From my understanding TIM is used to fill in microscopic grooves between two less-than-flat surfaces. I'm curious though, if you lapped both the heatsink base and the IHS of the CPU ending up at say 3000 grit sandpaper, would you have a "perfectly" flat surface that wouldn't require any TIM (therefore having excellent thermal conduction)?
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
Not using any TIM - two perfectly flat surfaces
- Thread starter eddieck
- Start date
SicKlown42012
2[H]4U
- Joined
- Jul 6, 2008
- Messages
- 3,319
Even machined surfaces will have gaps. Perfectly flat surfaces are damned hard(if not outright impossible) to come by, be it in nature or by man's hand.
Mr. Pedantic
[H]ard|Gawd
- Joined
- Sep 19, 2009
- Messages
- 1,707
There is no way you could get them perfectly flat by hand. Even if you could use the IHS as a mirror there would still be microscopic imperfections in the surface that would affect thermal conductivity. It's just that if you lap it you don't have to use nearly so much.
Belt sander maybe? This was just a random idea I had; figured I'd post here since you guys would know more than me though I was pretty confident it wouldn't work (or it would've been done already!).
I haven't done this. This was just an idea. Read above.
- Joined
- Aug 10, 2009
- Messages
- 2,909
laser cut sounds like it'll work. hmm i'd like to see that
Derfnofred
Gawd
- Joined
- Dec 11, 2009
- Messages
- 617
Just give me a laser pointer and a quality optical microscope. No need for anything that fancy. Hit the lapped surface with a grazing angle and look at the scattering in the microscope. Super easy.
OP: Prime grade silicon wafers are pretty dang flat (far and beyond your capacity for a HS and spreader), and they still establish pockets when you try and bond them together.
Lapping will hopefully allow you to use LESS TIM, but eliminate? Not on my watch.
D
that is not actually correct. forget about microscopic grooves - TIM is used mostly to fill the largely uneven surfaces in the top of the CPU. the bottom of quality heatsinks tend to be much more flat. in a few core2 duos that i have owned (and lapped) there has been as much as a few hundredths of an inch in concave-ness in the top of the heat spreader. ive heard similar things in the core 3/5/7's and some of AMDs chips.
once you get rid of those large fluctuations, then it will be used mostly to fill those little grooves. if you have an ultra flat heat spreader and heatsink, then you wont have to use as much TIM, but you still need some.
some people argue that having the heat spreader slightly convex helps make better when applying pressure to the top of it because it flattens out and makes sure the very center of the HS has the best contact, not out around one edge, for the best cooling.
BillParrish
Supreme [H]ardness
- Joined
- Aug 25, 2006
- Messages
- 7,519
Keep in mind "flat" and "surface roughness" are not the same. I use a jewelers faucet cutting machine to lapp my stuff to .5 microns and they are so smooth and flat they "stick" together but I still work in a haze of AS5 with my fingertip in a baggie on both surfaces before installing. It is likely I will never go as fine again.
Long ago in the old time, billions and billions of electrons died over the arguments about surface roughness (size of microscopic scratches) and the particle size of heatsink compounds. (That the parts where flat to a high degree of accuracy was a given) The result being that most but not all felt that one should shoot for scratches (surface roughness/grit size) equal to the average diameter of the particles in the compound and 1/2 as deep. The idea being that if you can get the scratches lined up (impossible but in a random environment the best we could do) a thermal compound particle would 1/2 way fit perfectly in one surface and hopefully mate up with a scratch in the other. The situation to be avoided was to the the particles actually hold the surfaces apart. So you wanted to avoid surface roughness/scratches that where smaller than the particle size in the compound so the particles would "fall" into the scratches and not sit on top of the surface. This is why it is recommended that one lapps both the heatsink and CPU so that scratches go in the same directions.
This is extreme case. It is unlikely with the common "pane of glass" and wet and dry paper you can get anywhere near conditions where any of this will really matter (IMO).
note "scratch" is not really the right word for these microscopic surface irregularities when you get down to that level of surface roughness but I cant for the life of me think of the correct term.
AS5 spec.
Average Particle Size:
<0.49 microns <0.000020 inch
to get to .5 microns requires P30000 grit and there is no way outside a very fancy lab to ensure you get the aprox .25 micron depth one would in theory want to have.
3000 grit wet and dry will have about 7um grit and thus (at least in theory) 14 AS 5 particles could "fall into" just one surface scratch and would need to "stack up" as the scrach depth would far exceed the particle size in hopes of meeting up with particles or the metal of the mating surface. It is almost unarguable this "stack" of particles, even if silver would not have better conductivity of heat than a metal to metal interface. So smoother is better but it is all theory. Unless you dad happened to be rock hound and left you jewelers grinding/finishing equipment there is not a lot to be done with a sheet of glass and sandpaper after you get the big bumps out and get stuff reasonably flat. (and even then I no longer think it is worth the effort).
All my opinion. You do not have to beleive a bit of it. For sure other people disagree. Do what makes you happy. Even when I was aware of this and before I had the use of the machine I liked to make a mirror finish on the parts. It just made me happy. Next personal build I will slap a decently fine disk on the machine and get both surfaces flat and not spend 3 days mucking about.
Long ago in the old time, billions and billions of electrons died over the arguments about surface roughness (size of microscopic scratches) and the particle size of heatsink compounds. (That the parts where flat to a high degree of accuracy was a given) The result being that most but not all felt that one should shoot for scratches (surface roughness/grit size) equal to the average diameter of the particles in the compound and 1/2 as deep. The idea being that if you can get the scratches lined up (impossible but in a random environment the best we could do) a thermal compound particle would 1/2 way fit perfectly in one surface and hopefully mate up with a scratch in the other. The situation to be avoided was to the the particles actually hold the surfaces apart. So you wanted to avoid surface roughness/scratches that where smaller than the particle size in the compound so the particles would "fall" into the scratches and not sit on top of the surface. This is why it is recommended that one lapps both the heatsink and CPU so that scratches go in the same directions.
This is extreme case. It is unlikely with the common "pane of glass" and wet and dry paper you can get anywhere near conditions where any of this will really matter (IMO).
note "scratch" is not really the right word for these microscopic surface irregularities when you get down to that level of surface roughness but I cant for the life of me think of the correct term.
AS5 spec.
Average Particle Size:
<0.49 microns <0.000020 inch
to get to .5 microns requires P30000 grit and there is no way outside a very fancy lab to ensure you get the aprox .25 micron depth one would in theory want to have.
3000 grit wet and dry will have about 7um grit and thus (at least in theory) 14 AS 5 particles could "fall into" just one surface scratch and would need to "stack up" as the scrach depth would far exceed the particle size in hopes of meeting up with particles or the metal of the mating surface. It is almost unarguable this "stack" of particles, even if silver would not have better conductivity of heat than a metal to metal interface. So smoother is better but it is all theory. Unless you dad happened to be rock hound and left you jewelers grinding/finishing equipment there is not a lot to be done with a sheet of glass and sandpaper after you get the big bumps out and get stuff reasonably flat. (and even then I no longer think it is worth the effort).
All my opinion. You do not have to beleive a bit of it. For sure other people disagree. Do what makes you happy. Even when I was aware of this and before I had the use of the machine I liked to make a mirror finish on the parts. It just made me happy. Next personal build I will slap a decently fine disk on the machine and get both surfaces flat and not spend 3 days mucking about.
Last edited:
Bad idea. This entire plan is a bad idea to be honest. Outside of extremely specialized precision equipment like Bill has already pointed out, it is impossible to produce a surface sufficiently flat to achieve the results you are aiming for. A typical lap job is the best you can hope for (and even then, it really isn't necessary and it will void the CPU's warranty).
meatfestival
[H]ard|Gawd
- Joined
- Apr 5, 2005
- Messages
- 1,767
This may be of interest to the OP
http://indigo-xtreme.com/
bonds your CPU to the heatsink by reflowing a metal alloy
Only thing is you need to let your CPU get to scary temperatures when installing it
http://indigo-xtreme.com/
bonds your CPU to the heatsink by reflowing a metal alloy
Only thing is you need to let your CPU get to scary temperatures when installing it
I use Indigo Extreme. Installed per the directions .. no problem.
I also exchanged a couple of emails with them that further allayed my concerns about the reflow ... take the processor over 90C
They are a small company that developed the material as a part of gov contracts & were able to go commercial. Very difficult to do & I hope they succeed.
I also exchanged a couple of emails with them that further allayed my concerns about the reflow ... take the processor over 90C
They are a small company that developed the material as a part of gov contracts & were able to go commercial. Very difficult to do & I hope they succeed.
Even if you have 2 surfaces ground, polished, lapped, acid etched, etc to be as flat and smooth as possible, TIM in small amounts will still help and can't hurt unless you use too much and it holds the 2 surfaces apart. the TIM should move out of the way and allow as much direct metal-metal contact as would be present with no TIM present, but then adds more thermal transmission surface by filling in the gaps that remain.
This may be of interest to the OP
http://indigo-xtreme.com/
bonds your CPU to the heatsink by reflowing a metal alloy
Only thing is you need to let your CPU get to scary temperatures when installing it
I already use Indigo though I need to buy another kit since I had to pull my heatsink off. Great product. And 90C isn't a problem for a CPU... I ran my i7 without a heatsink for about a minute (was just seeing how long it could take the heat before it shut down).
Yeah, I'm not sure if I'll lap the IHS or just remove it entirely. Removing it would present another problem with the huge D14 heatsink. Not to mention it takes somewhere around 6 hours apparently to sand off the IHS (the safest removal method).
I'm not worried about the warranty on these things - CPUs don't really die within their voltage specifications so the only other real danger is a poorly designed Apevia PSU (LOL) but I've got a Seasonic S12D so I'm OK in that area.
drescherjm
[H]F Junkie
- Joined
- Nov 19, 2008
- Messages
- 14,933
I lapped my i7 and heatsink base with 1500 grit. (I might try to get my hands on some 3000 grit later but can't seem to find it locally, and the auto parts store I went to was out of 2000 grit.)
End result: *WITHOUT* any TIM, *WITHOUT* any fans I can run Linpack at stock i7 speeds maxing out at around 96C on the hottest core. Though this is throttling territory RT didn't seem to report any throttling. I shall of course be adding some TIM and later on ordering a new Indigo kit but this is definitely a step in the right direction.
End result: *WITHOUT* any TIM, *WITHOUT* any fans I can run Linpack at stock i7 speeds maxing out at around 96C on the hottest core. Though this is throttling territory RT didn't seem to report any throttling. I shall of course be adding some TIM and later on ordering a new Indigo kit but this is definitely a step in the right direction.
robothunter
Limp Gawd
- Joined
- Jan 19, 2008
- Messages
- 369
Even if you had two completely flat surfaces by whatever means possible, the forces applied by attaching the heat sink to the CPU would cause some amount of warping of the surfaces. Thereby defeating the purpose.
If the clamping force was perfectly normal to both surfaces, there would not be any warping.