That can be frustrating. Did you order the DIY model? It might be worth asking them to escalate your ticket and ask if they can offer you some kind of repair if you send the laptop to a repair center. I mean, assuming it’s not something you want to tackle yourself.
Yes mine is a DIY. I would be fine doing it myself, but finding the time is the issue. And like I said, I want this solution before warranty expires, but it looks like Framework may be delaying for that reason. Mine will on rare occasions suddenly power completely off without warning while gaming. My last email to them was informing them of this along with the request to expedite a solution. They just want me to provide logs when it happens again (logs that don’t really help because it suddenly stops/poweroff with no time for errors to be logged). So now I’m just in a holding pattern until my FW16 powers off unexpectedly again.
The feeling they are just delaying to waste my time until warranty is up is growing stronger every day.
One thing I would like to suggest with the guides: why connect the fingerprint reader to power on the system? Theres a power switch on the mainboard itself (probably placed there to give the option of standalone operation) that you could use. Iit is on the top left corner of the installed motherboard, to the left of the wifi card. The silkscreen label is SW1
2 Likes
Wish I would have known that a few hours ago - never thought to look for a standalone switch on the board, even though I just went through setting up a headless 13 board the other day…
I went through my LM removal today - still waiting on my PTM7950 and shims to arrive. Got mine from moddiy.com in an attempt to avoid Amazon and AliExpress where possible. Figured I would post my experience for posterity.
Followed the steps @taezea took in their approach, as my model appeared to be one of the more affected ones - I would struggle to hit 30w under all-core stress tests, Cinebench R23 scores around 12000. Specifically, I ran sudo stress-ng -c 16 for 10 - 15 minutes, then quickly closed the test, started a shutdown, closed the lid, and flipped it over until it fully cooled. This pooled most of the LM on the CPU cooler side of the assembly, along the edges between the shim and the gasket.
After everything cooled down, I removed the input modules, expansion module, and the midplate. Once the midplate was off, I loosened the three screws securing the battery (Red) and pulled it out with the tab. After that, I removed the Power Button and Vent plate which pulled away easily after disconnecting the ribbon cable (Yellow). Finally, I loosened all of the screws holding the CPU Cooler in place (Green). Note that the two screws on the furthest edges of the cooler - the wide ones - are not captive, so be careful to put these somewhere they won’t get lost.
To further weaken what was left, I heated the cooler block with a hair dryer because I was too lazy to go downstairs and fish out the heat gun. While it was hot (uncomfortably hot to the touch! If you can stand to touch it, it likely isn’t warm enough and you might damage something), I gently pried up using the plastic spudger on the screwdriver - it took a bit of patience and time, but it eventually peeled off without too much exertion. Once removed, I found that there was adhesive from the gasket in place as well, which was likely contributing to the amount of stickage I was fighting.
Luckily, the die cleans super easily. (I was just happy that the thing came off by this point, so I forgot to take a picture of the uncleaned die.) The spudger was more than enough to get into the crevices and lift out the errant LM, and cleanly scraped / flaked free the remainder on the die. Do not use a metal tool for this.
If you are not replacing your cooler with a new one (I bent a heat pipe slightly during my disassembly and ordered a replacement to be safe), make sure you also remove any LM that remains on the cooler.
As you can see, my coverage was pretty poor, especially after the flip before cooling. About half the die would have had coverage at best.
As I don’t have my PTM7950 and shims in yet, I just did a layer of Thermal Grizzly Kryonaut, as it was leftover from a previous build. If you don’t have a spreader for this, you could probably use the spudger here too. Reassembly was as simple as screwing everything back together opposite the way I removed it.
Results afterwards have been stellar. I am hovering around 15000 in Cinebench R23 now, with my CPU pulling ~47-50w instead of barely 30w. I still hit 100 C if fully stressed, but I’m getting far more out of it with a swap to simple paste than before, and it will be much simpler to open it back up and change out the paste for PTM.
11 Likes
I’ve been having thermal issues on my framework 16 and submitted the Phase Change Thermal Pad Kit Request form some months ago (around the new year), but haven’t received it yet. They say it’s supposed to be available in “early 2025”, but I’m not so sure since it’s already March and I haven’t heard anything new. Is there any news on this front?
1 Like
We are all still waiting…
1 Like
I reached out to support to try and get any additional info and they just pointed me to the form and said I’d get an email when they started shipping kits…
Really wish they would say something
4 Likes
You are all wonderful and I’m legitimately about to cry. I followed the benchmark and flip instructions to get the liquid thermal off the cpu, was able to pry it off easily and found it only covered about 35% of the chip, with an odd burnt gunk as well. I have applied my paste of choice, and it’s now humming beautifully, maxing out at 94 Celsius per core and hitting a 45w average for the first time. (I didn’t think to get pictures as I was too worried/excited when I saw how bad the liquid metal was.) I didn’t think I was able to do this before I found this thread. May your cores be cool and your games hot!
5 Likes
I really wasn’t looking forward to this job. Especially removing the heatsink from the cpu. But in the end that wasn’t too bad. Maybe because the system was still hot when I opened the case.
The LM on my cpu looked like that too. There was not much left of it.
1 Like
For the FW provided PTM for the FW16.
After completing the process myself, I think there might be a little easier and safer way to do it.
After removing the heat sink.
Lift the plastic sheet a little around the CPU and place anti-static A4 sheets of paper all around the CPU, and then put the plastic sheet down on top of the A4 sheets.
The paper, in addition to the plastic sheet, will then catch any LM bits that might jump or escape while you rub them off the CPU.
I received the FW provided PTM replacement pad and applied it following the official Guide. Worked fine, although the force required to pry/lift the heatsink off the CPU felt a bit frightening, though still far from braking anything.
I tried heating the CPU up before opening the laptop but it cools down too quickly. So a hair dryer would have made it safer I guess.
Before: 20°C core difference, 40W sustained CPU power, 809 pts cinebench r24 multicore
After: 10°C core difference, 50W sustained CPU power, 856 pts cinebench r24 multicore
Thanks Framework!
3 Likes
I just followed the advice from here (Uneven CPU thermals! - #153 by taezea) to hit the CPU hard, then turn off and flip over. I let everything cool for at least 5 minutes before starting disassembly, and I was probably 30 minutes in before getting to the heatsink removal step. I would recommend just following the official instructions since it wasn’t that difficult. Work around the edges slowly, using the provided spudger, and the heatsink should break free. I didn’t have any board or heatpipe flex as rotated the spudger to lever it upward, and I didn’t pull on the heatpipes.
The LM actually cleaned off really easily. Most of it broke free as a large piece, and the smaller pieces easily adhered to my isopropanol-soaked paper towels
3 Likes
I recently discovered all the information regarding over heating core temps, throttling, delta differences etc. and finally got in contact with support as I suffered too. They are sending me the PTM replacement and I’m looking forward to the project as well as new performance. Going to do as you did and stress test and flip upside down to hopefully making the process easier.
1 Like
I removed the motherboard (official instructions). https://guides.frame.work/Guide/Liquid+Metal+->+PTM+7958/402?lang=en
I already had to replace the motherboard once because it was defective. With both the defective and the new motherboard, the right-hand heat exchanger was slightly stuck in the housing. You need to assist here during removal (Motherboard with cooler). Special attention is also required during installation to ensure that the heat exchanger is positioned correctly deep inside the housing. If this is overlooked, the subsequent components (fan plate) will not fit.
Because I knew this, I did not attempt to remove the cooler while the motherboard was still in the housing. If something else gets stuck, it is easy to lose track of what is going on (slightly stuck).
Once the motherboard is on the table, heat the cooler with a hair dryer (as other users have already suggested). After a while, the cooler will come loose with little force on the framework-tool. A heat gun gets much too hot. With a little patience, the cooler will come loose.
In my opinion, you can save yourself the effort of “preheating” with a CPU stress test.
First of all, how do I know if my Framework 16 laptop cpu/gpu is liquid metalled.
Is there information on what batches/years/models were done with liquid metal?
If I am sure mine is, i would like to replace it.
I don’t like it, it’s a sleeping murderer of laptops
When did you buy yours? Only the first few batches were LM.
My FW16 was batch 16, and it had LM.
1 Like
Mine was Batch 20 and it had LM. I was able to get them to send me a replacement board with the PTM as I was uncomfortable to do the replacement myself.
1 Like