MSI gaming laptop randomly does not detect nvme (MS-16J51 board)

[Grouickman]

Hello,

Since 5 or 6 months my old MSI laptop (based on MS-16J51 board, I don't have the machine with me atm, i'll update the exact number when I'll get home) started randomly loosing the nvme drive.

At first I thought it was a bios issue, changed the bios battery that was a bit low, but did not solve the issue.

It used to happen like once every 2 or 3 weeks, but now it became more frequent and can happen several time in the same day if you restart the computer.

It looses the nvme at power-up (will boot to bios and report no nvme drive but still sees to SATA drive) but never when it's running.
The workaround is to unplugg the power, remove the main battery and bios battery and it will properly detect the nvme on power-up.
Sometime powerring down, removing power and waiting several hours also "fixes" the issue.
The reset switch does not help.

Of course first thought was the nvme drive. I swapped the nvme (960 evo replaced with 970 evo) but this did not help and the original nvme did not show any errors and works perfectly fine in another machine.

So this weekend I dowloaded the schematics and found that the 3.3V on the nvme port (+3VRUN) was not present when the nvme was not detected. but the main 3.3V voltage is present. So I thought about PQ42, but it looks fine.

I'm thinking maybe the wlan card could play some role as the +3VRUN is shared with it. Haven't tried yet removing the card.

Otherwise the RUND that comes from PQ16 could be pulled low for some reasons ? RUN_ON is generated by U35, I believe it's determined by some voltage condition, but since all voltage seems ok (and the computer startup and enter bios) I would assume it's not one of the main voltage rails.
My understanding is that RUN_ON is generated based on other voltage rails and if for some reason one of the rail does not turn-on within the proper timing, RUN_ON is not activated and stays low. But then I would expect the POWER_FAIL signal to be triggered as well.

I haven't finished investigating, but if anyone already run into this issue, that would help.

[Grouickman]

OK, checked the complete schematic and it does not make any sense, +3VRUN cannot be missing, it's used in many places, including the allsyspg signal, without which I doubt the computer will even turn on.

I don't know how I manage to mess up the measurement, but I must have, there's no other way.

Now I'm thinking more about signal "PCIE_SATA_PDET#" which connects to the PCH and is generated by +3VRUN and Q6, with input taken from pin69 of the M2 slot (PEDET (NC-PCIe/GND-SATA)).
PCIE_SATA_PDET# will signal the PCH whether it's a sata or pcie drive in the M2 slot, which would cause the NVME to not be detected if the PCH is trying to access a sata drive.. makes sense I guess..

[Grouickman]

Just in case : I solved the issue, it was the PCH. There were some intermittent contacts with the PCB (cracked solder joints on some of the balls). Especially when the board was hot.

of course stupid me cracked the PCH die after de-soledring it from the board by dropping it on the floor...
Gave a shot at some sellers on Aliexpress, and I was lucky enough to receive a working PCH, re-balled it and soldered it, and it's a fix ! no more NVME disappearing for no reasons and no random shutdown when moving the laptop.

So just in case you have a similar issue, check the PCH

[mcplslg123]

I too solved similar issue on hp x360 by replacing amd ryzen5 apu.