[Alienware 17R4 / 15R3] - Disassembly + Repaste Guide + Results

@iunlock any chance you can describe how you lap the heatsink and which sandpaper you use?

 

It is not a refurbished unit, I got it about a week ago. According to Dell tracking it left production on the 6th of March.

Thank you @iunlock @Pete Light I will open it again next week when I return home, bend the arm and repaste it again. I will take a picture of the arm before and after if that can help anyone.

For the pads I have no ideas, I was very cautious not to touch them and to be sure they stayed in place. I can try to make better pictures of the pads next time I open it.

Do you recommend warming up the arm with an hairdryer or something similar to soften the metal before bending it ? Maybe less chance of breaking it that way ?

 

I did this. I used P1200 and lapped it very gently for 5 mins for both and no more. Check the guide to repaste in my Sig for the instructions and pictures I provided

Sent from my SM-G935F using Tapatalk

 

You didn't use water & soap when rubbing? I never lapped a heatsink before but I definately see the need for it in many cases. Also I see people taking the entire heatsink and rubbing it against sandpaper when working with desktop ones, but this is not a possibility with a laptop heatsink? Lastly, isn't there a chance of actually removing too much on parts of the surface area when moving sandpaper by hand over the surface making it possibly an uneven fit?

 

Yes. For that reason I don't use sand paper to lap it although you could use a very fine grit finisher, but even that can be too much depending on the heat sink. It's tricky because some heat sink units are more forgiving then others. As for the soap and water, those are definitely reserved for desktop heat sinks.

::iunlock::

 

Ye, that's what I expected. I ordered a small piece of 3M 600, 800, 1000, 1200 & 1500 grit sandpaper with a small corkblock that I will cut to a very small piece and use to try to make a flush finish

 

Excellent. Happy Repasting!

::iunlock::

 

So about 2-3 weeks ago I've been noticing higher idle temps. I thought it was from my conductonaut being dried out or from a uneven heatsink, but that was not the case cuz when trying to remove it, it was still very much in liquid metal form (though some felt dry/crusty in a small spot).
So I got some cheap cooler master paste from heatsinks from years ago and was using that to see how the heatsink laid. I used small amounts of paste, cuz obviously too much paste will squeeze out everywhere and not give a good idea of if its truly flat. I did this about 2-3 times before moving to Kryonaut. Core 0 and 2 were about 6C higher so I bent the arm a little bit and tried again, then bent it a lot and tried again, no difference. So then I used the 0.1mm copper shims I had and pretty much cores 1, 2 & 3 are the same or within 1-2C at all times but Core 0 can be as high as 6C higher.

But whatever I'm fine with that. What I am not fine with, is why my idle temps seem much higher than 2-3 weeks ago. Pretty much they would always be below 50C, which meant the fans would never go on. Even when playing League of Legends, I'm pretty sure the fans never went on, but they do now for even browsing the internet if there is a video playing. My max temperatures are pretty much the same as before 2-3 weeks ago, just the idle is higher for no reason? My ambient temp is probably lower nowadays than when I had the lower idle temps too

 

the 0.1mm pad is actually white milky colour

 

You know the heatsink and pipes are shared between Cpu - Gpu in this laptop? Higher temp on graphics can affect the processor temp.

 

Yes. What I am saying is, browsing the internet and even playing light games before, never would turn the fans on due to temps being much lower. Or at least it did it such few times that I thought it never happened. I always remembered my CPU being in the mid 40s, or lower than 50. But now its like high 40s and hitting 50s which causes the fan to increase speed.

I thought since doing my first LM repaste ever that I had done something wrong and it dried up, but even that wasn't the case since cleaning it off and applying Kryonaut resulted in nearly the same exact temperatures and fan speeds. Either way, max temps seemed consistent and I went and used HWinfo to create my own fan profiles. Not exactly a fix but its a workaround.

Edit: I have to do some more testing... randomly, for no reason that I can think of, with a video loaded and typing this, CPU temps are 37, 38, 37, 37 and GPU 37. Fans aren't even running that I can hear (1000RPM set in HWinfo). Something is wrong somewhere with software. I have two 1080p videos running and it hovers around 43-46C on all cores, as it once did. Closing one video results in a near immediate drop to 40-42C with just the one video while typing this. Maybe there is a conflict somewhere with power settings, that's where I was clicking around before I saw the major drop, but I didn't change the power setting

 

Newer BIOS or VBIOS or OC profiles?

Sent from my SM-G935F using Tapatalk

 

None of that. I just did a fresh install of Nvidia and Intel video drivers, gonna see if that helps.

 

If you don't have any bloat or unwanted trash running from Windoze... You should see almost zero power utilization in task manager or in ThrottleStop monitoring in idle mode.

 

I think I figured out the problem, kinda. Running more than one display runs the GPU @ about 30W vs the idle 7.4W. Except, I am pretty sure for the past 2-3 weeks except yesterday I was only using a single display. The whole reason I reinstalled drivers yesterday was so I could turn off the laptop screen while using my 144hz monitor and 4K TV, but it was being problematic before and wasn't possible (I think due to not having Intel GPU drivers installed). So maybe since I would often unplug and re-plug monitors and switch around screens, maybe it got stuck somehow and always thought I was running two or duplicating? Idk.

Edit: so that was definitely the problem. Which raises a few questions... firstly being, I'm not sure how I didn't notice so it must have been a bug with something (i would notice in minutes if my mouse was moving further than the screen) and secondly...

Is it normal for temperatures to drop pretty much 10C for CPU and GPU when going from dual-displays to single display?? I mean I can understand a small increase, but 10C? Anyone else able to test their temperatures connected with one display and then set to extend to another, I am curious if everyone else experiences the same

 

Quick question, is the heatsink of the 6700HQ the same as for the 7820HK ?

As I posted before my replacement unit, a 7820HK GTX1070 is running very hot but I still have my old unit, a 6700 GTX1070 that was running much cooler. What if I swap the heatsink before sending back the old one ?

 

I doubt you'd get a difference since the heatsinks are same and only CPU is different. At what frequency 7820HK is running? Is it stock or OC'ed?

 

I agree, considering all the problems others have had its much more likely that someone had screwed up with the thermal pads/paste as opposed to one just having a better heatsink. Besides, I'd hope they'd include a better heatsink for the X820K's, but they don't, cuz the HQ's can't exactly use it (ofc everything can use a better heatsink, but you know what I mean)

 

The only difference is the 1080, 1070, or 1060.

With the 1080 you get the thickest heatpipe, with 1070 it's slightly smaller, and the 1060 model is smaller than 1070 and has less pipes.

What CPU is inside doesn't change the heatsink, only the GPU model.

 

Here ya guys go...

OP updated.

Enjoy...

 

thanks for the update, but the picture shown here belongs to kaby lake GTX1080 model? im asking because the heatpipes look way too small to be GTX 1080 and the picture looks awfully similar to the skylake GTX1070 u posted earlier. Doesnt look like a mistake an OCD person would do, does it?

 

No this is the same pic and I had just changed the values for now...this HS is for the 1070, so you're correct, good eye, but the information is correct despite the picture.

I'll update the OP and links once I get to it...I'm busy right now in the shop... some people needed this chart tonight so I just threw it together ...

 

@iunlock are the thermal pads still the same for the 15r3 gtx 1070 i7 7820HK, like in this picture? http://i.imgur.com/5XwqvYS.jpg

 

no worries there mate. thanks for taking the time to post the guide. awesome job

 

It's my pleasure, glad to be of help.

 

I had the same question. Thanks for the awesome guide!

 

Yes the 15R3 is the same.

::iunlock::

 

Great my 1mm pads got delayed o.0 Deploying tomorrow.

Any chance I can stack .5mm pads to make 1mm pads? I have 2x 15mmx100mm packs from fuji, not sure if that'll be enough for the new 15r3

 

What kind of gloves are you using / recommend?

Thanks!

 

17R3 user here would like to have a repaste done anyone willing to do it for money?

 

I would wait for the 1mm pads. Stacking pads will degrade the thermal conductivity numbers by about a factor of 2.

Any latex or non latex gloves works fine. No particular brand...just whatever is on sale.

 

My AW 17R4 fan diagnostic test failed so I have managed to convince Dell support to replace and am hoping to get the technician to repaste as well

I realise that Thermal Grizzly Kryonaut is preferred however I could only locally source Thermal Grizzly Hydronaut. Will this be better than the stock paste that Dell techs use?

They have agreed to replace the cpu fan. Is there anything else I should be asking for while they are on-site other than a re-paste using my Grizzly Hydronaut?

I have a AW17R4 with i7-6700HQ and NVIDIA GTX 1060 and am running BIOS v 1.0.8

Are there any thermal pads that I should be asking if they will replace and given my build date is mid Dec 2016 should I also be asking for a replacement heatsink?

 

Yes the Hydronaut will be better than the stock stamp/paste.

Also, the fan showing as failing during that diagnostics is normal. It's strange, but there is a glitch that causes it to read out as failed, when it is not.

Of course if your fan is physically damaged then get it replaced, but I would double check to make sure that your fans are working, because that message is a common bug that displays false info.

 

Managed to snag the last 2 of the 1mm pads on prime yesterday. Looks like everything should arrive in advance or just after my laptop next week. I will be doing some tests before repaste to ensure there are no issues with the unit or anything that would require an RMA from the factory. I also got some mini dividers to organize the screws when I take them out.

If I recall correctly I remember iunlock saying that he compresses his pads with a rolling pin or something prior to installation in order to allow a better mount for the heatsink, is that correct? It makes sense because the pads would decompress after some time but while compressed it would be easier to get good mounting pressure on the HT.

 

Did you get the Fujipoly 11.0's? I would just lay them on and not compress them as they will won't bounce back like regular stock thermal pads.

Good luck with the repaste and looking forward to hearing the results!

Cheers

 

Ah, understood.

Anyone know what bit sizes to use? I am not seeing it on the first page with the other instructions.

Only thing I am doing differently than normal is following this vid:


and not removing the motherboard as I am not using liquid metal.

I do plan to bend the tension arm and to replace the WiFi card with the Intel 8265 as well as remove the HDD for an SSD and install an additional M.2 SSD. I hope to record my work so I can post it up here for others to see more examples.

 

Are you in the US? If so, contact Team LHz
@iunlock @Mobius 1 @DeeX @zergslayer69

 

Here you go:



::iunlock::

 

Thanks, I knew it was in the thread earlier but I could not find it when I was looking lol!

 

Saw the UPS guy as I was driving to the airport, got him to hand over the goods.

2x 1.0mm pads, 2x 0.5mm pads, conductonaut, kryonaut, 33+ tape all in hand. Just landed in Kiev, debating doing it here at my hotel or waiting until I'm at my final station. Don't have an ETA for my pickup so I don't want guys to show up when I;ve got it half apart lol.

I read somewhere about using thermal pad to dam the dies in case of leakage, anyone done that?

Just want to make sure all goes well. The Prolimatech I used a few days ago is at 45C playing some music and typing this, think its just super damn old. Got kryo also just in case I fee bored and want to do it again ha.

Highest current temps got with the PK3 is 81 in ME3. 65 in OCCT. Really makes me think OCCT is a garbage way to measure your temps, although I may have some settings wrong in it.

I'm wanting to get idle temps much lower though.

 

Great stuff.

Just use the right amount of LM and never mind the thermal pad dam stuff...that can and often does more harm than any potential good.

Also use wPrime v1.55, change your cores to 8 (advanced settings) and run the 1024M bench.

Configure your hwinfo64 to show all the relevant info like the core temps, package power etc...

Good luck!

 

The thermal pad dam has not yet been attempted by anyone. The general consensus is that it would reduce pressure in such a way that would result in the very thing it is trying to prevent. In addition it would remove any benefit from using liquid metal TIM as a result of the effects of decreased mounting pressure. Thus the suggestion or using a traditional TIM as a dam around the liquid metal was suggested but was also shot down and has yet to be attempted by anyone due to worry of if such a thing would or could ever even work and how the different TIMs would react. Lastly there then was the idea that I came up with of what if you just used liquid metal on the heatsink to fill in the micro scratches and wipe access off and then just apply traditional TIM. This would in theory be no different than if you had applied liquid metal to start with, removed it (well you can never fully remove it from the heatsink to my knowledge) and then applied traditional TIM. However, this idea was also shot down under the idea that you would not see any real benefit to doing this and even if you did it would not justify the cost to use liquid metal in that way and or that it would only work if the liquid metal could cure of soak into the HT before being wiped off or something like that... (maybe bake it in an oven after application?). As a result no one has attempted that idea either yet and I do not plan to try when I do my repaste next week.

 

@iunlock I am going to repaste my Alienware 15 R3 with liquid metal and I was wondering: you have to take off the electrical tape after you apply the liquid metal and then put the heatsink back on or you leave the electrical tape on the computer?

 

You leave the tape on, it's insurance in case the LM leaks out.

 

Repaste done. Took my time, 2 hours maybe. Been running OCCT and playing overwatch for two hours or so, max temps 68-71-68-70 gpu 66. Seems about right. About 5-10c lower when I;m just browsing or listening to music too.

 

I've tested all those methods mentioned and for the obvious reasons that you've stated, is the reason I don't recommend it. Counter productive. This is why it is not included in my repaste guide.

Leave on the tape. It is a free insurance policy in the event of any leakage. It's important to put the right amount.

Excellent! Great job.

Can you configure your hwinfo64 to show the important info like all four cores, package Temps etc... And run wprime v1.55 (1024M) run and post up some screen shots?



::iunlock::

 

How do I contact sorry I cant seem to send a pm or even see how to do so.

 

Youy need post minimum 5 posts for the opportunity to send PM

 

This is with 4.1 on the cores, -.15 voltage.

1351 on XTU Bench. Not sure how you got above 1400 iUnlock lol. Guessing better bios in those ASUS rigs.

Running a ton fo crap in background also

 

Looks great! Good job. Conductonaut?