BSOD P650se ntoskrnl.exe ntkrnlmp.exe on power supply solved (sorta)

Hi all!

This is half of my own diagnostic of the issue, half question if anyone has come across this
-and half just-so-its-out-there if anyone has this issue, there's a hit if searching for such.
(scroll to the bottom for my work around)
-----------------------------------------------------------
So, P650se bought second hand really cheap, strangely mint condition with not a spec of dust or fingerprint,
(bar the Kensington lock being obviously used) .....with a "graphics issue"....
-and a cheap replacement Ac adapter

Re-sold as a Venom BlackBook 15 in Australia
( the Clevo sticker code doesn't come up on they're system, and they refuse to acknowledge this as one of theyres, )

I spent about a week screwin with this, so some steps may not be in actual order.

At first, it would randomly freeze and reset after at least an hour of intensive gaming, or not at all.
Fan on max, lights flashing, beeping, you know the deal.
-But no BSODs, just freeze, Clevo's "your a winner" pokie machine song and dance and then reboot.

So assuming it was a cooling issue, id run the fans on max using Fn+1, annoying but all good.
Decided to re-paste CPU and GPU and a board inspection. No dust, no damage, no burns,
no bulging capacitors.
Nothing, Cool, I had a week of no crashing.

Updated to Nvidia 381.89.... and the BSODs start up, ntoskrnl.exe & ntkrnlmp.exe.
And this is while just at desktop on the intel display adapter.
I uninstall Nvidia drivers using DDU, reinstall the old original off the Clevo site, still bsod'n same codes.
Did the intel driver with Nvidia still active, still same codes.

Reformat, disable Win10 Updates from services, Nvidia driver only.
Taskbar is unresponsive, bsods and reboots. I install every driver but the Nvidia driver-
Allgood.

I try 8 Versions of Nvidia drivers for the notebook GTX970m using DDU in SafeMode everytime.
Same codes, different amount of time before bsods and different windows issues.
If the Nvida driver was disabled from Device manager, there was no issues.

Tried every combo of Nvidia Control options and the ControlCentre options to no avail.
Same with Windows power options.
Tried flashing Prema's Mod V2 Bios, all sorts of settings, no difference,
( hats of to Prema BTW, love your work man!)
tried the stock Clevo Bios, no difference.
There's only one kind of GPU Bios for these soldered on GPUs so id didn't bother with that. (Should i?)
Onto hardware...

Memtest. 11 passes all ok.
SSD Raid checked out ok.
HWMoniter had all idle temps very low.
GPU-Z had stable readings, odd graphs,
Nvidia Inspector had the same odd graphs.
((the graphs seem to oscillate in a squaretooth, from 0mhz/135mhz to 938mhz/1038mhz from memory
every 3-5 seconds depending on which Nvidia driver is used.
Im not sure if this is a normal thing with shared graphics so could someone clarify?))

Checked the PSU voltage, 19.45v-19.55v
Unfortunately the cheap multimeter i could find cannot check the amps.

But im not sure with laptop PSU's feeding amps, if its a constant and the lappy only draws what it needs,
and it seems a bit hard to measure that under load with the rear cover off.
The PSUs ive come across that have crapped out had either worked or dont, making various high pitched noise or not, and within tolerance volt oscillation.

I then realized without being plugged in, i can run games on battery alone but obviously it throttles as the battery cant maintain such draw compared to mains.
So i open her up and disconnect the battery from the MB and boot her up.
Same BSOD's, so its not the battery, and HWMoniter has it at 7% batt wear.

So at this point im thinking this was an ex-demo model due to the lock wear, the mainboards power management is damaged in some way, the GPU lucked out on the silicon chip lottery/ has taken damage, or its all been overclocked and not been a happy lappy.

Maybe the PSU is dropping amps or voltage under heavy load when needed, driver crashes due to not enough amps or volts, or maybe the PSU has a lower end oscillation in current and its playing funny buggers with the MB or GPU in its low power state.*
Maybe a factory defect, inhouse privately sold and re-sold until I gut my grubby hands on it.
Maybe I don't know what I'm doing.
Maybe it just needs to be held and told its loved.

Ahhwell, i got 16gb of ram, a clean chassis and a 3k screen for bugger all.
It was a gamble to begin with, *but I'm leaning towards dodgy PSU.
Speculation makes me smoke and drink coffee.
Coffee makes me smoke and speculate..... help.

------------------------------------
The work around:

I work in live music and used to be in LX, and my mate had a similar trick with one of the DMX distros that would always crap out and not receive data from cold start.
Plug up for a min or two for the capacitors to hold charge, get ghost power, unplug quick,
get a data feed goin, plug back in the mains. Data is feeding out.
Sorta the same principle, but not.

So get this:

If i unplug from the mains, enable the GPU in Device Manager, boot up a game so it has switched the GPU into full power, plug it in, game runs as it should and doesn't crash!
Before exiting the game, unplug, return to desktop, Device Manager to disable Nvidia driver, plug in.
All good, will not crash.
I can repeatedly do this and its stable, annoying yes, I now have a Device Manager shortcut on my desktop
and a usable gaming machine for cheap.
>>This will only work when the battery is full or near full, otherwise Nvidia drivers will crash to desktop, it will recover but by that point it will need a hard reset as the fans are on max and she is beeping.


Winner.

I have yet to try this with Photoshop, PremPro 2 or any rendering software, im not sure this will work
as your effectively still in Windows, with the whole sometimes-Intel-sometimes-Nvidia ordeal, but it works for gaming.

So right now, im runnin the stock Clevo Bios
(I think PremaMod V2 has no issues with THIS P650se but im done with testing for the moment)
Nvidia 378.92( 381.89 still random BSOD with
Intel 4600, 20.19.15.4624
All drivers updated manually (and Airplane mode hacked driver
And a suspect PSU which i can hear a high pitch in my left ear as i type....


Thoughts anyone?
Anything simple and obvious i missed or didn't try?
Anyone have something similar?
Google comes up zilch for such a scenario/workaround

Il force some new minidumps soon as i forgot to backup them up before i recently formatted with my working combo. Same with the screenshots of GPU-Z, if you guys like il get them happening.

I would happily receive an opened hand downward *****slap if anyone can get this was to work as designed.
Cheers!