Uneven CPU thermals!

There’s a bit more to it. Unfortunately, it doesn’t look like we can find an easy solution.

I decided to remove each layer from one of the heatsinks and, as far as I can see, there are 4 different layers + the base.

@PSierra117 please have a look at the following photos.
I feel that this could help you find a solution.

Layer 1

Layer 2

Layer 3


Layer 4


4 Likes

Is that a copper mesh?

I’m quite puzzled by this design.

Yup. Layers 3 and 4 are a cooper mesh.

Same. I was expecting to find a block, not multiple layers with some empty space between layers 3 and 4.

1 Like

Well i knew before your Pics how it looks inside there. Because its a Vapor Chamber. There is near Vacuum inside and a tiny Amount of Liquid inside the Coppermesh. The Liquid evaporates on the Hotspot and liquifies on the Heatpipe Area and through the Copper Wick its transfered back to the Hot Area.
This is alot faster than transferring Heat through a solid Cube of Copper. A Vapor Chamber is just another Form of a Heatpipe.
Here a great Video which shows the Principals and the insides of Heatpipes.

What’s Inside the Worlds’ Fastest Heat Conductor? - Youtube “The Action Lab”

*edit
The Vaporchamber works perfectly fine, It’s just the added shim+LM+ Glue? where the Problem is. If the Vapor Chamber wouldn’t work as intended there would be all Cores sitting at a 100C. When a CPU has a high Delta between Cores there is a big variance on Heat Transfer directly above the Die. When all Cores are perfectly sitting under a copper Block they are all nearly at the same Temp until the Block is Heatsoaked and than starts throttling. Thats why the Heat has to be force away as fast as possible. A Vaporchamber+Heatpipe Design is therefore quiete proven. When the Air coming out of the Device is not really Hot but the CPU is than there is one Step between Transfers where its to slow.

3 Likes

could you measure the shim? Than i can order some CopperShims that Size and thickness and maybe some shims a bit thicker, for trying some different Pressure :slight_smile:

18mm x 13.4mm with a thickness of 0.4mm on the outer part
When I measure with the glue part, the thickness is 0.45mm

You could easily go with 21mm x 16.4 mm while trying a thickness between 0.45 and 0.55mm.

2 Likes

would 15x15 fit the die? I can only Source 20x20mm or 15x15mm in various thicknesses.
edit:
Well what a Question, i just get the 20x20 one. It fits save :man_facepalming:

It should be fine with 20x20.
As soon as I’ll have the new heatsink I’ll take some measurements.

Well i reread some Reviews of the Framework Laptop 16 and there was definitly something changed between the Review Units and the Customer Units

Notebookcheck Framework Laptop 16

“The CPU would boost to 4.6 GHz and 80 W for the first couple of seconds when initiating Prime95 stress before quickly falling and stabilizing to 4.0 GHz, 54 W, and a toastier-than-expected 96 C.”


I Really Assume there is alot of Potential in the Cooling System, but i furthermore suspect the Glue or the Shim to be the Main Culprit holding it back.

2 Likes

well that’s very interesting, i’m glad you dug that up.

i’m uh, not nearly as industrially inclined as the rest of y’all so modification of the thermal assembly is well beyond me, but i’m going to attempt in the coming days to try adjusting the tensioning of the screws to see if i can get better than the mid-11k to low 12k cbr23 results i was getting.

i’ll just have to be careful: this is my only ‘real’ computer (the other devices are a phone, a steam deck, and a stack of very decrepit devices not fit for much).

You would think they had review samples. Golden samples for the heat transfer solution.
I opened up the laptop again to replace the thermal paste for a thermal grizzly PTM pad.
The results are similar to the paste, if you use high quality paste.

I was thinking about the glued on plate, which I assume was done to use liquid metal (protect the copper) by design. It resembles also the principle on desktop cpu’s where the integrated heat spreader is soldered/glued on the die.
This design is that basically in reverse of that principle, in theory it should be fine, if done well. (see quote below).

I also replace the paste on the dGPU with the PTM pad, here the results were more significant.
In heaven benchmark a 50% increase in points and average fps.

I also remembered reading something about the thermal solution by framework:

New issues

  1. Yield issues on CPU thermal system - We set strict criteria for thermal performance on Framework Laptop 16, and we have a station on the production line that both tests for system stability and performance over an extended period of heavy benchmark load. When producing the first set of mass production systems for press reviews and internal use, we found that some had CPU performance results that were lower than our threshold. We first found an issue with liquid metal pad installation on some units. After adjusting the assembly process to resolve that, we found that about 20% of units still had CPU performance about 5% lower than expected. We determined that the issue followed the CPU heatsink, and that the failing heatsinks have higher than expected thermal resistance. We’re now taking two actions in parallel. First, we’re validating that the change in liquid metal assembly does indeed fully resolve the issue there. Second, we’re working with the heatsink vendor Cooler Master to understand why their outgoing quality control didn’t catch the thermal resistance issue as it should have. We expect answers on both of these within the next week.

Soure: “Fourth update on Framework Laptop 16 shipment timing” email by framework, send 28-12-2023 00:30.

It could be this problem was never really resolved and still the QC is lacking?

5 Likes

As much as I dislike many things about Amazon, there’s a reason they are popular, lol.

I actually ordered some thermal pad options over the last couple months. I figured I could do some experimentation with my FW 16. I ordered some Thermal Grizzly Cryosheet from Amazon, and it came like the next day. But I wanted to wait until I had multiple options, then spend a day or two testing them. I ordered some PTM7950 from MODDIY in July. Then I got busy with work and kind of forgot about it. When I remembered, I looked into it and it had quickly shipped and there were tracking updates through a couple days after I ordered it…then nothing for over a month. It was clearly lost in shipping. MODDIY support was very responsive and helpful when I asked about it. After a few days of looking into it, the shipping company said it was going to be 5-8 additional business days before they could give us an answer, so MODDIY just shipped me an entire new order. Now I’m waiting on that. I ordered some IC Graphite pads from Innovation cooling. They said they were in stock. The order confirmation email said they would wait to ship until my payment processed. It came out of my account that day. That was ten days ago, with no word yet.

In the meantime, I have been emailing support about my temp results, but I told them I was honestly fine trying some various thermal interface options myself, just to see if it would make a difference. My laptop has so far been reliable, so I don’t know if they need to spend the money to send me a new mainboard. But I wanted to see what they would say before I started messing with it myself.

After some testing and screenshots, Framework support escalated my ticket and they almost immediately determined they wanted to send me a new mainboard. So, after all that, I guess the experimenting with various thermal pads will have to wait, lol. I’ll report back how the thermals are on the replacement mainboard, once that arrives and I get it installed.

Also, I just wanted to point out that I don’t know what my problem was when I wasn’t noticing the air venting from the laptop being hot. Maybe I was just tired that day. When doing additional testing the next morning, the air coming from the side vents was clearly quite warm.

4 Likes

I had very good results with the thermal grizzly PhaseSheet PTM on the dGPU.

It didn’t do much for the cpu compared to the (insufficient) liquid metal installed by framework compared to the pad, but the pad at least hasn’t degraded yet as the LM did.

I would like to see your results with the kryosheet.

I will have to see how the new mainboard performs. If all is well, I may leave well enough alone, or I may decide to tinker anyway, just for the heck of it. We’ll see. I was curious about the Kryosheet too.

4 Likes

I’d really appreciate if we got a definitive answer from the team on this specific point. I mean, they’ve been replacing quite a few Mainboards, but even the replacement parts aren’t performing as they (and, by extension, us) expect them to. Personally, I’ve decided to forego the second RMA they’re offering me, as I am currently not confident that they have resolved the underlying issue and don’t want to waste their time or mine in sending the entire laptop back.

3 Likes

In my opinion based on my personal experience and what was shared on the forum, I think the email that I quoted, never was fully resolved, and they just let it slip again.
I am not sure if Framework would ever admit that, as that could have legal liability issues.

From the 2 samples I have personally experienced, the original and the replacement, the degradation happens because of the LM flows outwards. I speculate this happens because of a unaccounted for hot spot of where the plate isn’t straight enough or not glued/soldered on sufficiently/correctly. It causes higher temperatures of the LM than needed and thus more of the LM phase changes more than it should.
It is al speculation on presented evidence here and personal observations and tests with other thermal compounds and solutions. I think only framework has the volume of cpus and thermal solutions to actually to come to some conclusion, not sure if they are willing to share that?

So I am really interested in @PSierra117 plan to test out his own plate(shim) and see his results, it could confirm my speculation or maybe not…

It is very frustrating, I have replaced the LM with a thermal grizzly pad and I have been testing it daily on my machine, and I keep getting scores nearly or over 16k in CB r23, but still that one core that gets hotter than the rest.
If they every come up with a message that all the new heat-sinks don’t have this issue anymore I would replace my current one in a heartbeat, but seeing the same issue with 2 different boards and heat-sinks what are the chances a 3rd was fabricated without that issue? In the email I quoted they said 20% but for me its been 2/2 so…

4 Likes

My 2 Heatsinks are ordered. Thermal Grizzly Phase Sheet and a Kryosheet. Also i ordered 20x20mm Copper shims in 0,5 and 0,8mm. I will test both Heatsinks unmodified with the current Fake PTM i already have. The Better one stays untouched, the worse one gets its Shim removed and inspected. I will replace it with my own Shim in a PTM Sandwich.
The Shim as it looked on the Pictures looks like a Heat Throttle as its not fully attached to the Vaporchamber.
I suspect, that replacing it by a bigger Shim will improve its Ability to transfer more Heat into a wider Area in the Vaporchamber thus increasing its efficiency.

But im still not Home till the second October week. So you Guys have to wait on my Tests or do sone on your own :wink:

5 Likes

The sixth update on framework laptop 16 shipment timing mentioned the CPU thermal fixes.

CPU thermal module performance - Our thermal module supplier improved their vapor chamber soldering process, which reduced thermal resistance. While this was only intended to improve manufacturing yield, it actually ended up improving thermal performance too. All press units passed the same pass/fail criteria that we use for CPU performance on customer units though, so we consider press unit CPU benchmarking to be a fair representation of what customers will receive.
Liquid metal barrier adjustments - We made some adjustments to the liquid metal application process to prevent any leakage risk on customer units.

3 Likes

Well as it seems, it is still the same Issue. Exactly the same…

2 Likes

Was about to say the same thing, it’s basically what I mentioned before, that they let it slip, to what whatever happened initially…

edit: With they being the supplier, coolermaster.

1 Like