Laptop overheating, then not booting anymore

Lappen

Estimable
Jun 4, 2015
2
0
4,510
I've had the MSI GT780DX for more than 3 years now, and roughly one and a half years ago it started randomly turning off while the computer was on load (mostly games which would point towards the gpu) exactly as described in this thread: http://www.techpowerup.com/forums/threads/msi-gt780-laptop-shuts-off-during-games.176063/

I did several steps that kinda helped with the issue (cleaning fan etc, applying new heat compound to cpu and gpu, running laptop on a cooling pad) but it only reduced the problem a little, it kept reappearing. After the shutdown the laptop would turn on again after 2-3 seconds, but without booting, ie the screen stayed black, the fan was running on full speed (probably to cool down the previously overheated gpu or cpu) and the harddrive was running as well. Usually after a while it would start the OS by itself, or after I restarted the laptop myself. If I restarted the laptop to fast it would just crash again after a minute or so. I monitored the gpu and cpu temperatures before such crashes and they were way in a tolerable range, something around 70-80°C just before the crash.
A funny thing to note is the fact that most of the crashes were happening during the summer when it was hotter and the air was more moist.
I followed this tutorial for applying new heat compound and cleaning the heat sink and the fan:
http://m.youtube.com/watch?v=5pvVTgN0xZM

Had the same problem with the inital heat compound that went over all the connections around it. I tried to clean it as best as I could, but not sure if that was causing the problem. The guy in the video is talking about the paste short-circuiting some connections, but Im not even sure if thats true firstly because the laptop was still working on medium load and secondly because Im not sure if the heat compound can conduct electricity.

This weekend I had the same issue as described above the first time again after a couple of months, but with a way lower load (just skype on fullscreen) and it did the same thing. I could restart the laptop after a while, but it would shutdown right when it reached the desktop after startup (probably because it wasnt cooled down enough). So I let it cool down over night, tried starting up again, and it crashed just after reaching the desktop. Now it wont boot at all, but just turn on the fan on full speed and run the harddrive. Screen is black. Same situation as before, only that cooling the laptop down over night didnt help this time. I tried the same steps again (cleaning fan, new heat compound), but no success so far. Next step I did was a gpu reflow following this tutorial: https://www.ifixit.com/Guide/Repair+VGA+card+by+re-flowing+solder+on+the+board/2240
Once for 6-7 mins with 195°C and once for 10 mins with 200°C. No success with that either. Luckily as far as I can tell, none of the plastic components on the card were damaged.

My guess is that the laptop gets a wrong reading on the temperatures of one of the components (most likely the gpu), and therefore starts up the fan on full speed trying to lower the temparature before booting (which obviously wont work). Might be wrong though.

Any ideas what the problem might be? What other steps could I try?

Some more info:
- Just this weekend I reinstalled windows 7 using a harddrive image I made on this laptop with TrueImage a couple of years ago. And there were some problems concerning graphics drivers. But in the end I reinstalled the nvidia drivers correctly and my laptop was working fine for a day or two, even under load of playing games. Might be a coincidence, but I only reinstall the os once or twice a year.
- Following some other instructions in another tutorial video I checked whether the issue is with the screen backlight (?) by shining a bright light at it during startup and seeing if I can find any symbols. I also put the RAM module into the second slot. Neither of these steps helped.
- My screen is only loosely connected to the laptop body because the connection at the hinges broke some time ago. Its been working fine so far though. Not sure if the screen broke first or the crashes appeared first. But since the screen is connected to the gpu it might have something to do with it.
- I tried starting the laptop without the heat sinks after removing all of the thermal compound. Same problem as before, but I noticed that the temperature of the gpu rises very fast while the only thing that is shown is a black screen. The cpu temperature only rises slightly noticably. I left the laptop on once for 30s and once for 60s in order not to damage the non-cooled chips while testing this.
- starting the laptop either without ssd (which contains the os) or without the hdd (only for storage) doesnt change anything about how it starts. Without cpu and gpu the laptop will not start up at all. With cpu and without gpu the laptop will start running the fan on full speed and starting the hdd like before. The only difference is that now the laptop will power off right away when I press the power button, in contrast to the 5 seconds that I have to hold down the power button for the computer to shut off when both cpu and gpu are connected. This might indicate that the laptop is in a similar state to when the laptop used to start when it was fully working before reaching the loading screen of the operating system. at this time pressing the power button would shut off the laptop immediately while it would take 5 seconds to power off the laptop once the os is reached. Its unlikely though that the os is reached right away and that this causes the 5 second wait time when gpu is plugged in. It probably just means that a regular shutdown is permitted when it powers off right away (no gpu plugged in) and that a regular shutdown is not permitted when it waits 5 seconds to cut the power (gpu plugged in).

 

Lappen

Estimable
Jun 4, 2015
2
0
4,510
Like I said, I already cleaned the fan and the heat sinks very thoroughly. If the problem were actually the temperature, it would only appear once a component is too hot. On startup all components are room temperature, and after 60 sec black screen the gpu is the warmest component, but still pretty cold.