Hi!

I'm currently diagnosing a Lenovo X1 Carbon G5 system.
The issue is that it turns off randomly and cant be turned back on, only after removing the charger and pressing the reset button.
Sometimes it stays on for 2 hours, sometimes 2 minutes.
When it turnes off the USBC 20V remains but the 10V from the BQ25700A charge controller disappears and the switching stops on the charging inductor (L5) so VSYS15_OUT disappears.
The time that it stays on gets shorter and shorter each time it turns off.
So im pretty sure that its temperature dependent.

The buck controller is U41 -> TPS51285B
I was probing around and the 5VALW regulaion seems a bit off.
The 3,3V is spot on, but the 5v is pretty low, around 4,5v and constantly jumping around.
Before shutting down the 5V rail falls to 4,3-4,4V.
The LDO-s (VREG5 and VREG3, pin 13 and 3) measure 10K and 5K but VCC5M and VCC3M (measuring on coils L3 and L4) both are 100 Ohms.
The VCC3M consumes 0,012A, the VCC5M 0,02A.
The resistance between VCC3M and VCC5M is over 18 kOhm and climbing.

Since the board works (for limited time) i think that the pch and cpu must be fine.
I started to look around for potentially partially shorted components. Since i don't have a thermal camera i used the alcohol techniqe.

I found that U107 (RTS5234S pcie card controller) was getting unreasonably hot so i have removed it.
The result was: VCC3M and VCC5M havent changed resistance, they are still 100 Ohms, but VCC5M is not 5.1V instead of 4,3-4,4V.
So i might have found one of the potential faults, it seems like U107 was putting a load on the U41 buck controller.
However not much have changed, the board still shuts down randomly.

None of the other chips produced something like this exept for the ALPINE_RIDGE_DP Thunderbolt Controller (U182).
The alcohol quickly disappears from half of the chip while booting, but after displaying an image the alcohol returns to 3/4 of the chip.

After all this i'm stuck.
What could cause the shutdowns, how can i further diagnose the issue?
Can the thunderbolt controller be the problem? (At first i though not because it requests 20v)
The partial short still remains on both rails, how can i find the source of that with this low of a current?

Thanks in advance for your help!
I'm happy to provide further information, measurements and such.

*forgot to write that i have flashed the main bios chip with a clean ME firmware and the thunderbolt controller firmware, neither made any difference