I am waiting to get back. A hard-learned lesson. No apparent damage to my beloved 2.4c, which clock to 3.46 3D stable - on air - thank God.

Rather than ripping off the IHS - which I know would surely lead to me crushing the core for an encore - I'm thinking Nanotherm PCM+ for when the board comes back - does anyone know if this stuff is sticky?

[Peen]

I think im going to take off my heatspreader, but put it back on with AS3 cause with my luck the core will crack. I might just try to smooth out the bottom of the IHS with 600 grit maybe. Im still scared to take off the IHS also

Also about what someone was saying in the post above( cant remeber who) about the fact that when you sand it that the HSF doesnt apply as much pressure. This would be much more of an issue with no IHS

[Peen]

ok i took off my heatspreader too! it was easy as hell, right when u get under the IHS you dont even need to saw with a razorblade. Just a light push and its like the glue just peels. Anyways the grease that was under the CPU was hard as a rock. I also bent a few pins and they are OHHH so close to falling off :eek: , and the CPU is light as crap without the heatspreader.
As for temps? i didnt get a temp change yet, maybe cause its new AS3

Edit:6c drop in temps so far under full load

[PTK]

When you take off that IHS, how do you get that "rock hard" Intel original grease off ? I was thinking about lapping the IHS, but why not go all the way ;)

PTK

[nikhsub1]

I dont know that I recommend all of you take the spreader off! I did it because of what was happening with the cpu getting yanked out of the board, no other reason. I have been told that the die is very fragile so if you do this PLEASE be extra careful when mounting your cooling device.

[Nanotherm]

uclajd,

No, the PCM+ is not sticky. Just a quick note on the re-application of PCM+: Since the PCM+ forms a semi-solid, wax-like coating on and between both mating surfaces (which is how it's supposed to be) after application, you should remove your HSF or Waterblock within a few minutes after shutting down your PC while the CPU and HSF are still warm.

[uclajd]

Originally posted by Nanotherm
uclajd,

No, the PCM+ is not sticky. Thanks, Nanotherm.

Anyone who doesn't work for Nanotherm have a take on this stuff?;)

[Peen]

yeah but the temp drop is worth the taking off of the IHS. Plus i know i wont crack the core ;)

[nikhsub1]

Originally posted by Liquid3D
nikhsub1 how high have you had that 2.6C, i'm thinking of adding a 2.6C to my collection, but was told as the chips increase in speed, they don't OC as well.
I have had this CPU up to 3.7Ghz on my water in sig. This was when the ambient temps were much more reasonable...

[Liquid3D]

Well I certainly think the higher multipler will benefit me. 12x is great for bandwidth, however; if one is looking to attain the processors speed limit, multiplier should be high. But I couldn't afford a 3.2C (heck I cant even afford a 2.6C have to find a trader).:eek:

[Marvin]

The new stepping p4 IHS is harder to remove. As my 2.8 is a SL5WJ, i will try that tomorrow.
:stick:

[Liquid3D]

I wouldn't recommend removong the IHS from any current P4's. Using any method string or otherwise.

I beleive their now using a silver-epoxy just under the IHS on the core, and it will destroy the chip when you pry it off. This is true with the P4 EE's and I have a feeling the chage may have been made across the board as it would certainly increase thermal transfer.

You can, however; still lap the IHS to a mirror finish, and I HIGHLY recommend getting a kit from http://www.easypckits.com I got one and below is the result. That's my surfboard in the reflection on the left, and the right is focused on the chip. Almost as pretty as your avatar my friend;

[Marvin]

heheh, really almost.
So after reading this excellent topic, billa´s article and procooling topic, i came in some ideas:
1 The IHS is really concave(mine)
2 if you try to lap do it the right way.
3 Adjust your presuure clamp.
But how intel only assegure 15-20 mountings ?:confused:
sorry my english.
hey, off topic my under contrction water box.
soory my crappy english

[freecableguy]

I have had the IHS removed from my 2.4C for about 6 months now. I tried to get more people interested in this but no one seemed to car that I saw about a 7C drop in temps. BTW, DO NOT attempt to remove the IHS from ANY of the "Z" (m0 stepping) chips including the P4EE. Lapping the IHS to a mirrored copper finish is a good idea though.

[Liquid3D]

Hey don't feel bad, I didn't think peope cared either. But you must remembeter a 7C drop may not be worth the loss of the warranty, and if you don't have a heatsink mallueable mounting hardware forget it. For examle the standoff's on my SLK947, and SP-94 prvent me from using the P4 with IHS removed. You would have to sand down the stand-off, and place washers under the mounting screws, but then it's hard to get even tension. I actually ground down the screws because they tap out in the bottom of the standoff. That's carefully meas7ured by Thermalright, and just removing the IHS is enough to render the heatsing unusable. I tried it. I did write an article about it at Madshrimps here's the link;
http://www.madshrimps.be/?action=gethowto&howtoID=39

Your not alone in this. I too experienced a significant temperature drop. And the core is simply amazing. See below how it looks afdter you clean it. It has that deep dark blue/black color which I've heard is indicative of a quility core. And it just so happens I've had this 2.4C up to 3865MHz, and even higher, but not stable.

[st0nedpenguin]

Hmmm, I have been considering shaving a mm or so off my SP-94 standoffs, might be worth a try now just for a touch more pressure.

[Liquid3D]

Be careful my friend. I simply sanded measured, and sanded again. It's not easy, but I wouldn't take it to a grinding wheel or anything.

[Holst]

Originally posted by freecableguy
I have had the IHS removed from my 2.4C for about 6 months now. I tried to get more people interested in this but no one seemed to car that I saw about a 7C drop in temps. BTW, DO NOT attempt to remove the IHS from ANY of the "Z" (m0 stepping) chips including the P4EE. Lapping the IHS to a mirrored copper finish is a good idea though.

Why not remove it from a Z stepping chip ?

[lalPOOO]

Because Intel is bonding the heatspreader to the chip with some sort of silver epoxy. So far anyone who tried to remove it has ended up with a destroyed core.

[Spec3]

Here is a pic of my core after taking the IHS off. The reflection you see is a bottle of isopropyl alcohol.

http://computer.spec3.net/lapping/core.jpg

[feesh]

Beautiful :) I might have to do that when my 2.6c gets here today, provided it isn't a newer stepping.

[snow]

Originally posted by Spec3
Here is a pic of my core after taking the IHS off. The reflection you see is a bottle of isopropyl alcohol.

http://computer.spec3.net/lapping/core.jpg

well my 2.4c don't look as good as yours but it is running 3.4 ghz 1.55 volt at 30 C full load with a SLK900U and a zalman 50 cfm fan on it, don't think it would have so low temps with the IHS on it.

[Liquid3D]

Originally posted by Holst
Why not remove it from a Z stepping chip ? Holst, I'm shocked you wouldn't know that? :p: Now I'm scared.

I think I'll rip the IHS off my 3.0 using a hammer and big screwdriver. I mean why the Hell not "Self-Destruct Will Robinson...WARNING. WARNING..." (flailing arms in flexi-pvc tubing)

[st0nedpenguin]

Originally posted by snow
well my 2.4c don't look as good as yours but it is running 3.4 ghz 1.55 volt at 30 C full load with a SLK900U and a zalman 50 cfm fan on it, don't think it would have so low temps with the IHS on it.

I dunno, this chip I have will run 3.5GHz @ 1.6v, temps at 30c load, that's with the Tornado cranked down to 3500RPM.

The CPU is still in it's "virgin" state, IHS in place, non lapped.


Originally posted by Liquid3D
Be careful my friend. I simply sanded measured, and sanded again. It's not easy, but I wouldn't take it to a grinding wheel or anything.

Yeah, I just want to shave a tiny bit off, nothing insane, just to up the pressure a tad.

[kromosto]

i want to remove my ihs but not sure if it will touch 6000cu good anymore in my p4p800

[Liquid3D]

Whats the mounting hardware on your CU6000 kromosto? Maybe it can be done. Have a pic or link?

by the by Stoned P just place washers under the springs, and that will increase the pressure on the chip. That's all you have to do my freind?

[st0nedpenguin]

It's odd, but I tried all kinds of different mounting methods, I followed the instructions first, then tried it with only the plastic washers on the standoffs, then with washers on the springs, then with no washers at all, and a few other combinations.

I got the lowest temps using just one plastic washer on the standoffs, between the standoff itself and the motherboard mounting hole, I'm getting slightly higher temps at the moment, around 32 under load rather than 30c, but that's probably just because the heating is on now due to the weather.

I'll have a play about with it once I get the new motherboard through, I have an RMA number at last, so this one is going in the post tomorrow.

[kromosto]

this my cooler and in this link there is a information about installation i have to be sure that it makes good contact because i am using this rig at work and when i did sometinh wrong and my rig stops wprking my partners go angry lol :) they told me to stop overclocking but i cant :( :D:D

http://www.zalman.co.kr/english/product/cnps6500b-cu.htm

[kromosto]

http://www.zalman.co.kr/english/product/ZM-WB2.htm

also i got this water block but cant get my water cooler parts from one of my partners so havent tried it yet lol he is cooling his 2400+ with a home made water cooling system with a block to cool 9800pro block is also home made

[nailbomb]

Do the 1.8's have a thin layer of aluminum on top of copper? I want to know, so that way I can judge if i have gone far enough/too far ;)

Shaving off should NOT have an effect on the interface with my MCX478-V right? Since its clip mounted?

Thanks :)

[Marvin]

mine is a SL5WJ and still dificcult to remove the IHS. Lapped the IHS 2 time and not great success.
Nik, does your P4 is the same ?
cheers
:confused:

[berkut]

Is there any way to crack the glue on a M0 stepping cpu ?

As far as i know some epoxy glues (ati's for examle, nvidias also) dont like acetone, the intel / amd / nvidia glue used to mount IHS'es change in to some kind of goo when they have contact with acetone, maybe the epoxy glue would also act to acetone like this ?

[Liquid3D]

Trust me it's not worth destroying a $200 to $600 CPU for a maximum temp drop of 8C...

The time and effort, not to me tion the risk, would be better spent getting a Danger Den Maze-4 Powerkit for $199. You get ther excellent Maze-4 CPU block, Hydor L30 pump (320gph), Black Ice Xtreme radiator, clear plexi-reservoir, 1.5meters of 1/2" clear DD tubing, 120mm Sunon fan, and depending on how you set the kit up, a 20C drop in temps. I leave the radiator outside of the case on the cool floor, and leave all the components external of the kit. Usually my PC is under an open window, but in winter my temps drop to 8C, so peoapl are actually in disbelief.

To satisfy those who wouldn't crack a window in Winter, place their PC just beneath it, or leave the components outside of the case, I closed the windows, and let the heat rise in my room. Below are my current temps, running a 2.4C at 280FSB (1:1) at only 22C;

[nailbomb]

Liquid,

Can you show us a pic of the guts of your rig? What case are you using?

I'm real curious about the Maze-4 rig :)

[Liquid3D]

Sorry about the late reply. Here's the system, the top left photo is the primary reason why the temps are sp low. This window can remain open even in Winter. I live on the 3rd floor of a 100yr-old Stone building a few hundred meters from Newport Harbor. The cool water air, keeps the building cold, th stone retains this cold. This corner room has a small window pewrpendicular to the window pictured, so there's a constant flow between them. With the components on the foor the way they are, everything is disipating it's heat into the stone-floor/walls etc. Especially having the Radiator at the lowest point, it's as cold as the floor, then facotr in "wind-chill" from the fans, and you have 11C on the mobo, but only 10C on the CPU (although where I'm stting on the bed, is about 20C or around 68F).

One reason for this is the RBX has the brass plate installed, and it's dissipating/absorbing the cool temps as well. I have the Danger Den Z-chipset block fed first, and the RBX second, although it wont remain this way, because the block has many nozzle attachements and I have the Hydor Seltz L45 now to install with it. The added pressure is almost going top create a vaporizing effect. I just need a ;large dual radiator for the larger pump. I'm simply ecstatic with the Danger Den hardware, I love it! I'm not sure where the diode measures temps on the Asus, I doubt it's on the chipset (although they should be) but if it were it would certainly explain why many days the CPU is running cooler then the mobo. If that makes sense.

[nailbomb]

Nice :) Thanks for taking the time to make the pics. If I go with water I think it will certainly be Danger Den. I hear nothing but good things.

I got my CPU lapped flat down to the copper layer and didn't really see much of a difference in temps (on air). Kinda disappointed, but o well.

[Liquid3D]

By the way Nailbomb (crazy username :eek: ) welcome to the Xtreme I think if you stick around you'll find there's some really great people here!

What TIM are you using? How did you apply it? Laping the IHS is important, but if you've only lapped the IHS, and left the Heatsink replete with the micro-grooves exemplified below, your only half-way. The TIM is the medium of transfer for everybit of energy (heat) which is to be conducted away from the CPU cooling it.

Another lesson I learned from Nevin himself (owner Arctic Silver) is that, spreading the paste, incorperates air into the forumula. This incurs the same effect in the TIM, were trying to ameliorate through lapping BOTH surfaces. Simply place a small dollup (a "drop" equal to a hyper-thin layer which would cover the IHS surface) and then allow the natural clamp-down pressure to spread the TIM. Remember, the actual core size under the P4 IHS is only about 40% of the entire surface (as shown below). If using a mounting system which uses the boards mounting holes, tighten the bolts perpendicularly so that pressure is distributed evenly during the installation (tightening down process). If using the retention brackets or Ziff-tab clamps, you can center attach the heatsink, and then apply even pressure so the paste displaces evenly underneath. After a week, remove the fan, and shift the heqatsink slightly, this will actually expidite the "settling" and break-in process of the TIM. I find the best TIM's on the market are Silver based, period! AS5 is the best I've ever used, it's density makes it the ideal interfase.

[kromosto]

ok ok i am starting to remove my ihs from 2.6c if it works i will tell the results if not work then you can read the whole story from kunaaks thread :)

[Liquid3D]

Be honest I wouldn't suggest it, my point above was to lap BOTH surface meaning heatsink and IHS, not remove it. I was just showing how small the core is underneath, so when applying thermal paste one can understand spreading it to the edges, isn't necessarily as critical, as the smoothness of the mating surfaces, and the type of thermal paste.

If your going to do it, I'd recommend this method below, and if you feel the center being umovable, it's silver solderd to the core, so leave it (that's nylon thread in the picture NOT wire, it's grey from the epoxy) You simply place the processor into the motherboard (to clamp it down) and "saw" away with the string. It takes time, but using any other object can expose the interconnect trace wires, just under the chip's surface. And especially at the edges;

[kromosto]

and yes i got it off but problem with my zalman 6500cu so i take zalman water block and it worked nice with water cooling and ihs off i got nearly 10 temp drop :)

[nailbomb]

Hey Liquid, thanks for the reply :) I picked my name because I used to IRC a lot, and wanted something unique so that I would allways have the same nick on Efnet. Nailbomb is the name of a band from Brazil, so I figured nobody else would ever take it. Plus I like the band :)

To answer your questions:

Right now I'm running a new MCX 478-V. Nice heatsink, very nice finish on the bottom of the sink. The V version uses the stock retention bracket, and a pair of clips to hold it in place.

TIM: AS5 applied per your directions. I first learned how to do that when I started using Ceramique. I don't think a lot of people are aware and probably still just spread it. When I first showed my friend how to do it, he was skeptical. That is, until he saw his temp drop quite a bit.

I DO think that maybe the MCX478 could stand a bit of sanding thou, so maybe I'll try that and see what happens.

Unless you have any other ideas?

Thanks :)

Oh, just to clarify for me:


After a week, remove the fan, and shift the heqatsink slightly, this will actually expidite the "settling" and break-in process of the TIM.

You're not saying to remove the whole sink are you? Just the the fan right (easy to do on the 478-V) and then kind of wiggle the sink a couple degrees in both directions? I had seen that as well on AS's site. The 478 has some bad mama jama clamps on it thou, its pretty damn tight. Tighter than my Spark 7 was. With that sink I could wiggle it a couple degrees each direction.

So, pull the SINK after a week, or just the fan?

Thanks for the warm welcome too dude :)

[nailbomb]

Liquid, while I'm thinking about it:

Do your 'burn in' your paste by running a max load with something like Prime95, or do you feel its even beneficial to do so?

[Liquid3D]

About the "settling" procedure, Yes, simply remove the fan (actualy you don't even have to remove the fan so long as you grab the heatsink, and wiggle, it from side to side just a few mm. You want to be careful if using the retention ziff-tabs, perhaps avoiding this all together. In fact, if your motherboard has mounting holes, I'd immediately remove the ziff-tab retention mechanism. Ironically, what makes the 478 so effective at dissipating heat, it's weight; is countering that of the ziff-tab clamps. I'd say your only reaping 75% of it's full potential, due to the weight significanty reducing mounting pressure.

In so far as buring in the paste. I do not, but many do, and Nevin did say that turning on, and off power to the CPU, does help expedite the "break in" period. He didn't mention Prime-95, but my philosophy is not to stress the CPU if it's not being stressed under normal circunstances. If you want to stress it, just use it.

You probably have what is arguably, the best air-cooler on the market, Thermalright being the only other brand which in it's class. What is critical with these massive copper heatsinks is air-flow. It may be noisy, but if your spending the money for the best, I'd only match it with the best fan. For 92mm that would be the Vantec Tornado pushing 119CFM. I have this fan, and it's excellent. I imagine it would be quite a match with your heatsink. Finally, while the Swiftech does have an excellent surface, it's one of the manufacturers which use a "finish", and I beleive this protective layer impedes it's conductive capabilties. I would definately lap it, and bring in down to some really fine grits. This company www.easypckits.com (or person) offfers an excellent kit, and it's how I attained the finish below on my 3.0C;

[nailbomb]

If your motherboard has mounting holes, I'd immediately replace the ziff-tab retention mounts.

Am I reading you right: replace the whole P4 retention bracket assembly? I didn't realize that replacements were even available.

Tornado is allready on the way :)

Just to clarify: the 478-V fits in the stock retention frame, and the old 478 uses the thru the board mounting, or
at least I think it does. MCX4000 certainly does.

http://www.swiftnets.com/assets/images/products/mcx478-V/MCX478-V-installed-300x252.jpg

I might try to hit the bottom of the sink, but I don't know. I feel the flu coming on I think :( Maybe later in the week ;)

[Liquid3D]

Well I went to the Swiftech site, and don't see where they've provided for a through-board mounting system for that cooler? I'm perplexed because of it's weight? Perhaps they believe the plastic push pins that secure the Intel bracket are sufficient to hold the weight, which I'm sure they are, but I wonder how much pressure can be excerted, with the unit, and large fan "hanging" there, gravity pulling the unit and reducing contact pressure?

[nailbomb]

Well, I think you raised a very good question. However if you look at their C/W numbers, they seem very good especially with the Tornado.

And, I've got to tell you, the clips themselves are BADASS. It takes a LOT of force to get them clamped down, so that being the case I'm not sure mounting pressure is much of an issue.

Maybe you could hit them up for a review sample, and see what you think?

:D

[Liquid3D]

I've tried, Swiftech only sends samples to the largest sites. I can get anything from Thermalright, as their much better in that respect. They give small sites a chance. Funny thing is Swiftech sent a sample to [M], and I submit articles to Madshrimps?

But I'm going to write them anyway. Thing is, when I last wrote them I requested a Water-cooling kit, which is somewhat costly.

[STEvil]

I think you should give this a try if you can find a pair of 60mm fans, Liquid.

My numbers have been great, but i've been hesitant to state anything final incase i've pulled a boner..

http://www.wcroller.com/los/linkpics/slk10k.jpg

Right now I am running 2444mhz 1.968v with 21c ambient (measured behind an intake fan with ratshack indoor/outdoor thermometer). SETI@home (setihide and cmd line client) gives me a temp of 37c. Loading the CPU down with a game or any of the burn-in wizards I can find (those 64k demos work great for this, too) only maxes me out at 40c.

Idle temps for the few times I have been idle long enough for them to drop significantly were about 26-28c.


Thermalright should give this a try too, and sell SLK-10K's.. i'll let them as long as they send a few to XS (and maybe lanageddon.com) for give-aways :D

[Liquid3D]

Is that the cooler I gave you, that got there in like five days? Email me, and I'll send the photo to my contact at Thermalright, you never know my freind.

[nailbomb]

Well, I have to say I'm really perplexed:

I got an 80mm Tornado in today for the 478-V. Why the 80mm? If you look at Swiftech's site, the 80mm actually has a better C/w rating than the 92. My guess is the larger hub on the 92mm gives a bigger deadspot in the center of the heatsink.

So, WHAT am I perplexed about? I'm perplexed about how the Tornado, running at full blast, is no better tempwise than the 39CFM fan that I was running before.

How could this be? :(

[Liquid3D]

Wow this is a puzzler. I imagine it's your case ventillation, or lack there-of. Although it's seems your knowledgeable enough that you alrteady have a well-ventilated case. I'd recommend removing the side-panel if it's in-place, even if you beleive your case is well ventilated. leave the side-panel off for a good long time (3 hours minumum) for temp equillibrium. I gaurantee you'll see a minimum 10C temp drop.

And as I've said before, I'll bet my infamous Silver Solder Dropping Collection on it! Which by the way has become a valuable commoditity in it's own artistic right.

[Marvin]

I did it. After a´pain i removed the shim

[Liquid3D]

Good job, what method did you use? Did you try the string method? Most don't have the patience, and to be honest I cheated a little, by taking a couple of prys at it, with the ole Xacto knife. hehe

Now let's see the core cleaned up.

by the by, DON'T LAP the CORE!

[Marvin]

Originally posted by Liquid3D
Good job, what method did you use? Did you try the string method? Most don't have the patience, and to be honest I cheated a little, by taking a couple of prys at it, with the ole Xacto knife. hehe

Now let's see the core cleaned up.

by the by, DON'T LAP the CORE!
i have used a dead mobo(socket)a steel blade and lot of patience

[Marvin]

Originally posted by Liquid3D
Whats the core look like? Is it blackish, or what?
sorry, hee is the core a bit cleaner

[Liquid3D]

Yes that's look like a "spicey-a-meata-ball". Should overclock like a madman.

But I must tell you that, you may using too much thermal paste. This is an awful lot my friend. With all due respect.

By the by, be careful mounting heatsinks with through board mounting hardware, they are designed to stand-off at a certain height. Enough to apply adequate pressure on the Pentium 4, but removing that IHS, creates a gap. I tried mounting my SP-94, and thank the Lord, the CPU shut's down before frying, because contact between CPU and Heatsink was non-existant. The paste remained un-touched.

You eitther have to sand down the stand-off's, whihc means you have to sand the bottom of the mounting screws as well, because they "tap-out" in the bottom of the standoffs.

[Marvin]

Originally posted by Liquid3D
Yes that's look like a "spicey-a-meata-ball". Should overclock like a madman.

But I must tell you that, you may using too much thermal paste. This is an awful lot my friend. With all due respect. yes, i know that. Also i was not the first owner of this chip. i bought from a friend.
Thanks;)

[Liquid3D]

As I said no offense intended. Just looking out for one another.

[nailbomb]

I spoke too soon. Temps have dropped 3C with the 80mm Tornado, running it at 7V on my Rheobus. Interestingly enough thou, I don't see much difference thou running the fan at full nuts 12V.

One thing I did have to change was my exhaust fan. Bumped it up to a 39CFM fan. So, I DO think its somewhat of an airflow issue. Its a Kingwin 424 case. Spots forf 2 80 intakes, who's airflow will get somewhat impeded by the hard drives, 1 80mm exhaust (now 39CFM) and the 80mm blowhole I cut.

Intakes: Pana H1's at 39CFM
Exhaust: Vantec Spectrum at 39CFM
Top: Sunbeam UVCCFL at 32CFM?

Plus the power supply fans, which are thermally controller, ie SilentPure Power 480. Good PSU BTW, voltages look rock steady measured on a digital multimeter.

Its rumored that some early batches of the Asus P4PE might have problems accurately reading temps, so I have that to contend with as well. Batches Made prior to January 2003.

All my wires are sleeved, and round cables for everything else. It's nice and clean, and the sleeving really isn't hard to do yourself, its just time consuming the first time you do it. :)

[Marvin]

i lapped the ihs after removing it just to see. God, tis ihs is very concave.

[Marvin]

Strange, lapped my other 2.8C. I donk now if the IHS was more flat than this one or i did i bad job. I didnt apply to much pressure on the cpu, neither put the finger in the middle of the cpu, so the temps ate the same as unlapped.:rolleyes:

[nailbomb]

Mine was partially convex, partially concave ;)

Draw an X on it with a permanent marker, and then when you sand you can see the low spots. Low spots are where you have marker left.

[Marvin]

second attempt. no difference in temps. sanded with 400, 600,1000 paper.
this is before relapping

[Marvin]

After the lapping. Almost the same.:mad:

[Liquid3D]

For the love of Pete, fella's all I'm asking is that you TRY removing your side-panels for several hours and watch the temps drop. I bet you wont put that side-panel back on.

Today it's warm, but recently I've had the windows closed, because of a review I'm writing on the RBX block, I need to find temps "normal" people would experience. Although what I don't understand is I'm not so special. How hard could it be to find a window, put your PC beneath it, and open the window a crack. It's Dec in R.I. on the shoreline, and it's almost 60F outside, so I'm stting here in a Tee-shirt, so I'm not freezing. These are the temps I'm getting at the moment using the RBX block, with H20 components outside of the case.

[Marvin]

i will do it. Here in Rio is very hot (40C).
Liquid3D, what did you think of my lapping ?
:(

[Stang_Man]

Originally posted by Marvin
i will do it. Here in Rio is very hot (40C).
Liquid3D, what did you think of my lapping ?
:(

not enough

and why'd you take the shim off the processor before lapping it?

[Liquid3D]

Shim?

Your lapping looks pretty good, but you want to go a few more finer grits down, and in reversing circular motions. Remember to place the sand-paper on a piece of glass, or a mirror, to ensure the surface is flat. Carefully spreading your weight (using a cloth or your fingers but don't bend the pins) across the pins, go in circular, and/or figure eight (8) motions, keeeping the IHS flat on the surface. If you can't get the figure eight thing down, just go in circles, but be sure to reverse the direction. Keep cleaning the processor with alcohol, and tapping the sand-paper to get the powdered copper off. It will go much better if you keep cleaning it.

I'm bummed because I have one of the famous SL6WT Malay 2.8C's L323A832, and it's crapping out above 250FSB, or 3585MHz. Here's my OC temps now;

[nailbomb]

For the love of Pete, fella's all I'm asking is that you TRY removing your side-panels for several hours and watch the temps drop.

LOL! I used to run my Hx08 with the side panel off all the time, that was before I was OC'ing and monitoring temps. It was just a pain in the ass to take all the screws out to get inside the case to work on it.

Its fairly cool in my room here. Winter outside, and window cracked. I don't see much a difference with the panel off on my Kingwin, but I think I have good airflow anyway so. The other issue for me thou, is I don't think the P4PE temp sensors are accurate. Nobody really seems to know WHERE the mobo sensor is actually located for one thing, at least on the forums I have asked on. Several people on abxzone.com have reported goofy temps with boards shipped before January 2003. I got mine in Dec. 02.

In regards to lapping: Liquids suggestion to use a figure 8 motion is a great one. Doing the figure 8's really seems to make it easier to keep things FLAT, especially if you do it in conjunction with my Sharpie trick. Draw an X on the IHS, and then when you send you can see your high and low spots :) Also, you should probably keep in mind, the more mirror like your finish, the LESS TIM you want to use, because there theoretically less voids to fill.

[Marvin]

thank you nailbomb, i did that lapped in figure 8 motion. I didnt appy to much force while lapping, so i am afraid of dead spots.

[nailbomb]

Hrm, you must not have sanded very much Marvin, because I can still see the X there. The way you know its flat is when all the X is GONE.

[Marvin]

sorry the mistake Nail, i draw the X pattern, as you said, before lapping. the final is this.

[DjTonic]

Nice tweak. Do you remove IHS before lapping? I'm worry about this hole in the Heatspreader whats got happents if came dirts in?

[nailbomb]

The hole is just a leftover from the manufacturing process. When they epoxy the IHS to the chip wafer there is a lot of heat generated. The hole is there to prevent the IHS from blowing off the top of the assembly.

So, I wouldn't be too awfully worried about getting stuff in the hole, especially if you follow Liquid's suggestion of cleaning the IHS with alchohol as you do your sanding.

[Liquid3D]

I got an excellent finish using a kit I got from www.EasyPCKits.com they sent me a small piece of glass, and all the paper you would ever need. At first I didn't even know what the different colored paper was for, it was so fine, it just felt like paper hehe. Here's (http://www.easypckits.com/images/007.jpg) a picutre, and on the sheet you can read they supply 400 - 1000, and then 40-micron - 15-micron paper. These are the results using fine papers;

[nailbomb]

Did 2 things today:

1. bent my Swiftech clips to get a little more mounting pressure.
2. went up to 1500 grit on lapping. Had stopped at 1000 previously

Result: -2C :)

[Liquid3D]

Ever consider Watercooling my friend? You simply can't it for the money; http://www.dangerden.com/mall/Kits/kits_retail.asp

DangerDen Maze-4 Intro Kit is only $174, and is awsomw, but for $199 I'd get the Powerkit, because it includes a slightly more powerful pump, and better radiator.

The kits come with MAZE-4 CPU block, Hydor L30 (320Gph) pump, Black Ice Extreme Radiator, 120mm/92CFM Sunon fan, 1.3-Meters 1/2" clear tubing, Water Wetter, clamps, AS3. It was the best move I've ever made in my PC experience, and it's so versatile. I now use the D.D. Z-chipset (http://www.dangerden.com/mall/blocks/zchipset.asp), and have their GPU block as well. Just waiting for a larger Radiator. You can experience as much as a -25C overall temp drop switching to H20. The screenshots below were taken the other nite, when I had my window open about 6", and yes it was cold in my room, but not freezing. I did put on a sweater. Obviously no one would do this all the time, but these temps allowed me to do some OCing.

[nailbomb]

Yes, I have been looking at it. A LOT actually :) For some 2.4c action.

Hrm, biopsychiatry.com eh?

[Liquid3D]

Actually it's more like Neurophilosophy. Neuroscience encapsulates several disciplines, at it's base is Cognitive Science, AI (Artificial Intelligence), Physiopsychology, Neurochemistry, and Linguistics. Neurophilosophy delves into topic's such as the "Nature" of consciousness. Are our thoughts purely physiochemical brain states. Such that the color green, or the thought of green, can be defined as, such and such an exact neurochemical state, at such and such a time.

Semantics, a sub-discipline of Linguistics talks about things such as "Possible Worlds" as it applies to "Truth Values." Mathematical methods in linguistics are integral to AI, and programs such as Parallel Distributive Processing, are excellent analogies for the Brain's learning process. Funny thing is philosophers have been arguing such ideas even before PLato's Allegory of the Cave (http://plato.evansville.edu/texts/jowett/republic29.htm). Reading THIS (http://hem.bredband.net/arenamontanus/Mage/neuro.html) will most likely lead you to understand why I find this science fascinating.

[Gef]

I just tried lapping my 2.6C yesterday, to see if it would make any different. I used 1200 paper all the time. I don't think my temps were any better than before...i only use sucky Tsunami cooler with some new AS5 that i just got. I just think that my temps are higher now than it were with my shinetsu . In prime it got up to max 54C, with my AS5 and lapping the CPU it whent up to 58C in prime! Maybe the temps are so high now because of the goop is recently put on. I also used as little goop as i could, so you barely could see the core of the CPU.

Here is the result:

[Gef]

NR2.

[Alexandrus]

Damn, Liquid3D, that's some fine lapping. I'll post some pics of my results soon, I took some pics but haven't downloaded them to the PC yet. I used 500-600-1200-2000 grit, took me about a week to do the lapping(my HDD was dead and I had a defective mobo, so for a whole week my only contact with my PC was when I lapped the CPU :D).

[Liquid3D]

This is how cold my Danger Den water-cooler RBX and Z-chipset NB blocks got, with the radiator outside of the case, and the window cracked a tad. This is appraoching phase-change cold and Yes I did see my breath that morning hehe;

[dioSRL]

temps didnt change much. CNPS7000 Al-Cu, 2.4B SL6RZ, AS5. I posted alexandrus picture, I've had it downloaded to my PC ;)

I also did my last CPU. 2.0A, AS3, SLK900U, SCF2@1600RPM. 2000grit, both. I experienced higher temps, ~2-3C.. But after a week or so, I experienced a temp drop of ~2C of the original.. I suppose it's the low RPM of the fan. Picture will follow(sorry but a bit blurry.. about 1 year ago and shacking hands at 5AM)

[dioSRL]

and my old cpu&current heatsink... I could read with ease the AS3 notes in my CPU..baad picture :mad:

[Liquid3D]

Nice lapping job there! It's possible if you didn't lap the heatsink (or CPU) on a piece of glass you actually created an un-even surface. This would explain why temps went up. Also don't spread the paste, simply place a small dollup in the center of the CPU, and mount the heatsink.

The manner in which you lap is critical as well. If you don't continually clean away the copper particles, they tend to become re-ground into the surface, making deeper pits, so all the work ends up being defeated.

Nevin the owner of Arctic Silver told me that moving the heatsink "shift it just a bit" actually seats, and sets the paste. After about a week of applying new TIM on a CPU, you should give the heatsink a slight shift. being careful not to loosen the fastners, but enough to cause a shift in the paste, in which any air-gaps are eliminated. (he's also the person who taught me how to apply paste)

[dioSRL]

OK, thanks a lot for the advices. Actually I don't clean the surface when lapping, I like the way the sandpaper became more smooth. At the end, for the final touch, a bit of soap with water is used, I think it increases the smoothness.


if you didn't lap the heatsink (or CPU) on a piece of glass
I've chosen myself the piece of glass, it's near the SLK900 heatsink. I will give it a try with AS5. I always spread very well any kind of thermal paste, maybe it's time for a new approach. Happy lapping to everybody :)

[Marvin]

just put a small dab os paste the center of the IHS.your lapping looks fine.:D

[ctgilles]

Anyone using a special thing to hold on to their chips while lapping? I found the Xeons have more pins and they're truly a hell to hold on to when lapping :D

[Holst]

Id get an old socket off a board and use that If you have one avalable...

Otherwise pushing the chip into some stiff foam will work pretty good.

[ctgilles]

The foam thing is a good idea, I'll try it :)

[Liquid3D]

There's absolutely no way to discern if the surface was lapped flat by looking at it. I agree it's "looks" well done, however; if temps went up after lapping, I'd most definately take a piece of glass / place the sand-paper on it / distributing pressure evenly over the pins with your fingers (so as not to bend any) / press down firmly / go in figure eight motions, or circles / periodically reversing / wipe clean every few seconds / always use clean sand-paper / work from 800-grit to 10-micron.

The reason it becomes "smoother" at the end, is because the particles become trapped in the paper. This is not a good thing, because your simply using the material lapping is intended to remove, and rubbing it back into the surface. This is what can cause uneven pit's and microscopic grooves. Remember the intent in lapping is to create as flat a surface as possible. If both surfaces are perfectly flat, then they will mate together perfectly. The reason we use thermal paste, is because it's almost impossible to create two equally, and perfectly flat surfaces. SO we fill the microscopic pits, and groooves with the Thermal INTERPHASE Material (TIM) the key term being interphase.

[Alexandrus]

Too late for that Liquid3D, I did the lapping the wrong way it seems and I am too lazy to start it all over again, I have to clean the CPU, do the lapping, clean the heatsink, lapp it again, and it takes too much time. Not that I am a busy man, but I am an extremely lazy one. I should be called XLD(eXtremely Lazy Dude) :D

[dioSRL]

yeah.. I did half wrong.. But since, I did remove so many times the heatsink.. changed so many times cpu's.. it's already gone that lapping. So maybe.. sometimes when the sun will rise, I will do it perfectly. But for that I'm gonna need that lapping kit.

[Liquid3D]

The best lapping kit I've seen, and used. Included is: a piece of glass for flat backing surface, 400 to 10-micron grit sand-paper sheets, and TIM of your choice (depending on kit). www.EasyPCKits.com :D

[dioSRL]

yeah, that's what I'm talking about. I saw you posted the link a couple of pages earlier. Nice one ;) Also, could you please post a scanned image of "Suggestions and Tips" from that lapping kit? I am curious what they are saying in it. Thanks.

[Liquid3D]

I don't have a scanner, however; I'll reprint a few pertinent items from the included instructions;

Your Premeium Lapping Kit contains a piece of glass to be used as a flat backing surface...sheets of sandpaper consisting of the following grits in order: 400, 600, 800, 1000, 40-micron, 25-micron, 20-micron, 15-micron, 10-micron (color coded)...Superfine particles of aluminum or copper will be very dark and hard to remove and clean up. Keep plenty of paper towels on hand...put a micro-bevel on the the outer edges of the surface to be lapped...The trick is not too remove too much metal...but a slight bevel on the outer edge will prevent snagging and tearing of the sandpaper...start with coarser grits, using lower grits only if you have to as they remove too much surface....wrap the sandpaper around the glass edges to prevent it from slipping....use a few drops of water for lubrication (mineral oil or 70% ispropyl alcohol will work as well)...move in figure eights or circular motions, reversing direction every few seconds...Rinse (or clean with alcohol damped cloth) your surface (heatsink or IHS, etc) and sandpaper periodically when changing grit, even while working on the same grit. A few stray grains of larger grit then your currently using will really scratch and be noticable (keep cleaning it!)....Go slow and easy, don't bear down too hard...let the sandpaper do the work... Those quotes in parenthesis are MY own.

[Alexandrus]

I wish I could get my hands on one of those lapping kits, but it's not easy, since I live in Eastern Europe. I think I'll redo the lapping using 600-1200-2000 grit, do it the right way this time, and see if anything good happens, like temp drops ;)

[Liquid3D]

They ship Internationally all the time, and I beleive Eastern Europe would be no problem. In fact I thought they only shipped to USA, and got an email yelling at me (just kidding). It's as little as $3 for shiping just about anywhere in the world.

[Alexandrus]

Awesome :)
Thanks for letting me know.

[Liquid3D]

I just finished an in-depth review of their Premium Lapping Kit at my site. If you have time, your opinion would be appreciated;
www.ksBrainstorms.com

Temps were attained with Danger Den RBX & Z-chipset blocks, all H20 hardware (BlackIceX, 2x120mm/92CFM Sunons, Hydor-L30) sit external of the case below a window I open periodically. The low temps let me run up to 3907MHz stable. Only 93MHz from my H20 goal...:eek:

[nailbomb]

Nice results :)

How big is the piece of glass they send? I was thinking about hitting the local glass shop and see if they could hook me up with a 12x12 piece of scrap plate, and have them braze the edges so it isn't sharp anymore.

[nailbomb]

This is just NUTS:

Newegg has some Intel Stock heatsinks for 5 bucks, so I decided to order one. It's the same unit that ships with the 3.06HT, ie the nice looking copper base jobby with thin Aluminum fins.

Anyway, I thought I'd throw it in my rig to see how good it performs in comparison to my Swiftech 478-V which was 10x as expensive.

So, what are the results?

The 5 dollar stock copper Intel sinkgives me the SAME temps as the 478-V. That's no §§§§ either. 35c idle and 41 load running
Folding.

Wow, I'm sure glad I spent all that cash on the Swifty ;)

I lapped the Intel sink before installing it.

[Liquid3D]

The glass EasyPCkits provides is 4" x 4", and I wouldn't recommend buying a 12 x 12 because it will have a greater chance of cracking, shattereing. The size they send is ideal, because it allows easy wrapping of the lapping-paper, and keeps your figure-eights, or circles in small concentrations. This is better. I'd get their kit, because it's not easy finding 10-micron paper and beleive it or not that's the key. I NEVER thought I'd use that paper, in fact I didn't even know what the Hell it was at first. It was so smooth. But that's how I got those perfectly smooth, flat surfaces.

In so for as the heatsink. If you go to Thermal Integration's (http://www.thermal-integration.com/) site, you'll see their heatsink's are specifically used by Intel to be included with the 3.06, and 3.2 CPU's. That's a compliment to T.I.T.I. I used to recommend the Dr Thermal to people, because the design is excellent. You only need contact directly over the core of any CPU. Especially the P4 given it's IHS is 70% larger then the core beneath it. My Danger Den RBX actually allows edges of the P4 IHS to be exposed, and this is the best heatsink/block I've ever used. The same is true with Thermal Integrations. Their concentrated foot-print, and the fins being cut to the base, allows the air from the fan to cool right down to the chip itself. The reason Swiftech, and Thermalright design's work, is because they use the concept of MAss (and in Thermalright's case a small foot-print, and unrestricted air-flow to the base), however; only a high powered fan 80mm/84CFM or 92mm/119CFM Vantec Tornado's really cools that copper-mass enough to extact the heat from the CPU. Remember two basic laws explain heatsink conduction where MAss is concerned; Fourier's Law (http://www.efunda.com/formulae/heat_transfer/conduction/overview_cond.cfm) and Zeroth Law of Thermodynamics (http://hyperphysics.phy-astr.gsu.edu/hbase/thermo/thereq.html#c2).

[nailbomb]

Welp, I'm back. Just put the Swifty back in.

I 'think' I can account for my surprised reaction to how the Intel sink performed.

I just got a bottle of Akasa TIM-Clean today. So, pulled the Swifty off, cleaned the CPU with the Akasa, and the Intel sink and installed. Temp's the same as the Swifty.

Pulled the Intel, cleaned the Swifty with the TIM-Clean, and the CPU. Reinstalled the Swifty, and now my Swifty temps are -2C. Idle 33C load 39C.

I gotta say, while I can't for SURE attribute the change to the TIM-Clean, the stuff is just flat out AWESOME at cleaning heatsinks, and cores.

I took an old Slot1 PII, that had bunch of caked on goo on it, hit it with the TIM clean, and it just peeled it right off like butter. It looks like its brand new now.

I got the Akasa stuff from Sidewinder Computers for those that might be interested. There is only one other retailer right now in the USA that has the Akasa, and they charge (I feel anyway) an exhorbitant amount for shipping.

Liquid, I allways enjoy your insights, thanks :)

[ctgilles]

I'll be getting results when the mobo revives :D

[Alexandrus]

For the past week I have been experimenting with my rig.
I lapped the CPU IHS(I have a 2.4B SL6RZ) long time ago, and I perfected it with 2000 grit about 2 weeks ago. I also lapped the Zalman CNPS7000-AlCu the same time, 2 weeks ago. Both are mirror smooth, and almost perfectly flat(I cannot say perfectly flat because I would lie, nothing is perfect). Did the same to the NB heatsink on my 4PCA3+ rev. 1.2, which is now smooth and shiny :)
Anyway, before last weekend, I didn't use ANY TIM between the CPU and cooler or between the NB and NB heatsink, and my temps were something like 49C idle for the CPU and 39C idle for the system temp(which I guess is the NB temp). Full load temps were above 60C on the CPU and around 42C for the NB.
I put AS5 on the NB and on the CPU, and the temps ... well, raised. I noticed that because of the shim, the NB heatsink was making proper contact with the NB core, so I removed the shim and replaced it, replaced AS5 with Ceramique for the NB, and yesterday, after about a day of gaming(SW KOTOT), the temps were : 57C CPU and 51C system temp.
My case is Antec PerformancePlus 1080AMG, and I have absolutely no fans installed on it, so no airflow, I understand that. I also have a GFFX5900 Ultra which get pretty hot and is very close to the NB, but still, these temps look a bit high to me.
What do you guys think ?

[nailbomb]

nd I have absolutely no fans installed on it, so no airflow

Houston, we have a problem :slobber:

You need to get some airflow going man. I would recommend getting either some PanaFlo H1's or M1's to get some air moving thru your case. www.sidewindercomputers.com has the new model Pana's that support RPM monitoring if you want/need that. They are good fans, and don't put out as much noise as other fans that crank out the same CFM's.

I would also recommend that your amount of exhaust CFM's is greater to or equal than the amount of intake CFM's, thus negative pressure in your case.

Maybe Liquid has some other thoughts, he's a sharp dude :)

[dioSRL]

yeah I also need 6x80mm and 2x92mm Panaflo for my case/heatsink/psu .. that's going to be ~$200 fans+shipping+vat and whateva' . It's hard to be at most 1/4 xtreme(I would like so many things.. like a bowman modded prommy :slobber: ) in Romania ;)

[Stang_Man]

nailbomb, where'd you get the P4 retail heatsink for 5 bucks?

i searched up and down newegg's site and i can't find it :(

[SPQQKY]

After lapping my IHC and my SLK-947U and switching from OCZ Ultra thermal compound to Geil's copper and silver mixture, I got zero difference in temperatures. Oh well, nothing ventured, nothing gained.

[lalPOOO]

Originally posted by Alexandrus
For the past week I have been experimenting with my rig.
I lapped the CPU IHS(I have a 2.4B SL6RZ) long time ago, and I perfected it with 2000 grit about 2 weeks ago. I also lapped the Zalman CNPS7000-AlCu the same time, 2 weeks ago. Both are mirror smooth, and almost perfectly flat(I cannot say perfectly flat because I would lie, nothing is perfect). Did the same to the NB heatsink on my 4PCA3+ rev. 1.2, which is now smooth and shiny :)
Anyway, before last weekend, I didn't use ANY TIM between the CPU and cooler or between the NB and NB heatsink, and my temps were something like 49C idle for the CPU and 39C idle for the system temp(which I guess is the NB temp). Full load temps were above 60C on the CPU and around 42C for the NB.
I put AS5 on the NB and on the CPU, and the temps ... well, raised. I noticed that because of the shim, the NB heatsink was making proper contact with the NB core, so I removed the shim and replaced it, replaced AS5 with Ceramique for the NB, and yesterday, after about a day of gaming(SW KOTOT), the temps were : 57C CPU and 51C system temp.
My case is Antec PerformancePlus 1080AMG, and I have absolutely no fans installed on it, so no airflow, I understand that. I also have a GFFX5900 Ultra which get pretty hot and is very close to the NB, but still, these temps look a bit high to me.
What do you guys think ?

Sounds pretty odd. Perhaps too much paste? Its also possible that the paste didn't get to burn in, but that'd only be a 1-3C difference.

[Slickthellama]

Hm... Ive never gotten this whole lapping the IHS idea. I mean, the max temp gain I have seen is like mayyyyybeee 4C. Even then, is it really worth voiding your warranty? I recently had a 2.6C die on me at stock volts. Boy was I glad I hadn't lapped. I would have to advise against lapping. Just seems like the risks outweigh the rewards.

[dioSRL]

what time did u use that C? It was faulty for sure.. very strange.

[Slickthellama]

what do you mean by "what time"? Thanks.

[dioSRL]

oh sorry for how long :)

[Slickthellama]

no problem, im not here to argue symantics. But i was confused ;) . I used it for about a month. It just died one day. Everything at stock. The weird thing is, get this, one day, my AOpen AX4SPE refuses to boot. OK so I figure, dead mobo. I replace it with a 4PCA3+. Refuses to boot. So I try some new RAM and it boots. But get this. In my BRAND NEW 4PCA3+ it will not even POST when the RAM is in dual channel mode. So I get into BIOS on single channel. Temps are 54C. At stock everything. Im like... OK.............................................. So i shut it off. Remount and go in again. 53C........ so im like WTF? So then I figure, oh it must be a faulty mobo sensor. So I put my external temp monitor on the CPU. Nope it reads 55C. And I know it is accurate. So im totally stumped. It has to be a dead CPU. BTW this is with stock cooler. And I was able to get into windows but the temps were 89C then it immediatly locked up. The board isnt overvolting either because I checked the volts and they were at 1.488Vcore 2.5V ram.

[Liquid3D]

I agree it does depend on the situation. My 2.4C was OEM, so after 90-days it didn't matter. My 2.8C is retail, so I haven't lapped it yet. My 3.0C is lapped, and OEM, since I've removed the IHS from my 2.4C, and am now removing the IHS from my 3.0C. Water-cooling blocks have minimal, and infinately malleable clamp-down pressure therefore, it's safer to do these things, because you don't have worry about a pound of copper bolted to it.

Lapping is beneficial, depending how you do it. This is why I keep pushing the EasypcKits (http://www.easypckits.com) lapping kits on you guys. I don't own stock in the company, but until I used their Premium Lapping Kit, and did it correctly, all I got were 2C to 4C drops. I'm willing to bet SPQQKY (with all due respect my friend) Either didn't use a complete flat surface like a piece of glass, or use 400-grit all the way down to .10-micron paper. Perhaps he didn't continually clean the surface with 70% isopropyl alcohol (70% because then it's like wet-lapping) and continually clean (by tapping it pff) the sand paper. Also one move the heatsink/CPI in small figure eights or circles reversing directions. And alwsys allow the sand-paper to do the work. I f you press down with too much papaer , you may create an uneven sureface. I'm wiiling to bet that's what's happining in many cases, not enough lapping, and most P4 IHS surfaces I've lapped were uneven before I started.

It may sound complicated, it's not. It only takes about an hour to lap both surfaces correctly. Then you will see 6C to 8C drops. BUT you must use a high quality paste like AS5 once you see how thick it is, you'll understand.

Also you can practice "Tinting" rubbing a grain sized amount into the heatsink and wiping almost all away until it literally leaves a stain. Then place a small (tiny) drop of paste on the CPU center (not enough is actually better then too much) and simply tighten down the heatsink.

I'm doing an entire article on this step by step, removing the 3.0C IHS in the process, and cmparing temps. It'll be done next week. I have three reviews to do in 48 hours before that. This picture below is what your surface would look like when done, and I put the exact amount of paste to show that as well (the camera flash reflected the silver in it, it's the dark, thick, AS5);

[nailbomb]

nailbomb, where'd you get the P4 retail heatsink for 5 bucks

Refurb/Heatsink/Intel Section. They are now out of stock thou, but just keep checking. I don't think they ever last long.

[nailbomb]

Liquid,

If I'm allready close to a mirror finish on my IHS, do I need to restart with 400? I might order a kit here shortly. It seems flat to me because when I draw an X on it with a marker, it gets sanded off equally, with no obvious high/low spots.

I"m using 1/4 inch plexi to sand on.

[Liquid3D]

NO never go back, you need a finer grit then the one you left off wioth, otherwise you'll actually lap it back to being rough.

I just removed the IHS from my 3.0C, and am doing a write-up on comparing before and after;

[nailbomb]

Sweet, you got it off.

If you wanna clean the goop off easily, I can't recommend the Akasa TIM-Clean enough. It's amazing. Plus it won't eat your liver for lunch :p

[Slickthellama]

didnt you guys hear what fugger said? He said that the P4s actually are better cooled with the IHS than without. But I guess we will find out wont we.

[jmke]

Originally posted by Liquid3D
I just removed the IHS from my 3.0C, and am doing a write-up on comparing before and after;

that camera's MACRO mode is awesome :)

[Liquid3D]

yes it's the Fuli FinePix 210, I'm going to do a review on it as well. You won't belive it, a very cool friend of mine sent it to me to enhance my attchements, and bring a real visual quality to my reviews. He's most definately an Uberman! The world is a better place because of people like him (hehe). I'm still learning the Macro mode, which allows you to foucs on a non-reflective object, then scan back over to the original, but once focus is locked, you have to be careful not move it.

[Liquid3D]

Oh by the by, here's a pic for those asking exactly how much, and how to apply AS5 or any thermal paste. The silver is reflecting in the flash, so it's looks like Ceramique, but it's AS5. You might use less then indicated below, the macro mode is sort of deceptive. It looks much larger then it is, because it's reflecting in the lapped surface. Once you do this, simply mount that baby, and allow the paste to be compressed between the two surfaces, thus minimizing any chance of air incorperation.

[st0nedpenguin]

If anyone out there is doubting this application method by the way, don't.

I got a drop of about 3c just by applying my Ceramique properly, and have managed to shave another 3c off that just by experimenting with the amount applied, so it's clearly worth trying.

Keith, if you don't mind me asking, do you know how much that camera costs, I just need a ballpark figure, I'm going to be in the market for a Digicam myself soon, and that looks like a beaut.

Currency conversion isn't much of a concern, thanks to all our taxes and VAT, it works out at pretty much £1 = $1 anyway. :D

[Liquid3D]

Sure it's $200 or I beleive (how do I do the lb icon?). And here's it's zoom capabilities. It's an unbelievable camera! (This is the view out my window, which is open) from the snow you can now see why my water-cooling temps drop so low. The H20 hardware (reservoir sitting in front of push/pull radiator fans) sit on the ground below this window... The Pink building in the foreground is 300 meters away, the first row of houses, across the first section of Bay are about 2 Kilometers away, and the houses across the entire Bay are about 12 kilometers away! The sun is rising, reflecting off the Yankee-white paint, so it's seems blurry, it's not (by the way this is lowest setting 640 x 480, the pictures at 2048 x 1536 are amazing!).

[PTK]

$200.

DAMN... gotta try to look for one here in Europe.

Pe-Te

[st0nedpenguin]

My sentiments exactly, cheers Keith, you just cemented my camera choice for me, I didn't think I could afford to spend over £200 for a digican I might not use too often, looks like I won't heve to. :D

[PTK]

So its this one ?

http://www.geizhals.at/eu/a56420.html

175€ in Germany. Gotta see what that would be delivered here.

Delivery ~30e with UPS to finland.. ~18e to GB.

PTK

[nailbomb]

Keith my man,

I think you were right about my P4PE, and the mounting pressure not being adequate.

I just got a DFI board, slapped in the MCX478V, and it does WAY better in the DFI board. For instance I can see a diff in temps between 7v and 12v on the Tornado now. So, I think what I should do is get another retention bracket for the P4PE, and see if I get improvements. The P4PE was running my 1.8a at 2400 mhz, and the DFI is a 2.8C. I get lower load/idle temps on the DFI with the Swiftech, than I
did with the P4PE/1.8a (which ran stock voltage)

Thoughts? It would appear I get better mounting pressure on the new boards retention bracket.

[Liquid3D]

Originally posted by SPQQKY
After lapping my IHC and my SLK-947U and switching from OCZ Ultra thermal compound to Geil's copper and silver mixture, I got zero difference in temperatures. Oh well, nothing ventured, nothing gained.

This is somethjing I'm about to write on but I have so many things to write, I'm sitting here doing nothing. I'm so confused I don't where to sart? I have to find a new girlfirend just to help me get organized.

Anyhoo....I had an interesting talk with Nevin (Arctic Silver) and he made some very interesting points about lapping. Although in your circumstance it might not seem applicable, I beleive it may be.

Albeit a P4's IHS, or a Swiftech heatsink base, there's usually either a slight concave, or convex curvature in one or both surfaces. What happens is, when people lap, they might eliminate a convex surface where the protrusion made better contact with the heatsink surface. Lets say your heatsink base is slightly concave, and the CPU was slightly convex. You lap both surfaces. Because, your measuring your sucess by laying the heatsink, or CPU on a flat surface, the concave's deepest point may still exist. because the edges of the indentation will support it over the flat surface you wouldn't know. Ergo, by lapping the convex surface, you've now eliminated the point which protruded up from one surface, and made contact by fitting into the concave indentation.

Ergo the lapping of a single surface, albeit the IHS or heatsink, only gives you a 50% chance of improving contact. If the heatsink is convex, and the CPU concave there's a possibility that combination may fit together better then a poorly done lapping job on both surfaces.

Solution. Take a razor-blade and lay it over the surface of the CPU, then over the heatsink surface. The resulting discrepancy will enable you begin lapping with a much clearer picture of your goal, or at least upon which surface your going to have to lap down beyond the deepest part of the convex surface indent.

nailbomb (still love that username hehe) I'm not sure why but i don't rmemebr talking about mounting "pressure" being inadequate with the P4PE? Don't both those boards have mounting holes? That heatsink should be mounted through the board. I've yet to see a high-quality heatsink such as a Thermalright, or Swiftech which doesn't offer such hardware, and also recommends this as the preffered mounting method.

[Peen]

HMmmmm i took my IHS off about 6 months ago, seems to be running good since, but I wanna stick it back on just to see if my temps go up at all. What would be a good thing to stick it back on? im talking about the little black adhesive around it

[nailbomb]

I've yet to see a high-quality heatsink such as a Thermalright, or Swiftech which doesn't offer such hardware, and also recommends this as the preffered mounting method.

Man, this thread is so LONG now, but jeah I think we did talk about it pages ago.

The MCX478-V is designed to fit in the stock retention bracket with clips. There is no thru board mounting.

Now, I noticed that when I installed it on the DFI that it took more force to secure the clips, thus leading me to believe I'm getting better mounting pressure. With the 478 in my P4PE I would turn up the Tornado from 7v to full nuts, and not see any diff in temps, but on the DFI board I most certainly do, and its quite a lot actually like -5c.

So, I wonder if I got another rentention bracket for the P4PE, if I could get better mounting pressure.

[Liquid3D]

OK that's right sorry about that my friend. Now I remember what I thinking back then. How the Heck could Swifty build such a phenomenal heatsink which MUST weigh 600g, and not have offer through-board mounting? Anyhoo, my answer is yes, I think you should find a way to apply more tension. OR

You never took the IHS off that did you? I guess if the Swifty uses rentention bracket mounting you wouldn't have to worry about standoffs? Maybe that would be a better option then finding an ad hoc method of increasing pressure?

See what I'm saying? In that circumstance the lack of pressure would do well to mate with an Intel minus it's IHS. I'd go for that option. Can you show me a photo of the bottom of the swifty, or even better a close-up of it mounted on the board? Or would the plastic "cradle" Intel board-makers pin through the board obscure the portion of the photo? Anyway a photo similiar to this;

[nailbomb]

Keith, thanks :)

http://www.swiftnets.com/products/MCX478-V.asp

I think that will show you everything you want to see. Let me know if not.

I think I just need another retention bracket, because it seems like the bracket in my DFI gives increased mounting pressure, and way better temps, and that's on a proc that dissapates more heat to boot, ie the 2.8c.

[st0nedpenguin]

Could it not just be that your DFI gives different temp readings to your other board?

[nailbomb]

No, I don't think that's the issue.

On the P4PE, i saw no drops in temp from jacking the Tornado from 7v to 12V, whereas on the DFI I do.

Also, on the DFI, it took more force to get the mounting clips secured. Hence, my belief that I"m getting better contact on the DFI. Its clamped down better.

[st0nedpenguin]

I suppose it could just be a slightly different height CPU socket on one of the boards then, or maybe the older CPU runs hotter anyway, some of the old P4's used to run pretty toasty.

I don't get any differenc in temps when cranking my Tornado up to full speed either, some people would be annoyed by that, me, I'm glad, if it performed better I might be tempted to run it at full speed, I doubt the neighbours would be happy. :D

[BioBlade]

should i remove the spreader first befor lapping it ?

[Liquid3D]

OMG NO Intel now uses epoxy between Core and IHS to provide better contact for thermal transfer. Trying to pry the IHS off will destroy the core. Remember lapping the IHS will void the warranty as well. So if your prescott is a Retail version, given the fact all motherboards currently available are merely ad hoc modifeid for Prescott's (that is only BIOS changes have been made allowing a lower Vre), there's simply too much risk the chip will be damaged. this would leave you with no recourse, and a loss of $185. Until motherboards have physical modifications made to their power circuitry, giving them true 478 Prescott support, there's an inherent danger to both mobo, and CPU. Warranties are going to very imporatnt to maintain until these changes have been effected.

EDIT: basically the last step in Riska's AMD64 guide below where he uses Arctic Silver epoxy to re-attach the IHS is what Intel is now doing to all chips. Of course Intel isn't doing such an excllent job. I also doubt they're using high quality epoxy (actually the best) such as Arctic Silver!

What concerns me, is mobo makers won't expend the capital needed to make these physical changes. That we 478 Prescott owners will be using motherboards with inadequate power supplies, which will eventually become damaged (check this (http://hardocp.com/article.html?art=NTg5) out). I hope they won't leave thousands of Prescott owners with systems having a limited life. Tejas Prescott's will have the red carpet rolled out for them in the form of Socket 775 LGA (Line Grid Array) boards featruing Adlerwood/Grantsdale chipsets, DDR-II and PCI-Express (3GIO).

Even if you had a version without epoxy bonding IHS to core, it's a mirror finish and each I've tested (placing a razor blade edge along it's length) was perfectly flat. Lapping is to ensure both surfaces are as flat as possible, and as free of striations, and micro-pores as possible. Only when both surfaces (heatsink base and CPU) are perfectly smooth, and flat, can they mate together without any air pockets remaining.. That's the reason thermal paste is used, it becomes THE catalyst for thermal transfer, filling any grooves lapping may miss.

[Riska]

I have done this for my Amd 3400+ 64 cpu here is a guide but it is in danich but i have taken alot of pics!

http://www.burn-it.dk/index.php?state=artikler_read&artikler_nr=38

[Liquid3D]

That is the most thorough heatspreader removal guide I've ever seen! Excellent work, and excllent photo's. May I place a link to it on my website mentioning you as the Author? www.ksbrainstorms.com

[nailbomb]

Keith,

I like you're new avatar.

I agree, the whole Prescott story seems a little fishy. I'm glad I got a 2.8c instead.

[ultron5]

I've done this /w my 2.4C too.

http://mitglied.lycos.de/coolinglimit/Hardwarepics/IHS-Off/IHS_Off%20017.jpg

http://mitglied.lycos.de/coolinglimit/Hardwarepics/IHS-Off/IHS_Off%20013.jpg

http://mitglied.lycos.de/coolinglimit/Hardwarepics/IHS-Off/IHS_Off%20007.jpg

Got 25MHz more out of it (3-4 Deg) :D

[dioSRL]

thats a SL6Z3, it supposed to be glued.. How did you manage to remove it without breaking the core apart?

[ultron5]

yeah its a SL6Z3!

I removed the IHS with a small cutter and a razor-blade! I was afraid of killing the Chip but it survived. ;)

[Riska]

Liquid3D
Sure put i op there!!

[fraguzz]

Is there any chance, it can work on a celeron 2,0 "NW"
the laping thing??
it looks like a p4 and basikly is....

[Liquid3D]

Any CPU with a heatspreader can, and perhaps should, be lapped. HOWEVER; prior to lapping, you should take a razorblade and lay it's edge across the center of your CPU to see if its' flat (and mating heatsink). Many CPU-heatspreaders are slightly convex in shape. Some are even concave, but most are not "perfectly" flat. IF your CPU's IHS center has a convex protrusion, lapping may acutally reduce it's cooling ability. This is especially true if the heatsink, or waterblock your mating it with is slightly concave in it's center (convex to convex would be better). You cannot determine flatness by laying the CPU or heatsink on a flat surface. If it's concave the edges will supoort it, and you'll never know. That's why using a razor blade, or some other simliar object is preferable. Your using the edge of the razor-blade and can place the dge on the CPU diagonally, or across it . etc.

Secondly; there's a difference here between "lapping" and removing the IHS. I don't suggest removing any Intel IHS manufactured after 10/2003. And all Prescott's and Extreme Editions.

Finally; with all due respect that CPU really doesn't look like a common Intel Northwood, it's retangular shape is very strange? Perhaps it's the angle of the photo, however; the retangular shape looks like it's accomodating the 2mb L3 cache of an Extreme Edition? Intel uses Shin-etsu and I've never seen it so "silver". It does look more like the Silver Epoxy used in the EE's? If so it might be one of the few succesful EE Intergrated Heat Spreader removals. Are you sure you don't have an EE there? when and where did you buy that chip ultron5?

See how square this 2.4C core is below;

[nailbomb]

Some are even concave, but most are not "perfectly" flat.

And in my case one side was concave, and the other was convex ;) That's no lie either. I could tell by the way the copper was coming off. Also I draw an X across the HS with a Sharpie to see what's going on.

[Liquid3D]

I Definately beleive that's feasable, my friend, most definately.

By the way I haven't forgot about you, I'm still working on that and may have some good news. Don't ever feel shy about sending an email to remind me, if more then a week or so passes. Sometimes I get overwhelmed with stuff. hehe

[nailbomb]

No problem Keith, and if I get to feeling better I might give you call to BS for a bit. I
have had a nasty cold this week. Thought I was going to squeak thru this season without getting it too :p

[saaya]

ultron5 has an m0 stepping chip (gallatin core with defective or disabled L3 cache) thats wey its that much larger.

[ultron5]

Yes the chip is a gallatin with disabled L3 (are all m0 gallatins?)

http://mitglied.lycos.de/coolinglimit/Hardwarepics/SL6Z3/SL6Z3_Costa_Rica.jpg

I bought the Chip XMAS'03! ;)

[fraguzz]

okey they dont look the same under the IHS but ontop!
i do not want to remove my IHS!!

2000@2800 (standard Volt)
my temps:
room temp: 23C
cputemp: 33C idle (load: 46C)
motherboard temp: 32C

thats with good airflow inside of case..
standard heatsink from intel (still same thermic tape)

----------------------------<

im thinking of lacping then use artic silver 5..
any idear what that would bring me? im stuck at 163mhz*20
my system:
soltek SL-85ERV
kingston pc2700 512mb

i ran the core at 1,664V
and the DDR at 1,78V
seems like its the ram thats setting me bag :( because at about 3000mhz i had to raise my ddr voltage to even boot! now at 3257mhz it can go fine (with enought cooling) but over that, no matter how higg i set my Core voltage it keeps §§§§ing up windows! blue screen... win2000

but my ram should take 166mhz right??