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

It's not to say that it doesn't work, but it's not able to achieve its maximum thermal dissipation, unless the amount of thermals was under the W/mk's limit.

Also, not all chips heat up to the same temp, so there could be many variables going on ...

I remember back when the 15R3 came out, where I had stacked smaller pads on purpose to test it out and when redoing the test with a single pad (correct size) it ran 5-8C (avg) cooler, so maybe try swapping out your pads to a single pad, because there may be more room there for you. - Of course there are limits and uncontrollable factors with high ambient temps etc... but it's worth a shot.

Hey good job on taking the precautions to keep things clean haha...

As for trimming the pads, I make sure that the pads have the protective plastic on both sides, then I take a spacer (you can even use a plastic old credit card and cut a block out of it the size of the thermal pad. Then use use two solid objects to hold the pads in place (so that the pad is in between the two solid blocks) and place the card/guide (with the cut out) over the pads. Then using a razer use the old credit card as a guide to slice down.

* The CC may be too thick so if the pad is thin, then adjust the guide/card accordingly of course... You could even free hand if you have robot like stability.

 

I still use 15 r2 and arctic pads. With Fuji, I might not even stack it because of high W/mK and its too brittle for me.
I thought you used Razer Blade laptop when you actually used a Razor blade. A typo can give different meaning.

 

Have to take photos when im back in NL. Out of rubbing alcohol to clean the heatsink for another repaste.

But i tried both 0.5mm and 1mm pads to check for imprints. 1mm is fine for the gpu memory chips but no gpu contact (only the center barely touches the thermal paste).

 

Very difficult situation which i never came across with repasting other alienwares, my own past machines with multiple brands. The margin is just too thin. 1mm too thick and the heatsink cant give enough pressure and 0.5mm is not enough. Have to say that the contact with stock pads wasnt the best either though for the memory chips, light imprint as well.

No high wmk 0.75mm thermal pads on the market. Might just get 0.1mm copper shims. And place those on the heatsink with some good thermal paste and place the thermal pads after for the perfect fit. Having ocd is a curse (actual medical ocd). My temps are as food as they are going to get with paste though. 2c core temp difference now and with an ambient temp of 25c the gaming temps on cpu never exceed 85c In far cry 5 ,80c in bf1 and R6 siege doesnt even go over 75c.

 

I just compressed the 1mm thermal pad with 2 flat objects. Until the right thickness is met.

 

Hmm should try that too. but I worry that the pads are a bit too elastic for that. arctic pads fairly quickly regain shape after no pressure. even though the shore hardness is quite soft.

 

Stretch the pads. Have been done and showed in one of the Clevo threads. Works wonders. Try it bruh
Look after one of @tanzmeister ’s Clevo threads.

Edit. I think it was this one http://forum.notebookreview.com/thr...71dm-p775dm-thermal-paste-replacement.789825/

 

I use gelid and they're shore is 35. I have compressed them too much actually and had to start over again with a new pad.

Has anyone put thermal paste with the thermal pads?

 

After running stress test for about 20min, I notice the RING: Max VR Voltage, ICCmax, PL4 is flagged as YES on Maximum. Does anyone know what this reading mean?

 

@rinneh Gelids once compressed (before use) to specific thickness (specially achieving good fit for memory chips) do not come back, but still give after (specially when under screwed heat-sink and heated up), therefore for your situation this would be a perfect choice. For example, using pressure you can make a 1mm Gelid pad easy to 0.5 mm and it will stay 0.5 mm.
For fun, I had a left-over to throw out from Gelid pad and could make a solid ball out of it, just like chewing gum.

1mm should be perfect for you to get 0.75 mm.

@propeldragon typo lol, Gelids are SHORE 0035 (Arctics are 0025). You had huge success with Gelids!!!

 

First of all, thanks everyone for your views and opinions on this, next to the advice ofcourse!

Going to try it out. Have never encountered margins like this, even not when I repasted AW models of friends. But those where pre-revision heatsinks which had the exact measurements as provided by @iunlock

If I have the thermal pad contact with the GPU memory chips sorted I have the perfectly balanced heatsink with good temps.

Yeah compared to the arctics the Gelids kinda seemsto crumble a bit and seem to be more clay like? Have to reorder some 1mm pads for that. Only used 0.5mm pads on parts needed from Gelid and they perform excelent. Highly prefer them over the better specced Grizzly minus pads.

Yeah going to try the 1mm pads out from Gelid for the GPU. Used the 0.5mm pads for some VRM's. They do their job well.

 

Has anyone tried this?

 

What is it, modified AW17 R4 heatsink?

 

yes

 

What are the modifications? I only see a slightly different top arm. Is it also available for a 15 inch model?

 

What does this "SHORE" thing mean, and how or why does that make Gelid better than Arctic?
You can get giant 145x145mm arctic pads.

I mean come on, if you're not rich, would you rather buy something this tiny like this:

https://www.amazon.com/Gelid-GP-Extreme-TP-GP01-B-1-0-Thermal/dp/B01AAHM1WW/
(that's 45mm x 78mm for $10 dollars)
Or This

https://www.amazon.com/ARCTIC-Thermal-Efficient-Conductivity-Handling/dp/B00UYTTXSM/
(145mmx145mm for $15).

And the 0.5mm arctic pads are only $10.

 

They (Gelids) are 12 W/mk vs Arctic 6 W/mk and they simply work even on places where elastic pads failed (Gelids vs. Dell default elastic pads above processor) = with 0.5 mm Gelids I get no more core temperature differencials vs. 0.1 mm Dell pads.

You definitely get more Arctic than Gelid for 10 EUR, lucky people who are happy with Arctics.
Now with shore the lower the better (for laptops) for proper heat-sink vs. dye contact, especially with LM. In this case Gelids are harder then Arctic but they do not push heat-sink back as much as elastic pads. This is good for my memory chips around GPU where the area to be pressed down is large.

I had really hard time to get proper contact with VRMs,memory chips and chokes AND at the same time LM dyes and heat-sink.

EDIT (saw rest of your post): I agree, but elastic pads do not work on my laptop.

 

You have to try everything if the headache is awful as a TRIPOD. MONEY aka the price should come last.

 

Replacing stock thermal pad with K5-Pro paste

One of the biggest problem that I have while repasting + repadding the Alienware 17R4 (i7 7820hk GTX 1080) is finding the perfect thermal pad thickness. Using the wrong thickness will cause poor contact on either the VR on the CPU/GPU. This problem is worst when you try to repaste with Liquid Metal since the paste layer needs to be very thin to be effective. So I find this on Amazon https://www.amazon.ca/PRO-viscous-thermal-paste-replacement/dp/B018XR6TOM/ref=sr_1_3?ie=UTF8&qid=153... . Apparently, it's a kind of paste that claim to be able to replace thermal pads (up to 3 mm thick). And since it's a paste, you don't have to worry about getting the correct thickness, just slap a bunch of them on the motherboard and the mounting pressure from the heatsink will create the perfect thickness for you( or at least that what they advertised). So I just want to ask if this is a good idea? Have anyone tried this?

If not then I'm still looking a good thermal pad to replace the stock thermal pad, something that is easy to use and can be reused (I'm think of fujipoly but they are extremely expensive and if the repaste job fail, I'm afraid that I have to replace them every time I remove the heatsink).

Also, I would like to ask how important are the thermal pads really? Do they make a huge different in CPU/GPU temperatures? Because so far from what I understand, they are used to provide heat transfer for the VR, which I don't think have anything to do with CPU/GPU temperatures. So unless you are doing some crazy overclocking then your VR should never overheat right?

In the end, I'm actually wondering if Alienware laptops are even worth buying anymore. I've spent a lot of time and money just to try to fix this one thermal issue. Owning an Alienware has always been a dream to me but I'm starting to regret it. I guess I'm just gonna move on, get rid of this laptop and buy something else this time. Does anyone have a good recommendation on what brand to buy? I'm thinking MSI, Sager or Clevo, just looking for something that works out of the box or at least easy to fix/modify. Thank you.

 

Don't use this... Really) You never clear it from the motherboard.
If you can not choose the thickness of the thermal pads use Laird TFlex 3xx series.
https://item.taobao.com/item.htm?id=547906263262

 

Clevo P870Tm with single or dual graphics. From Prema partner. F.eks HIDevolution.

Too hot VRM’s will push stuttering and freezes.

Edit. And you haven’t temperature monitoring sensors on those components. Only the Graphics cores.

 

That was a joke right? LOL ...(About the razer/razor blade...)

I'm pretty sure given the context, that people here are smarter than to mistaken that for using two laptops.

I'm speechless if this wasn't a joke.

Precisely, which is an example that you got what I had meant... Glad you got it to work.

Tip: Definitely try different things until you get it to your liking, but do keep in mind that the shore rating of something soft doesn't mean that they all share the same max, "squish," point... a lot of the thermal pads reach the max peak at different limits when you compress them.

I don't recommend using K5 as it'll just dry up, is not very good thermal conductivity wise and will start flaking.

This has been tried and tested many times over.

 

I agree 100% with @Papusan on this. If you get single 1080, have them provide the vapor chamber for dual 1080.

I try to live by the Golden Rule. The Clevo P870 (with @Prema BIOS) is the only high performance laptop that I can recommend, as it is the only one I even remotely entertain the notion of purchasing for myself. I am simply not interested in buying broken garbage that cannot be everything I require it to be. Nor am I content with paying a lot for something new that costs an absolutely ludicrous amount to fix if the mobo dies outside of warranty. I'm fine with BGA CPU and GPU on a cheap (under $500) notebook like a low end Inspiron, a cheap Acer or even a Chromebook, because I can throw it in the trash and buy another one when it craps out before I am done using it.

The sad part is, everything that comes off an assembly line is a pile of trash now. Even the better products have to be fixed before they work well. HIDevolution takes care of that nonsense for you, so you don't have to deal with it. Doesn't matter what you buy from them (Alienware, Clevo, MSI, etc.) it's going to be tested and the broken crap fixed before it is sent your direction. They actually care about their customers and the stuff they buy.

 

TBH, I wasn't joking. My dad thought Razer(laptop) was new version of Gillette Razor when I actually told its a laptop brand.
Razer 2017 were POS, now they've made amends to cooling system which are far better.[/QUOTE]

 

@iunlock What are your thoughts on using thermal paste on thermal pads? Have you done any testing? I know EK tells you to have thermal paste on both side of the thermal pads when applying a water block.

 

highly depends the thermal pads, For example arctics will absorb silicone based paste in its pores.

 

I guess I had arctic pads and gelid pads on my mind. So the thermal pads would improve with paste? I would think anyway. I wonder if it would be worth the mess it would make.

 

Hello, new Alienware 17 owner here. Any thoughts if liquid metal is an acceptable solution, if I'm occasionally travelling with the laptop.

 

If it's not recommended, what kind of thermal compound would be best to lower temperatures?

 

Phobya nanogrease extreme is the best paste I have used. It's what I use. LM is up to you. If you take the right precautions it's safe.

 

Hi i'd like to ask if which thermal pads is best for AW17 R4 7820HK GTX 1080? I'm planning to repaste this soon with Kryonaut but i'm still stuck at which thermal pad is best for this unit (Can't decide from Fujipoly, Arctic or Gelid). I'm having stutters during gameplay with only 85-87c on CPU temp and 80-82c on GPU with stock thermal paste and undervolted to -120 this is why i wanted to change it with best quality thermal pads to fix this issue.

And I'd like to ask if this site is still working FrozenCPU? Thanks.

 

New Alienware owner here (i9 8950HK, GTX1080, 32GB RAM, 1TB m2 SSD, 4k GSync).
As soon as I've enabled the performance mode in BIOS temperatures went to 100°C on the CPU, the fans were screaming and the CPU throttled like hell. GPU under "Heaven" loop reached 80°C as well.

Started with tuning the voltages on both, CPU and GPU which worked quite good. CPU is ended up fine with -142mv under all conditions (-150 wasn't stable anymore). This brought the temperatures down to 87°C (max) on full load and it was able to permanently hold the 3,8GHz which I've set it to. Also tweaked the GPU to use max. 0.925V at 1850MHz which neither increased nor decreased the temperature but greatly improved performance.

But this thing was still very loud and temperatures are not considered "good" in my opinion. So I've followed the instructions for repasting this beast and decided to go for LM + repadding.

After that the temperatures went down quite significant with the same conditions (about 24°C ambient, no lifted back, no cooling-pad, fan-profile "balanced"). After about an hour of load as a first test I ended up with that:
CPU 3,8GHz max 74°C (-10°C and WAY more silent)
GPU 0.925mv 1850MHz max. 68°C (-13°C and WAY more silent)

I'm pretty happy with the results. Will test more today. Also got my Cooling-Pad. Silence is more important to me than 500MHz more on the CPU that I don't need yet.

 

After taking advice that the temps I was getting after a second repaste/repad was far too high I think I finally have got it right on the third try!

This time I used Cooler Master MasterGel Maker instead of IC Diamond 7 - no idea how old my IC was. Mostly the pads are now 1mm and 0.5mms. There's a couple of places where the gap was larger so they were stacked to make up 1.5mm.

I also took the time to change over to some larger head and stronger screws. I was finding that the screws were so soft that it was going to get to a point that I couldn't unscrew anymore.

I swapped out the Killer WiFi to the latest Intel 9260. https://www.ebay.co.uk/itm/Intel-Du...-9260NGW-BT-5-0-Card-01AX769-W25/292462114421

The temps in the picture are Idle/browsing for about an hour and the hot temps are from 2 hours of FarCry 5 at Ultra everywhere.

 

Speedfan readings are good. Use HWINFO instead of Speedfan. With HWINFO you can control the fans as well.
Which card you swapped out? killer 1550? If its 1550 then the card is same as intel 9260 with change in brand name. These cards have worse signal quality and range compared to 1535 acc. to Phoenix.

 

Unfortunately mine had the 1435 in it - I think it was built/assembled in September 2017.

 

Found something that you guys might be interested in https://www.dell.com/community/Alienware-Laptops/AW-17-R4-POOR-PERFORMANCE/td-p/6108601
So according to a Dell representative, it's normal for an Alienware with GTX 1070 to have those in game FPS (1080p all low on Overwatch average around 120fps on a 1070 is normal?????).

 

I have about 90~100fps on average with a 1070 on FHD with everything on Epic, on low it goes way over 160.

 

Yeah I have the one with GTX 1080 and on EPIC with QHD I got around 100fps, so it's not normal at all.... I'm surprise by the Dell Representative response saying that low fps is normal for Alienware machine, like WTF???? Also I just realize something, you don't have to disconnect the fan when doing the repaste, you just have to unscrew the 6 tinny screws on the fan and lift the heatsink up which save A LOT OF TIME and effort repasting this laptop. Also you dont need to remove the wifi card @iunlock should really consider updating his guide.

 

Anyone can help me which thermal pad is best used for stuttering issues? Thanks.

 

O chose Gelid and will stay with it until something MUCH better comes out. I like how Gelid keeps its form, they are soft and high 12 W/mK. Since I put them in my laptop I have no CPU core differences, and they adjust good to uneven height of components to be cooled.
I have experience with stock and Alphacool (about the same as Fujipoly, maybe re-brand).

Both pushed my heat-sink too high over dyes and I had too high GPU temperature or CPU overheating/uneven core temperatures differences.

 

I second this

I put gelid on 1, Thermal grizzly minus pads on 2, and Arctics on 3. I dont like Fujipolyu's they are just too dry and dense.

 

Finally had the time to register my findings.

After repasting multiple machines of my friends I had some trouble with my own machine. Their machines where early Skylake & early kaby lake models. Mine was an early 2018 kaby lake model.

From day one my core temp differences weren't bad but after repasting they became larger and after a long searching, test with different pastes I came to the conclusion that the stock pads where too thick on some places (not compressing enough).

My action after this was trying different pads and I settled on Gelid pads, why? Arctics even though really nice and I prefer them over Fujipoly's even though the heat transfer rating of Fujipoly's are better they were easier to work with. But the heatsink of this AW is quite finicky. 0.5mm pads on the memory chips weren't enough and 1.0mm Arctic pads are too thick. The heatsink cant generate enough pressure to compress them more so I settled for Gelid pads.

These are the pad thicknesses that provide the right pressure and contact for all components.



So what are my temps?

These are the highest temps after a full timespy run on both CPU & GPU. CPU reaches a max of 74c and the GPU a max of 68c. Ambient temps of 26c.





Conclusion:

It takes some work but the cooling can be good if you put in some work, no need for liquid metal for the non HK CPU's. Just make sure you have the right contact on the CPU and GPU die's and your pads are the correct size for each component.

 

Are the gelid pads any softer than the arctic?

The gelids are pretty expensive, practically fujipoly level (I've grown to hate the fuji pads too).

 

I paid 15 euro for 2 packs of 1mm.

They are on paper a little bit harder than arctics. But get into shape easier.

 

can i use that same pads thickness with the R5 i9 ?

 

No, i havent tested the R5. This is only for the later kaby lake 15R3 heatsink revisions.

 

Even if it was the same computer your heatsink will/would sit differently, resulting in different thicknesses. This is why a lot of people have problems.

 

when I disassembled mine about 1 month ago, I added a zigzag line of thermal paste on the factory ram pads .. I just disassembled my laptop to check on the LM .. and guess what !! the zigzag line was not touched at all by the heatsink !! I don't know what Alienware is doing, but they really messed up

 

First I would like to thank Iunlock for the repaste guide. I now have LM, and Fujipoly applied to the Alienware 17 R4. I also undervolted the CPU to -.135. When playing games, the max GPU temp is now 58c. My fans are still loud. I thought with such low temps, the fans would be running quieter. Is there a way to adjust the fan speeds?

 

How is your cpu temp?