How to FIX AW17R4/15R3/13R3 CPU Core Temperature Differential Issue

Makes sense. I suppose I could have gotten myself some pressure paper and tweaked the heat pipes to perfect the differentials. I watched the YouTube video you did with OWNorDISOWN, by the way. Thanks for the good info!

 

You actually dont need fancy pressure paper for this. To perform a "budget" pressure test, just apply traditional paste, screw the heatsink onto the motherboard, let it sit for 5-10min, remove the heatsink and viola! the uneven pressure is clearly revealed on both the CPU and heatsink.

and thanks for watching. it's a info sharing video which we hope can help owners with stuttering issues

 

What is the fundamental solution to this problem?

 

Thanks! Will give it a shot on my wife's laptop.

 

Your CPU still has temperature differences. (core0)

 

I'm still doing some test on my end, but the fundamental solution is to adjust the heatsink and i'm afraid this is something not readily shared. there are people making a living fixing this issue.

 

I think dell is responsible for this problem.

But they have no opinion on this matter.

 

his core differential is in acceptable range after his repaste with LM. it remains to be seen how long and well will this hold especially LM is known to dry up over time and we are not sure if the gap issue is really resolved. but Sunon heatsink has been known to be a lot better than CCI in terms of manufacturing tolerances.

 

for dell, as long as your CPU does not hit over 95C under OC L3, it does not qualify as overheating regardless of the temperature difference between the cores. so don't bother with dell.

 

I fix it for yoo 100C is the limit. From Dell by themself.

 

lol.. actually we are both wrong. it is 93C
https://www.dell.com/support/articl...-or-throttling-during-high-cpu-stress?lang=EN

 

Still wrong bruh... Re-read one more time. This is more for "In case the average temperature is above 93C and/or the processor does not sustain the speed as described in the table contact support for assistance."
Sorry I haven't the link in front of me now, but processor temp up to 100C is not a good enough reason for a repair

 

one line above it, also highlighted in bold, says "T he average temperature should not exceed 93C". Either way, anything over 90C is way too high

 

Average temp not exceed 93C isn't the same as spikes up to 100C and almost thermal shutdown. Spikes up to 100C isn't a reason for repair as stated by Dell support!! BTW. From the same Dell article you pushed above... Don't trust anything coming from Dell tech or Support.

 

Eventually we have to buy a sunon heatsink.

Do you have links to sell this?

 

If you look at Bios thermal configuration with a bios editor, you'll see that Critical Trip Point is set to 119°C !!!



And here's Intel opinion about 100°C being overheating and core temp discrepancy: https://communities.intel.com/message/441160#441160

Pathetic.

 

This is to avoid extra work from worried customers. The cheapest way to avoid having to fix the laptop due overheating and thermal shutdown. A suitable firmware is the cheapest fix.

 

Just checked and discovered they replaced my CCI Heatsink by a Sunon's one for my 15R3.

Here's a PIC



I had a near 20°C differential with the CCI's, and still have a near 10°C differential with the Sunon one.

Part N° AT1JM005SS0

 

did you repaste as well?

 

Yep i repasted with both heatsink, but still had troubles with the CCI's one with new pads.

The Suson has the stock pads.

 

Are you finished discussing this issue? What is the conclusion? Do I need to replace the sunon heat sink?

 

DELL sent a tech to change the mobo and the heatsink again. This time with a CCI heatsink and a Samsung VRAM mobo. I asked to keep the Sunon heatsink and the tech left the CCI heatsink as souvenir.
This time again, the CPU was curved and had still 10°C differential. I swapped the heatsink for the CCI's and the differential skyrocketed to 26°C like my unit the first time out of the box.

Conclusion : Curved CPU + cheap paste + CCI heatsink = Huge thermal issue. And it still happens dispite of what M. Azor said.

To be fine without to much work, you need a flat CPU, a Sunon Heatsink and a simple repast will do. Otherwise it will be trials and errors with thermal pad and LM to fix this mess.

Seriously, who thought a flexible arm alone could counter-balance the 2 other rigid arms, espacially on a lighly curved surface ?

I kept the Sunon heatsink in a box and sent the results to DELL with the CCI, they're considering a swap and asked me a one hour OCCT stress test with HWInfo's logs in OC3. I had to disable the thermal security on OCCT. The CPU reached 99°C easily. I'm now waiting for their answer since wednesday.

 

I live in South Korea. South Korea dell said they only have a cci heat sink. Is it possible to purchase a sunon heat sink?

 

No idea, i only saw CCI heatsink on ebay or aliexpress.

 

cicihen on ebay or taobao sells the 1080 heatsink

 

To be honest, I have never ever seen or heard about Curved BGA chips. @Mobius 1 have yoo?

 

Don't give them ideas like this

 

Look at the reflect of my screwdriver on my shiny BGA CPU:



And on another brand new swapped mobo :


Here's the reflect on the GPU to compare with:

 

Following your experience with keen interest Leeloyd. I've posted in some other threads here, but I ordered a 17r4 "certified refurbished" from the Dell Outlet UK with a 6700HQ and GTX1070. It arrived on 16th June and it was switched on less than an hour before I realised I had a problem. Don't know how it was "refurbished"....? Basically, by the time I got through the Windows setup the fans were already going crazy. HWiNFO confirmed it was overheating with a differential of over 30C!!!


Called Dell, who did a remote stress test. As soon as it started the CPU was already running at 100C (less than 10secs into the test!!). In HWiNFO I once had it record 101C doing nothing more than downloading a game on Steam.


To be fair to Dell, their customer service has been very good so far. No complaints with their response, but your experience had me nervous about the chances of it being fixed. That said, I will of course give them the benefit of the doubt.


I have a guy coming out on Tuesday to replace the heatsink and fans (also having the QHD panel replaced for the newer one that doesn't have the horizontal lines). Fingers crossed it does the job, but I'll be sure to report back one way or the other. Hopefully to confirm the issue is fully resolved.


It's a shame. I haven't been able to use the laptop at all so far and I was really looking forward to getting it (keep in mind, my old gaming laptop is a gtx460m so it's quite an upgrade...). Other than the overheating and panel, the machine feel very solid and looks great. Just hope everything can get resolved first time.

 

Just got DELL's answer for OCCT reaching 99°C in OC3 mode :

Rejected : system is running all the time at 100% of usage with 3890.5 MHZ as frequency, this is a normal reached temperature with this kind of use.

 

Do you see scratched GPU and possibly warped CPU die because of uneven high pressure from heatsink. Did you apply more pressure on the arm of heatsink?
Can you post some macro/close up shots of CPU die and heatsink?

 

Here's the scraches:

Heatsink


and CPU


Looks like a match.

Edit: I think i figured what's happening:

Alienware 15R3/17RA bad heatsink design by leeloyd posted Jun 26, 2017 at 6:27 PM

 

URL Link Please

 

https://world.taobao.com/item/54349...1.14.32.ebb2eb2Ieox3c&ns=1&abbucket=16#detail

 

This one is a CCI, don't get fooled by the Sunon fans. Both CCI and Sunon Heatsink have Sunon fans.

 

never said that he only sell sunon

 

... I'm on the phone right now, but I'll comment later when I get to a computer.

::iunlock::

 

I have repasted 4 AW17R4 (1x6820 3x7820) with LM so far and none have core temp differential. The trick is the use ALOT of LM, 0.5g for both CPU and GPU.

 

Just make sure that none of it squirts out of the sides.

 

That's way too much, poses a danger to spill out.









One way to "fix" the stutter is to force-flash the "non overclocked" 150w vbios originally shipped with the AW17. The OC vbios being shipped out and offered on the dell website is 180w.

This has worked on most AW17s that have not been properly repasted/repadded given that the ambient temperature isn't too crazy and the CPU isn't overclocked too far.

@Prema notes that most BGA notebooks have a power circuit / power circuit cooling design that can barely sustain the original gimped TDP that it ships with. So that's something to keep in mind. A 150w -> 180w bump, while seems insignificant, may be the straw that broke the camel's back.

And god knows who thought that the 4 mosfet chip on the AW pascal motherboard can be cooled with just thin protrusion of aluminum position furthest away from the heatpipes.

 

Not strange you see people have stutters and freezes with the new vBios with raised TDP. I hope the buyers went for maxed Premium Support warranty. As you know, AW's CEO Mr. Azor Was forced to push out vBios with higher TDP. As he bragged that Alienware was one of the few ODM who used maxed out 1080. The other ODM's used Crippled down powered Gtx1080 version. The results was totally opposite.

 

well it outperformed the 150w 1080s for around 2min before stutter

 

Yeah, the cooling + components on the MB is designed for lower specs long time before Dell decided to change for higher power for graphics. The specs used ain't designed for 20% higher power draw from graphics. Aka pushed out the new vBios without the components can keep up with higher power.

 

the funny thing, eventually most manufacturers view the 1080 as a titan xp for the mobile platform, then they go all out on the cooling like msi/clevo/asus etc with higher power draw directly from the factory

 

Yeah that is opposite of what Dell did. It's pretty stupid by Dellienware first brag that the cooling is designed for 10-15% higher power draw to give headroom for increased temp due dust will be filled up in the laptop and paste degradation, then uncritical push out firmware who increase the power draw with 20%. Damn nice engineering. Do exactly the same for tools intended used for oil platforms or other critical infrastructure You would be sued Buyers who skipped the maxed Premium Support warranty will pay the price!!

 

Premium warranty isn't cheap either.

 

Hello everyone...I'm here....okay let's see...as I *exhale*....

.....

I would not recommend this as I've already tried lowering the stud height during my testing long ago.

There is just too much risk and over-tightening the 'tri-sadness-arm," can do more harm than good...I've tested this with pressure paper as I do with all my testing. That arm is also thin and with the dynamics of heat and metal...well you get the drift...

ie...I don't approve of that method.

I wouldn't do it. It is not the ideal for most people. Not saying that it doesn't work, but there is a reason why I haven't made any mention of this.

To each his own, but I've already tried this long ago...

The pads should always be switched out as the stock ones are inadequate.

Yes that's correct. adjusting the GPU arms does help as well.

Because it's one unit and it does help toward the cause...

Here's a 17R4 I've finished up last night.

This HS was a mess stock with 10C differentials.. It was a revision 1 cci... Go Figure.

The inconsistency lies in the different revisions of the heat sink. Sunon and revision 2 of the CCI are the best ones. I get even cores with those, every time...granted I do much more to mod with my repastes.

Sunons have a more matted stealth like paint scheme and as you've seen from other posts, it says sunon on the bottom right of the GPU heat plate.

When using LM it's imperative that there are no gaps. This is why I spend the majority of the time with my work on balancing the heat sink. Gaps and LM is a nightmare waiting to happen...

Pressure paper does have its place. I use it daily for a reason.

It's quite comical. It just amazes me...

Anyung...PM me.

Nice animation. Neato...

The issue here is that the heat sink needs to be applied flat, straight up and down...and not in a rocking fashion...that's the culprit, so it's not really an issue per say of a new discovery...it's a common issue that is inevitable if you lay down the heat sink like what you've shown in the animation. That's not how it is suppose to be done. This is a fault of that person at the factory....

I work on these systems daily and see these come across my work bench daily...I've spotted a few, but that is out of many....

Over promise, under deliver. Welcome to the industry. It is sad that this has become a norm.

If people don't understand @Papusan when he lays out facts like this, you have to understand that he is basing it off of FACTS and not opinions. So if anyone is confused, I highly recommend that you go do your due diligence and you'll come to see how the game works in this strange 'over promising and under delivering," industry.

AW/Dell has issues, of course, but so does other Brands...they all have issues...

 

hi iunlock I was sent an engineer with a sunon heatsink he installed everything went fine but I felt like I needed to do it so I stripped it back re padded and repasted. much better temps than CCI 85 max down to 75 with kryonaugt.

question to you what is the best method to balance the heatsink?

when u say straight up straight down can you elaborate on this. do you hold mb and heatsink together then screw all screws bit by bit

Sent from my Nexus 5X using Tapatalk

 

That's quite expensive. It does its job daily. So, 1 paper comes with a one time use only, right?
I'm thinking of my own method to check the HSF pressure. The idea sounds stupid though.

 

Received a brand new system, built last week. i7-7820hk and it still has CPU temp difference issue.

I'm pretty sure they're just using old parts at the factory.. this one even has the old bezel and tobii eye camera from last year's Skylake models..

Dell can't get their **** together.