Condactonaut fail

I just repasted the alienware 17R4 with condactonaut, and my cpu temperatures suddenly jump all over the place, from 70-100 degrees and it thermal throttles every other second. Does anyone have any clue to what the issue might be?

 

If you did not replace the pads on the heatsink. Then goof chance there is a gap. And that is what causing Conductonaut to be effective.

If you did change the pads then i would check if the heatsink is warped causing a gap.

 

You have to take it apart again. When finished cleaning liquid metal off, try seating the heat-sink without screws and without liquid metal and check for proper contact between CPU and GPU dye and heat-sink by pressing down and/or by tapping on top of the heat-sink above CPU and GPU. If no sound is produced, you have proper contact. Then put liquid metal.

First idea after reading your post was that some thermal pad got out of place and does not allow for good dye contacts with heat-sink. Good luck and let me know how it goes.

 

Played around with pads that looked like they were too thick (using polyfuji 17W) and added some liquid metal (I think the glossy surface was just too little and it wasnt making proper contact as said) rescrewing everything again and will let everyone know.

 

So yea, basically you are definitely right. I have a ****ty heat sink (wow what a surprise) and when more LM was applied the temperatures fixed themselves to normal values however my core differential when I stress the CPU is humongous.

I just wanna ask if the temps are actually pretty OK and core #0 and #2 spikes and is around 20 degrees higher then any other core? I know we wanna aspire to the lowest core differential possible, but the temps when stressed in OCCT hang around 60 degrees while Core#0 is just spiking hard. I tried bending the tripod arm however as we can see it is not working properly. Is it safe to use such a CPU and what else could I do to solve it?

EDIT: So basically my cpu has 20 degree differential (never had it that big, not even on stock thermal paste) however the core differential CPU temp jumps high (up to even 97 degrees at times in for example witcher 3 while normally it is chilling around 80-86 spiking up to 93-95 for a split second, while core 1 and 3 are chilling at 78 degrees). Also the GPU goes up to 75 degrees (which I know is not bad for a Witcher type game in 1440p) but still I thought it would maybe go up to 70 with liquid metal inside.

 

LM is not for noobs, i am sorry but is the truth. LM is not suitable for every laptop or desktop, do you have to know your machine before doing it, not putting an electric conductive liquid inside your laptop and see what happens. i am not going to waste time teaching you how should you do it. here you can find a lot of threads talking about that. for the next time dont do things that may entirely damage your machine without reading

 

Some basics on liquid metal pasting:

1. Make sure entire core is fully covered.
2. Make sure heatsink area that contacts core is also fully covered.

Liquid metal does not spread, so if it's not covered when the heatsink is screwed on, it's not going to be covered by TIM.

 

Definitely did that. That's why most of the work was done by a friend who knows more about hardware than I do. We actually had huge problems with temperatures so we added a little more than a thin layer (ofc not too much, just so the coating is more firm and just looks better). First we did the repaste with kryonaut and it didn't work very well so we decided to do LM. We did a precise coating, secured CPU and GPU with electrical tape. Spread it so we had a nice glossy surface. Repaded it, dry tested the heat sink a couple of times, pads I think are rather OK. So yea, I don't think it's about being a total noob with applying LM (though I completely am) I think it is more of the ****ty S-J-1 revision of the heat sink and maybe the padding might be to thick?

 

in your case i should removed all the thermal pads and had applied IC diamond 7 on vram, to avoid a bad pressure of a wrong tickness thermal pad, and conductonaut on the cpu and gpu with the third arm "mod" suggested by iunlock

 

If you open it again, can you take photos with the heatsink assembly removed / upside down etc? Looking at some other heatsink from some other Alieware around these forums, it seems to me that the rigidity provided by the three interconnecting copper heat pipes between GPU and CPU, cannot be overcome by the mounting screws alone, of which we have only three on the CPU side, and that on a thin metal tab rather than a cast metal support. Maybe, just wild guessing here, there is a mismatch on the levels of the CPU and GPU as far as the heatsink plates are concerned and maybe you'd need to align it by sight off the motherboard and make sure it's a perfect fit before even trying to put on?

 

The application sounds correct. You do want slightly more than a thin layer. I didn't mention that before because people have globbed it, which then got dislodged and wedged under the GPU BGA or CPU socket or other very bad places.

If application and die contact are good you will see the entire shape of the die imprinted onto the liquid metal on the heatsink when the heatsink is removed. Do you see that? If not then the heatsink is uneven. Other pastes will also show this.

 

Mister, there 2 simple but important things to be careful: padding thickness and coating height. If the thermal pads have very deep dents around cpu or gpu there is a lot of uneven pressure on them and that might rise the heatsink and gives you little contact on dies. Second... the coting with the electrical tape is sketchy cause depending on the product used...you might or might not use it properly. The 33+ needs stretching... A LOT ...trust me... already happen to me, the coating level must be under the shiny part of dies. Keep an eye especially on corners where coating overlaps.
Hope you can reach what you hope for...


As you can see... the stretching level...and the height difference between coating and gpu die.

 

^ This is why I prefer nail polish coating of the SMD'd around that GPU (and the exposed traces or SMD's around some CPU's). Avoids that problem.
Another reason you NOT want to use an extremely thin layer of LM, regardless of what you have read elsewhere, is because you need enough to compensate for the battery effect on copper, which will cause partial absorption of the LM (and if the fit is very poor, a LOT of absorption, due to air, no pressure and extreme thermal stresses). Painting and then vigorously rubbing in a LM layer on the copper above the CPU area will start the battery effect directly on that layer rather than what's on the CPU, so this is very useful. Even on desktop LGA IHS delids, doing this under the nickel plated IHS will always give slightly lower temps!

The drawback of using more LM than the minimum is runoff threats, especially when dealing with convex heatsinks or transporting laptops in backpacks.
You are STILL going to have to deal with possible dislodging or runoff even if you do use Super 33+ tape or decent Kapton tape. Tape won't prevent runoff at all

Runoff is spare traces of LM balls that escape from the housing, do NOT get trapped on insulated SMD resistors and slide ACROSS the CPU or GPU housing and off onto the motherboard, where it WILL fry something. That's where barriers come in. A common barrier first used by an airline pilot over here (who takes his laptop with him, and who NEEDS a barrier) is highly compressible foam.

The foam must be VERY porous and not dense at all and *MUST* compress to the width of a human hair, with your fingers, with MINIMAL resistance, with a cutout shape in the shape of the CPU or GPU, with a few millimeters to spare for proper clearance (you don't want the foam blocking the heatsink from touching the cores). THESE DAMS ARE NOT EVER, EVER TO BE USED IN SUBSTITUTION FOR NAIL POLISH INSULATION COATINGS, OR SUPER 33+/KAPTON TAPES, EVER, but in COMBINATION with them.

The job of the foam is to catch any spare LM balls that escape the chips, and block them from going anywhere else. They will sit on the foam with nowhere to go. If you had a bad mount and the temps started skyrocketing, you will have saved your hardware.

The cutout foam MUST be very thin. On high pressure heatsinks like desktop heatsinks or AIO's, if foam happens to be used, that is not important, just compressible. But laptop heatsinks have atrocious mounting pressure so ANY resistance from foam will decrease resistance even further. So do your work.

Let's say your foam is 3mm thick and highly compressible.
You MAY get away with this, but to be safe, you should CUT the foam width in half to 1.5mm. This will be ideal. If you're good and OCD, or just plain pro, go for 1mm. Thinner is better. Just don't have it so thin that it breaks apart.

Don't try to cut an 'uncut' (meaning: square shape of foam, that has not had the CPU or GPU layout trimmed from it yet to trim it, that will be too difficult. Instead cut out your shape first, trim the outside to the exact shape of the CPU and GPU housings, and THEN use extremely precise scissors like titanium scissors, and then start working around both the inside and the outside to trim some of the thickness away. Once you start this, you will see why the "cutout" is much easier to trim than an uncut block

To secure the foam dams so they don't shift and move when applying the heatsinks, put a dab of transparent nail polish (the same stuff you used for insulating the SMD resistors) in each corner and secure it.

Left image is original cutout foam dam, 3mm for BGA CPU. Right three images are trimmed 1mm cutout foam dams for BGA CPU and GPU's.
Trying to use 3mm (or thicker) foam requires VERY low density. The thinner the better.

Some substitutions for foam dams can possibly be thin compressed layers of silicone gasket sealer--the same stuff used for relidding IHS's, although this requires preparation--this MUST sit BELOW the level of the CPU or GPU as this otherwise adds resistance! There are ways to prep this, including applying super thin layers and squashing it with Super 33+ tape and later removing the tape a week later after heat does the work--and even extremely thick (VERY thick) pastes like that super old Radio shack thermal paste, or partially dried or old Arctic Ceramique (NOT dried out, just very hard to squeeze) will also work as barriers too, although this will be messy, and a pain to clean up if you have to redo your LM application. A VERY big pain.