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

I a re-paste with Thermal Grizzly Kryonaut the other night. I have had some interesting results.

First off, I think the repaste seemed to help at least a little bit. The CPU and GPU hit 85C and 82C, respectively on an AIDA64 cpu and GPU stress test after 45 mins. These number are down from the mid to upper 90Cs before the repaste.

But here is what is odd, I would think that a stress test like AIDA64 would be enough to tell me how hot it would get in the most extreme settings (no rear elevation, no undervolt, ect). However when I play Overwatch, after about 10 mins the temps have spiked dramatically beyond what any stress testing will do. The temps are about 95 for CPU and (a somewhat comfortable) 86 for gpu.
These temps are helped very little with a -0.140 mv undervolt.

Here is the next odd thing, when I raise the rear end about a half inch or so, it massively reduces the temps. I played Overwatch for 3 hours the other night and the CPU never went above 75C and the GPU stayed at about 68C. The idle temps are also very impressive sitting at 40-42C between the core (there is about 4-5C diffs between the cores most of the time). This is also with little fan speed.

I of course, expected it to run cooler with better airflow; any computer would, but what I don't understand is why it is so drastic. I guess I expected it to maybe be a little better lifted in the rear but not cooler by 15-20C.

What could be causing this? Clearly airflow is a big part, but in my mind it should be able to run at least mid-upper 80sC when level and not have these drastic 95C+ temps. Is this a problem with my paste job? Wouldn't I see that with the idle temps then?

TLDR: Why is the temp difference so massive when I lift the back end?

The picture attached is of the temps Idling after an OCCT stress with a -0.140 Mv undervolt.

 

No, not normal. Perhaps check your Heatsink fit again, you should get much much better temps with LM.

 

I dont know how to post image but after OpenGL the result are 121.54fps and the max temps are 68. CPU result is 848 cb and max temp at 75. CPU clocked at 4.0

 

Airflow!!! can't cool without airflow!!! You can fix this by buying elevated rubber pads to raise the laptop off the bottom. Bad design by DELL CANCER!

 

Thanks for the reply. I agree, it clearly lacks adequate airflow. Do you know where I could buy these rubber pads that I could put on the bottom of my computer?

 

Search on google, amazon or ebay? I don't know. This is something you can find yourself, Sorry I can't help.

 

It's called Cool Feet . Or just use a couple of Coke bottle caps.

 

@JRBruett try these

https://www.amazon.com/dp/B01M8I6YLH/ref=pd_luc_rh_sspa_dk_huc_pt_sub_2?psc=1

 

Thank you. just ordered some.
I actually did a repaste with Conductonaut last night and It helped overall, but if I overclock to 4.0, it still goes up to 95C while sitting level and will sit 80C while elevated in the back while playing Overwatch. It also sits in the low 40s while idling. I don't know how iunlock is getting such low temps while overclocking, while playing over watch, WHILE HIS COMPUTER SITS LEVEL. This would seem like just a design flaw that no matter what cooling compound you have, if it isn't level, it can't keep itself cool. But clearly this must be a me thing.
That's kind of why I went full on LM to try and see if that would help. Admittedly, it gave me the ability to overclock and better temps in general, but I still wish I didn't have to have the thing elevated in the back to do that, but why have the risks of LM if it doesn't do anything for me? \
If anyone has any ideas/insight or similar experiences while using LM, please let me know.

Pic is after an hour of Overwatch.

 

Your temps are going to be 10c+ higher when the back is not elevated. The rubber feet just are not high enough to give enough airflow. Flat on desk is fine for light gaming and media tasks... but I would definitely not game without the back elevated, it's a design flaw.

I'm pretty sure everyone's low temps are from having the back elevated.

 

Errmmm I game always on a flat desk and my temps are fine. around 80cfor cpu and 65~70c for GPU and no I dont pop up the back for more airflow.

 

It would be lower if you prop the back. And his system is a 7820hk with gtx 1080 which runs significantly hotter... it's not going to be hitting a max of 80c and 70c flat on desk after an hour unless you are limiting the FPS and settings.

 

I also have the smaller 15inch unit with a smaller heatsink. I saw plenty HK cpu variants run cool. Every laptop runs better when propped up

 

I mean you can just look around this board to see flat on desk with 7820hk + gtx 1080 is not going to get you sub 80c/70c .There's multiple accounts of people posting their differences of flat on desk and raised. I dunno why you seem to be insisting it's possible when your system isn't the 17in with these specs with significantly higher TDP..

 

Because i tested it on multiple machines pf my friends. Again all laptops with bottom vents show an improvement when lifted up.

But a good repaste (not need LM) and a good heatsink fit will keep your temps cool even on a flat desk. I have repasted multiple 17inch versions of friends and they all use it on a flat desk. 80c on kryonaut is pretty good so why do you need it even lower?

 

I repasted with kryonaut and changed pads, but I didn't get any improvement overall (maybe I simply had a relatively decent unit to begin with).

My gpu is satisfying, stays around 70 degrees under heavy load, never gets hotter than 80ish.

But CPU, while having reasonable averages, has big differentials (core 0 and 2 average around 85, 1 and 3 around 75). The hot ones get to 96 max temp and also got thermal throttle.

I tried bending the arm but nothing changed. Is there anything I could look for/do about thermal pads?

 

I recommend returning the machine

 

I'm sorry?
Maybe I should have specified I was talking about overclocking and with stress testing (occt and furmark), not idle temps.

The core temps difference is bothering, but everything else seems fine to me, especially considering the average r4 user experience here.

Am I missing something?

 

Everything seems fine? 96C and thermal throttling? Nope.

 

No matter how much effort you put into the alienware, it'll always have bad temp problem.

The lack of 1 mounting arm and lack of strength on the 3rd mounting arm will cripple your system for its lifetime.

There's no fix for this. Only option is to return and get a different brand.

 

You're in the Alienware board, in the repast thread, what's the point in a comment like yours?

For the record I got an Alienware because it was the cheapest and easiest to get option in the country I live in. Buying from overseas just increases costs and puts you in possible trouble and longer wait times in case you need to use your warranty.

Not everyone has the same possibilities, not everyone lives in the US.

 

Okay these came in the mail the other day and WOW. Thank you, Thank you, THANK YOU. I realize this was a very simple fix but it seems that it was just enough to height to give it the airflow it needs. It never loads past 80C on an 4.2ghz overclock on the CPU and never past 73C on a 200/250 overclock in GPU.
I played Overwatch for two hours last night and the temps mentioned earlier are what it settled on.
I still wish that all the pageantry around the LM repaste would have given me better temps (like never going in the 80s on CPU) than just this, but I'll take it.

Also you don't at all notice the rubber feet when moving the computer, or using it. It is almost as if Alienware put the feet on, and quite frankly, they really should have.

So thanks again for finding these, they are perfect!

 

welp

 

AW has slight problem in core difference temps. Did you lap your heatsink before putting the paste?

 

No, I haven't.
And it's something I'm not sure I'd feel comfortable doing..

I noticed that most of the thermal pads don't get any kind of mark from the various components they should be touching, but at the same time adding another 0.5 layer would feel too much.
I'm not really sure about what I should actually do and look for when trying to get the right pads thickness..

 

You asked for help. Bro @Mobius 1 gave you his best advice. And he have owned this model, as well re-pasted many of them for people who can't handle this overheating mess by themself.

Try K5 Pro.

 

If you live in a 3rd world country with people stupid enough to believe the Alienware is best gaming machine, you can sell it for a profit even.

That's the great thing about Alienwares.

I bring 2 home and I cover 1/2 of my plane ticket from the profit of selling it to people dumb enough to buy it.

Spoiler

 

Thanks, I'll give it a thought

 

Didnt

After a repaste the machines run fine.

And pretty much most machines run with the same temps put of the box. Except msi but msi their chassis is always piss poor around it.

Saw asus machines out of the box doing 90c, HP, Aorus, Gigabyte. So whats left?

 

death

 

Maybe I can make a huge profit too, Hahaa.. But, I am afraid the machine will perform badly if I give it away!

 

How do they look and where did you place them? Post some pictures?

 

Just curious a friend of my had some cooling troubles with his alienware. when he said that it was running hot but not when placed on his laptop stand I checked his Alienware.

When I check the bottom panel it was bowing slightly outward, closing the gap between the table & laptop. Especially in the middle of the panel was close to touching the table. Check if that is the case with yours as well. The panel should sit flush ofcourse but in this case it didnt.

 

So his bottom cover wasn't on all the way? How did you guys fix it? Did his laptop act like yours afterward? I'll go look at my laptop. I always game with a cooler though. Edit - I looked at the bottom and on the front pads its sitting on the outer edges of the pads

 

No we couldnt. For some reason it was just slightly bowing outwards because of the pressure when mounted. But ordering a bottom panel is quite expensive.

 

Oh wow. Well that sucks. So you have zero difference when playing on a table?

 

Probably didn't push down in the center areas where there are 3 clips.

 

It was correctly assembled. Triple checked

 

Not zero. But not that huge of a difference. About 5c.

 

Hi there, I have AW17 R4 with 7820HK and GTX 1080 and from the first day I am facing the 1 second lag issue. My shipping date is before December 8, 2017. After reading this form I realized the problem with the machine I have and now its time for re-paste. The problem is I live in Asia ( Pakistan ) where things are not easily available. The real problem is 0.1 mm thermal tape is not available here. can I use 0.1 mm copper shim cut with exact same dimension as that of the sink inner just below the CPU ?

Thank you.

 

After many tries (had to open it six times...) and different approaches, I finally managed to get good temps and differentials.
I have an AW17 R4 with 7820HK and GTX 1080 (bought in February, in Japan).

Part of the issue was my total lack of experience, part was me rushing a few times, because I was getting frustrated.

First I tried to repaste with Kryonaut, and changed thermal pads using Arctic.
I improved things (actually, my GPU was decent at around 75° with furmark), but CPU was still very bad, 96° on core 0, thermal throttling, and 10° differentials..

So I bought Condoctonaut LM, together with K5 Pro (suggested to me in this thread).
Again, things got a little better. Almost no difference on the GPU, CPU got cooler in general, but still too hot, and still with big differentials.

I got rid of K5 Pro. I probably simply sucked at using it, but I couldn't find much info around.
For the record, cleaning it was a royal pain in the ass. While it's not like normal thermal paste, it's not like it doesn't stick.

I got back to Arctic, with LM.
Now things improved again, I wouldn't throttle anymore, but core 0 was always way too high compared to the others (core 2's differential was fixed tho).
Also, even if temps were decent, they were still too high, especially compared to what other people were getting using LM.

Then I opened it AGAIN, got rid of everything, cleaned up LM (in case you need to do it, I used a wet cotton swab, just water. It will make the LM stick to it. Had to use many because they would get saturated).
Then I lapped the heatsink. I used 1200, 1500 and 2000 grit.
Applied LM again. For some reason, I couldn't use the excess LM from the CPU on its heatsink (as suggested by iunlock), it wouldn't stick. I put LM directly on the heatsink, but after spreading it out, it looked like it was way too much. I cleaned it again with a wet cotton swab, and now I could take some LM from the CPU and spread it nicely on the heatsink.
Another change I did this time was using 1mm pad on the CPU side (the thin one that looks like stairs) instead of 1.5. I noticed that the mark left there was a bit too prominent.

And then, finally, good results.
GPU doesn't get hotter than 70, overclocked and with different tests (furmark, heaven and superposition).

CPU has now decent differentials (within 5 degrees), max temp under 80, overclocked at 4.2GHz, and average temps between 70 and 75, using OCCT.
With benchmarks and gaming, everything doesn't reach 70 degrees.

To sum it up, especially for noobs like me, how I solved the issue was:
-Use LM. You can live without it, but it will give at least 10 more degrees compared to normal paste.
-Lap the heatsink. I was afraid I'd make a mess, but there is honestly not much of a mess you can make.
-Be sure pads don't get marks that are too evident.
-Avoid K5 Pro. It may have its uses, but my experience wasn't good. And it's a pain to clean up.

Bonus thing:
I had a small aluminum heatsink with adhesive laying around. It's a few mm shorter than the PCH's length, and 4mm tall.
Just using it (no paste, no LM, and didn't have to use tape and glue) I get 36° on idle, and 60° when stress testing.

 

Thank you, very much brother Genryuu. I want to ask something of vital importance.

I am a bit confused over the right selection of thermal pads. I think there are three ways to do it. Kindly tell me which one is the best:
1) just do not remove stock AW pads except the 0.5 mm ( like stairs covering chips above cpu ), changing them from 0.5 mm to 1 mm
2) remove all AW stock pads and use new Fujipoly Thermal pads as mentioned at the iunlock page, changing 0.5 mm ( like stairs covering chips above cpu ) from 0.5 mm to 1 mm.
3) remove all AW stock pads and use some other thermal pads ( like ARCTIC ) which retain their shapes as soon as the load of sink removes, since I read that Fujipoly if pressed do not return to their previous shape. If other than which ones? Please mention

kindly tell me that the thicknesses of all the thermal pads are correct as shown at iunlock page, especially, the one which is above GPU( 0.5 mm ). Because what I think that the height of all the ICs. around the GPU is same, then why does he used 0.5 mm above except 1 mm?

Please tell me how can I get the the heat sink and adhesive you used on your PCHs. Lets say I get the sink from some where, from where would I get that adhesive. Is their any name of this adhesive?

One more thing, Is their a way to repaste with just removing a cover on the mother board ( internal plastic cover ). Actually I do not want to open up this whole computer.

Thank you.

 

For thermal pads' thickness, it's case by case.
You should follow iunlock's guide, and then adjust in case you see you're not getting good results.
In my case the one that was off compared to iunlock's guide was the "stairs" one, but in your case it could be different.

But yeah, you should probably just replace all pads, and then work from there.
Fujipoly are really expensive and not reusable, that's why I followed someone's suggestion and went with Arctic.

As for the heatsink, I had it around. I had bought it on amazon for really cheap, and it came with the adhesive already on its bottom.
If you can't find anything like it, just go with iunlock's method, which is just a little more time consuming (but allows you to use thermal paste, so it should give better results).

If you have trouble finding stuff in your country, I guess you could buy everything from amazon.com making a single order.
I had to do the same because most of the stuff I used cannot be found in Japan at reasonable prices, and overall shipping wasn't that expensive.

I read that you can remove the heatsink if you remove the small screws on the fans. But people do not recommend this because it makes it difficult to make pressure on the screws when putting it back together.
To be honest, when you got to the motherboard, removing it is just a matter of a few more minutes.

 

With the "stairs" do you mean the vrms on top of the cpu? Those need to be thinner for an even core temp balance. They use like 0.1mm sheets for that now.

 

Sorry, had to be more specific I guess.
I meant the 1.5mm ones.
I didn't replace the ones above as suggested by iunlock.
These ones were supposed to be 1.5, but in my case making them 1.0 did the trick.

(I didn't manage to upload an image or link from imgur lol, what am I doing wrong? ahah)

 

Image doesnt work.

 

Hey guys, it's great to see all the discussion and the positive results people are having.

As for the thermal pads, as some of you have reiterated from what I've stated in the OP, those definitely are unique in size to each system as the tolerances do vary.

The best thing to do is to eye ball the stock thermal pads once you take off the heat sink, then note the ones that have imprints and ones that do not. From there adjust the thickness of the pads accordingly.

K5 Pro is tricky... at first it may seem like a great idea, but the integrity of K5 decreases at a very fast rate and you'll end up having to replace that often. I've done many tests in the very beginning with all sorts of stuff and the reason I never published K5 was due to it's weak performance. On paper and in thought it sounds great, but thermal pads are the way to go. As always, to each his own.

Hope all is well ..it's great to see a lot of new faces along with the old.

 

I didn't do the best job putting them on, as you can see. I will probably redo it sometime to make it look more even. Having that said, they are small enough and out of the way enough, you really don't notice, it is a little harder if you want to slide it on a desk (friction of rubber), but I think these feet are made for furniture which could weigh 200-500 lbs. They really do a fantastic job of providing better airflow to the computer.

In the picture, the red circles are the pads that actually touch the table, and the blue do not. The blues are there to basically provide a little more support if I unintentionally tilt the computer. the main heroes in this set up are the two by the grill. You can really tell a huge difference in physical airflow from the CPU vent on the side with these. I have a anemometer that I use for work I did a test before and after the rubber feet, the speed of air coming out of the CPU vent was 27% higher with the rubber feet than without, so clearly this makes a difference.

Here are some pictures of the feet, and a ~10 min OCCT stress test.

 

Images are still broken. Please upload on sites like Imgur and then paste the direct link