I also have two Heatsinks sitting here 
One unused and one tested. I am still thinking about modifying the Tested one to even get better Thermals than i have right now. Maybe even soldering a new Shim to the Vaporchamber.
2 Likes
There’s probably a market for modified heatsinks. You could make them with new shims and sell them packaged with PTM7950, charge the cost plus labor and shipping. I know I’d buy one.
4 Likes
Figured I would add my progress.
I have an RMA in progress for a 4th mainboard. I directly asked if they have an updated thermal solution given the number of RMAs I have now had and they told me it was the same as before. So either they aren’t willing to share any fixes publicly, or this new board will degrade again.
I feel like the trick at this point is to figure out the exact cause and send as much data to LTT as possible. My best guess right now is thermal cycles combined with leaving the laptop on for extended periods.
If I have a 4th board degrade, there is no denying this is a factory defect with the cooling solution and the excuse of “only a few owners having thermal issues” is no longer valid. That just means owners aren’t noticing or reporting it.
6 Likes
I think most are not even aware they have the issue. I didn’t realize either until I read this thread and followed the steps to recreate it on-demand. I am going thru the RMA process and I am afraid it’s the roll of the dice or it’s a fundamental flaw with the cooling system design.
2 Likes
yeah same. mine has always ramped the fans up quick since day one so assumed it was just a trade-off of having a high performance laptop.
Update with a new board send by framework support (r7 7840hs) :
Cinebench R23 Multicore : 14359 (previously 11803)
Peak package voltage : 53 Watt (previously 36 Watt)
Sustained Cpu Clock : 4.2 GHz (previously 3.6 GHz)
As you can see, It’s a big improvement but I still have uneven CPU thermals : Core 3 is still 20 to 40 degrees hotter than other core, I don’t know if it’s good or within the framework 16 r7 baseline and if will hold in time but for now I have better perf
1 Like
My Board i got replaced had 14600Points. It is still far far off, what the 7840hs could run in this Device. Mine is getting above 16k
Thanks for sharing. I will test my machine when I get the replacement board as I am very curious and if they improvements are temporary and as the Liquid Metal degrades and back to square one in 4 months.
I will check with support but my replacement board was also a little bend around the left heatpipe which made it quite difficult to replace. My only worries if they ask me to send the entire computer as I don’t have any replacement and I need it for my studies
@Clement_Fouche @PSierra117
I had my replacement like 2 weeks ago. When I received it it can boost and sustain a 55w power. Now, it is back to 39w, with a temperature delta of 32C
Thanks for your advice, I will ask support to send me another board that is not flawed
Thank you. So another Indication, that every Unit is flawed. Thats the LM RunOff Degradation. They should send their Units with PTM7950 instead of this Crap.
This needs a major Rework, but nothin is happening.
2 Likes
And, with a proper heatsink where we don’t have a shim glued 
1 Like
It isn’t glued, but sodlered as i removed it on mine 
But not fully soldered, that’s causing the uneven Temps, as the Thermal Energy of the Outer Cores has to push through the narrowed Transfer Area.
3 Likes
So the shim is not only too small but the solder didn’t flow underneath the whole shim?
The Shim is big enough. But yes it is not fully soldered as i documented. That is the Main Reason for the uneven Thermals.
The Degradation of the Remaining Performance is the Second Issue its the Liquid Metal Run off over Time.
2 Likes
I’m sorry 4th board??
What is the problem you are facing?
One board or two is like, ok maybe manufacturing or whatever. But 4 is like, nah nonono something is amiss.
Why is your EDC 120A?
If I run Cinebench it is like, 80 ish. Max out at 100.
Your SVI3 is also significantly higher, mine was 0.758V highest during a cinebench run.
Now, I am using the 7840HS, so I guess there are significant differences. But wow, what is going on
brand new, like, “2 hours into the machine” new?
Wow what is going on.
My highest temperature delta is … like, 20. Core 1 being the lowest and Core 2 4 5 being hottest.
Score isn’t fantastic, 14500, 14600 ish. But it stable.
Connected to 100W charger, expansion bay shell. display set to 165Hz. BIOS 3.03
I am a batch 1 owner, so I have had to most time to go through these boards. Based on evidence in this thread, the thermal issues are a factory defect in the heatsink. I noticed games weren’t using 100% GPU after some time and traced it to the CPU thermals. That was on my first board which lasted about 4 months before I noticed. To get my second board replacement, I had to send in the entire laptop only for them to agree there was an issue and replace it. Now I am getting a 4th board since the replacement they put in themselves is degrading. This time it is just a board replacement, no need to send in the entire laptop.
I also believe a Cinebench R23 score of anything less than 16k is unacceptable. I do have a 7940HS so I expect top performance since I paid the extra 200 USD for the small performance boost. We also know the 7940HS is capable of over 17k since it is used in other laptops with better cooling. The heat pipe design in the FW16 is being limited by the thermal interfacing in the heatsink itself.
Getting even core temperatures is the key to performance for this laptop, and it seems the F16 heatsink solution just can’t deliver. I would accept a 10C differential, but 20-40C is crazy. My most recent board was 64C to 100C, way too big of a gap. I bet FW could crank these all the way up to 54W sustained if they would just refine the heatsink a bit more.
4 Likes
So is there any point to do a RMA if you’re going to just get another flawed board? It’s becoming more evident it is a fundamental design issue and switching boards is more psychological satisfaction going thru the motions… it is not resolving anything.
3 Likes
This is wild to read. Does yours regularly go to 100C? Do you do heavy workloads?
The cooling, though. Although you might be able to hit 16K if you tune it a touch. The actual thermal capability of the cooler is way beyond 45W. I can game on my 7840HS at 55W total package power and have lower temps compared to a cinebench pull.
It’s akin to the 14900K problem. Except this time, we don’t have a IHS.
I want to say it’s a paste issue. I think framework got scammed.
Chassis bending you say. Interesting.