Well did some more Experiments.
First i tried putting Liquid Metal between the Shim and the Vaporchamber.
Worked okay, but the Risk was to high for me. PTM7950 is easier to apply and not that risky.
I made a Thermalpastebarrier arround the Shim.
Well for my next Test i sadly destroyed this Heatsink, i broke off one of the Retention/Holding Screws around the CPU so i had to further improve my Soldered Heatsink.
Reaheatet it thoroughly moved and pressed the Shim as good as possible.
Quiete alot Solder came out there. Had to lap the whole assembly to get it sitting evenly again.
The Result speaks for itself…
80W Peak TDP and around 63W Sustained TDP
about 16,6k CBR23 Score.
I think this is what i will keep. Have to improve the Spillagebarrier and make it really spill prove. But for now i will be stationary for some time, so the LM can settle with the Copper
Without an undervolt I sit slightly under that on my R9 with shim sandwich. I can only imagine what undervolt, modified power limits, and proper liquid metal application can do to my system.
Yes, that was my intent as well to open RMA so issue is officially reported and logged in their system but hold RMA until an official fix is available. I do want FW to be successful.
@PSierra117
Amazing results.
Is my understanding correct:
You are using the FW design, with LM between CPU and Shim, and solder between shim and heat sink. The difference being, you actually have not messed up on the “doing it” part so that it now actually works as designed.
I use Universal x86 Tuning Utility for modifying/unlocking the Stock TDP Settings
Stock Settings with maximum performance power profile, boost to above 65w for 1-3seconds and than cap at 54w ustained.
@James3 Yes all in all alot more janky and Backyard Garage Style, but working yes
And conventional Liquid Metal is alot more problematic to spill.
but well i dissambled the whole thing 4 times to readjust and check, but i build myself a spill barrier now that fits somehow with the Shim and thermal pads as a second layer of defense. I have alot of LM sitting next to the Die now, but as FW did put Conformal Coating there it shouldn’t be a problem. otherwise it will be an expensive Board to buy
That’s pretty impressive that with a non-defective heatsink, the system can be capped at 80C and still sustain the full 45W limit.
I replaced just the LM with a more generous application of TG LM, and even at around 30-40W usage I get close to thermal throttle sustained. Have you measured if your chassis temps reduced over time when you apply those caps?
This seems like a fully operating heatsink assembly would allow much reduced fan noise for the same performance we get now too or a cooler keyboard deck area.
Is there an associated design mistake here.
If a CPU hits a 100C threshold it should throttle back.
If the temperature of the heat sink hits a threshold, the fans should start.
There does not seem much point in spinning fans, if the heat sink is faulty, and not actually drawing the heat away from the CPU efficiently.
If you start the fans based on CPU temp, you get fans spun up and not doing much, because the heat sink has not heated up much.
Also, with those extra temp sensors on the heat sink, one would be able to very quickly diagnose a faulting heat sink, even on the production line.
I.e. raise CPU to temp X, Heat sink temp should rise to Y within time Z. If the heat sink does not rise in temp quickly enough, one has a faulty heat sink.
To complicated to be viable. The Fan is controlled bei the EC based on the Power Plan and CPU Temp. There is a hysteresis to prevent frequent ramping to damp fan rpm for little heatspikrs. No one designs a Fancontroller with a defective Heatsink in mind.
Impressed by you guys tinkering that much. Thank you for all of that, it give me hope!
Just a follow-up, it’s now been 14 days since I got my mainboard RMAd. I don’t even get 14s in CBR23 now. Max I get is 13,950, cold start, cold room.
The laptop is still WAY quieter than it was before RMA during simple/regular tasks but thought I’ll keep contributing even thought that’s not much compared to you guys.
My ticket is still open with FW, I just told them I don’t want any fix as for right now as I don’t believe getting a new mainboard could help on the long term.
Good decision to log the RMA ticket so it is officially registered in their system as yet another confirmed example. Unfortunately not everyone is aware of this fundamental design flaw. However, the more people logging the ticket and more awareness we can generate the more likely something will be done sooner rather than later.
Good news is that FW is aware of this issue. Cooler Master needs to step up and come good (on their partnership with FW) and work it out with FW so a fix can be released.
