The Notebook Review forums were hosted by TechTarget, who shut down them down on January 31, 2022. This static read-only archive was pulled by NBR forum users between January 20 and January 31, 2022, in an effort to make sure that the valuable technical information that had been posted on the forums is preserved. For current discussions, many NBR forum users moved over to NotebookTalk.net after the shutdown.
Problems? See this thread at archive.org.

    No IHS mod - Direct chip to heatsink Mod, HELP!!

    Discussion in 'Sager and Clevo' started by wyvernV2, Jan 8, 2018.

  1. wyvernV2

    wyvernV2 Notebook Evangelist

    Reputations:
    177
    Messages:
    567
    Likes Received:
    386
    Trophy Points:
    76
    Hello,
    So i have a p775tm laptop with a i5 8600k,
    Now my new i7 8700k and rockitt delidding tool, is on its way to me.
    Now was just thinking could i do the "No IHS Mod" for best thermal performance in OCing?

    So here is my plan:-(sorry for bad image editting and feel free to correct me)

    IMG_20180107_192036.jpg
    Blue area: will delid it and cut out the blue portion off the IHS and stick it at its place ON the chip so that the socket clamp on the motherboard can hold the cpu at its place.
    IMG_20180107_191117.jpg
    Green area: will stick the cut out blue portion from ihs and stick it on its place at green area on the chip to hold it on its place.
    Red area: Now this is where i am confused, lets say the LM drips off the chip-heatsink contact surface, will it damage the cpu? If yes then i am thinking to cover all the red portion with thermal pads!

    So this is going to be one of the biggest experiment/mod on any laptop by me so far, so just wanna be sure is it even possible?

    BTW if i remove IHS then i know there is going to be a huge gap between chip and heatsink, so will cover it by adding some copper shims and thermal paste.

    Tags: @Falkentyne @Mr. Fox @Phoenix @Vasudev @Midas Touch @hmscott @XMG @Papusan
    Note: i am doing this madness because the i7 8700k i got is for 120$, i actually live in india, and my uncle who lives in usa got this sweet deal from a local store who is selling out all his things for cheap beacause he is selling out his whole shop,(what a luck), so I thought why not experiment with it :vbbiggrin:. It wil be reaching to me by mid march, when my uncle would come to india!
    EDIT: new tags! @Johnksss @Khenglish @t456
     
    Last edited: Jan 8, 2018
    Beemo, Prema and Vasudev like this.
  2. Vasudev

    Vasudev Notebook Nobel Laureate

    Reputations:
    12,050
    Messages:
    11,278
    Likes Received:
    8,815
    Trophy Points:
    931
    You do know I haven't de-lidded by desktop CPUs ever?
    Papusan, Mr. Fox, Johnkss, Khenglish, t456, ajc866 are the ones who do the modding.
     
    wyvernV2 likes this.
  3. wyvernV2

    wyvernV2 Notebook Evangelist

    Reputations:
    177
    Messages:
    567
    Likes Received:
    386
    Trophy Points:
    76
    I taged you for help in the red area, where there is the danger of dripping the lm on the chip, so i thought you could help me on that!( frankly i didnt knew you havent delidded, i just knew that youdont have much exp with clevos though as you said to me before,sorry for that! ;))
     
    Vasudev likes this.
  4. Vasudev

    Vasudev Notebook Nobel Laureate

    Reputations:
    12,050
    Messages:
    11,278
    Likes Received:
    8,815
    Trophy Points:
    931
    Never had a Clevo in my life. I used only BGAs from Sony, Lenovo and Dell/AW.
     
    Vistar Shook likes this.
  5. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    8,396
    Messages:
    5,992
    Likes Received:
    8,633
    Trophy Points:
    681
    @TheBeastDragon

    With LGA CPUs, because of the way the socket clamp is designed, which BGA CPU's do NOT have, it's MUCH harder to de-lid and get proper contact on the CPU surface, because the entire clamping mechanism was designed so that the IHS is slightly elevated above the retaining mechanism, but when you remove the IHS, you're also removing clamping pressure as well, so you're at a lot of risk of the heatsink not even touching the CPU surface.

    Even if it DOES touch the CPU BGA surface, you have to be careful that not too much pressure is applied also. BGA CPU's have heatsinks that are designed for proper mounting (although the pressure is still very low). But for a completely de-lidded LGA, you have to worry about uneven heatsink pressure on the silicon also. Desktop CPU users don't usually do direct die contact like this, because with the heatsink sitting vertically in a case, there is also HUGE risk of cracking the die. while the risk may not be the same with desktop CPU's on laptops, you still have to be careful.

    First, even for *RELIDDERS* (the usual method--delid, put liquid metal on CPU surface **AND** on backside of IHS (putting on IHS backside helps improve temps)), a very useful thing you find people doing over on overclock.net -->intel delidded club, is applying THREE COATS of TRANSPARENT NAIL POLISH over the internal SMD resistors around the CPU housing. There is an IMPORTANT REASON FOR THIS---some people had LM leak off the silicon and eat the solder away off the SMD's, causing a component to just fall off. Using nail polish will insure this never happens, as well as protecting from any short circuits. Everyone who de-lids should take note of this simple effective step.

    Anyway, for you, you should do the nail polish trick regardless. That will stop short circuits.

    The next problem is preventing runoff LM, like a tiny drop escaping (like a rogue or renegade conductive ball of doom) from getting on the *MAINBOARD*.
    Usually the IHS would stop that completely. But if you're going to try your method, what you need now is something called a FOAM DAM.

    This foam dam *MUST* be VERY VERY HIGHLY COMPRESSIBLE. The starting thickness can be about 2-3mm, but it MUST easily compress with your fingers to less than 0.1mm, without any resistance. On heatsinks that have low downward static pressure, any resistance is bad, so you absolutely do not want to use thicker foam you often see in packing materials. you want the really compressible stuff (which isn't as common). Any hardware store should have something like this, even a crafts and hobby store.

    Here's an example of a foam dam around a delidded 7700K, from @Mr. Fox 's old system. but re-lidded, because there is liquid metal both UNDER the IHS (on the BGA slug) and on TOP of the IHS as well. (on desktop computers, you RARELY see LM on top of the IHS since many, many heatsinks are aluminum, as they are much larger and heavier, and copper heatsinks cause a serious weight and balance issue (google the famous Thermalright TRUE Copper for more information about the problems that caused).

    [​IMG]

    Since you want to experiment with bare die, the foam dam is no longer 'free insurance', like it is on Mr Fox's, and other people's systems (even mine; I have foam dams around my BGA 7820HK and TDP modded 1070 GTX, with LM), but now it is an absolute REQUIREMENT, since you have absolutely zero protection if you remove the IHS. So if you want to go with your mod, it's mandatory.

    Unlike the square cutout foam dam above, you will need a rectangle foam dam type shape, something ugly and ghetto like this:
    AtMMjdU.jpg

    And put that around the CPU socket. Unlike BGA, you may need this even thicker initially because you now have an LGA retaining mechanism in the way, which may also interfere with the --heatsink--side of the LM connection (remember I told you you should paint a thin layer of LM on BOTH the cpu heatsink AND the CPU, when doing a LM mod with a copper (or nickel plated copper) heatsink, if you want LM on top of the CPU IHS (or CPU bare?). So you're going to have to make sure your foam dam you make barriers both the CPU housing and the LM covering you made also on the heatsink., without the IHS getting in the way somehow. How you go about that is up to you.

    I SUGGEST if you are going to do a -full- delid with no re-lid, use normal thermal compound first, WITHOUT POWERING ON THE SYSTEM, as a 'dry test run' to make 100% sure that the CPU and heatsink actually have contact and enough pressure. I would put a single pea-sized dot of normal compound (white stuff, kryonaut, any cheap old stuff) in the center of your delidded CPU, mount the heatsink, and check for spreading. If it spreads extremely well, then you can disassemble, clean up and try the liquid metal mod. You don't want to risk powering on the system without the heatsink touching the CPU somehow!
     
    wyvernV2, Mr. Fox and Vasudev like this.
  6. Danishblunt

    Danishblunt Guest

    Reputations:
    0
    Those delidd tools are "idiot proof" so no worries about breaking your CPU.

    You can leave the IHS as is after applying lm. Also you can use scotch 105c isolation tape to isolate in case you're scared of shorting your CPU before getting the IHS back on.
     
    wyvernV2 likes this.
  7. Prostar Computer

    Prostar Computer Company Representative

    Reputations:
    1,257
    Messages:
    7,426
    Likes Received:
    1,016
    Trophy Points:
    331
    The isolation tape or Kapton tape should work well for sectioning things off. Good luck with the delid!
     
    Vasudev and wyvernV2 like this.
  8. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    8,396
    Messages:
    5,992
    Likes Received:
    8,633
    Trophy Points:
    681
    Transparent nail polish is used alot these days for painting over the resistors. Plus that can result in slightly better temps as well (less "materials" trapping random heat flow). And it's also faster to work with. But regardless of what path you choose, some type of protection must be done. That's user choice (me, I prefer transparent nail polish, which is very popular on the overclock.net Intel CPU delidded group club), but nothing wrong with tape either.

    Isolation or Kapton tape will NOT prevent runoff onto the mainboard! That is used (instead of nail polish) for protecting resistors (again). And this is where your foam dam comes in. The foam dam is your protection of the mainboard. Tape will not stop LM. Conductive Balls of Doom have a mind of their own and will run over whatever they please, when they please.
     
    Vasudev likes this.
  9. nedooo

    nedooo Notebook Consultant

    Reputations:
    40
    Messages:
    161
    Likes Received:
    67
    Trophy Points:
    41
    I would suggest that you make mould (3D print) for making pure silver IHS also to plan to make inner part of the IHS thicker just to shape it to make perfect contact to CPU die, no LM paste can beat that thermal conductivity...just my 2¢...
     
  10. bennyg

    bennyg Notebook Virtuoso

    Reputations:
    1,567
    Messages:
    2,370
    Likes Received:
    2,375
    Trophy Points:
    181
    OP, that's a lot of hassle and risk for what will probably end up minimal gains over a standard delid with liquid metal approach. I got fantastic thermals with LM on top of IHS as well, that was able to push the rubbish 6700K I had in my P870dm-g as far as it would go (a lowly 4.6 before some strange system agent current throttle) with plenty of thermal headroom still available
     
    Last edited: Jan 8, 2018
    Vasudev, wyvernV2 and sicily428 like this.
  11. wyvernV2

    wyvernV2 Notebook Evangelist

    Reputations:
    177
    Messages:
    567
    Likes Received:
    386
    Trophy Points:
    76
    I presumed that the chip-heatsink contact would be compromised, so i ask, would the coppershims be enough to get the heat properly transmitted to heatsink?
    So, i cant figure it out, nailpolish on the chip board(red area), or around the MOBO area around the socket?

    Als the foam dam idea looks promising rather then thermal pads, Thanks for that!!
     
  12. Falkentyne

    Falkentyne Notebook Prophet

    Reputations:
    8,396
    Messages:
    5,992
    Likes Received:
    8,633
    Trophy Points:
    681
    Nail polish ONLY on the CPU package area around the silicon shiny exposed slug, covering the little tiny resistors, not on the mainboard itself.

    Then, foam dam (Very highly compressible) cut out to fit around the CPU slug, MUST be very highly compressible, but still "thick" enough to properly touch the heatsink area too, since you will also "paint" liquid metal on the heatsink itself (in the shape of the CPU BGA). That will stop any run-off to the mainboard.

    if you're using a shim, this gets a little more complicated. Then in that case you should have two foam dams--one cutout around the CPU, and another around the "CPU area" of the heatsink (shaped but this time, more like the shape of the actual shim (my mistake, not the CPU), but in this case you just want a thin boundary since you want to maximize heatsink transfer area. You know what I'm talking about, so a square foam dam cutout, but with lots of space this time for the "heatsink" side of the shim). Make sure there is a little extra space (a few mm) in the foam dam cutout.

    To make sure the foam dam does not move out of place, you can put a drop of nail polish around the corners, attach the foam dam, and wait for it to dry a bit. Then it will stay in place :) :)

    Thermal pads as a dam have too much resistance to compression. This creates less pressures for laptop heatsinks going down. It's the same problem as using "non highly compressible foam"--the thicker foam causes resistance and a lack of downward pressure from the heatsink. Especially since you want the foam to act as a dam both around the CPU itself AND on the heatsink side (both sides). If you're using a shim, then ONE foam dam is required (on CPU), but I would use a second on the top side of the shim too (since you need Liquid Metal on both sides).
     
    Last edited: Jan 8, 2018
    wyvernV2 likes this.
  13. Meaker@Sager

    Meaker@Sager Company Representative

    Reputations:
    9,436
    Messages:
    58,194
    Likes Received:
    17,909
    Trophy Points:
    931
    You also risk warping the package and poor pin contact.

    You are looking at a 1-2C gain in temps at best too according to tests I've seen in desktop systems.
     
    Vasudev likes this.