Was playing a game and surprisingly I had disabled all Overclocks the others day as I felt a throttling issue was occurring when gpus reached 100% Usage so I was testing stock clocks (premas vbios)... All the sudden my PC just crashed I turned it back on and u coukd see this burning smell so I turned it off opened up and smell was coming from GPU 1 so I repasted it and loaded back up... Smell still persists ran a GPU Benchmark and the smell was strong... The GPU shows up and runs but smells like it is burning and the smell gives me a horrible headache. Any ideas on what to do?
Sent from my ONEPLUS A5000 using Tapatalk
-
Errr, let Dell know?
What else could you do?
Sent from my SM-G935F using Tapatalk -
Falkentyne Notebook Prophet
Smell giving a headache?
Sounds like a capacitor burned (like a filtering cap). Did you check for any blown caps? -
which thermal paste you are using. if liquid metal double check if it is not leaking and shorting anything if not check for metal partial between components.
-
Sent from my ONEPLUS A5000 using Tapatalk -
Sent from my ONEPLUS A5000 using Tapatalk -
Sent from my ONEPLUS A5000 using Tapatalk -
These help at all? Running benchmark for a good minute or 2 the pc will shut off...
Sent from my ONEPLUS A5000 using Tapatalk -
grab a needle and try to clear the debris between transistors on gpuPete Light and TheDantee like this. -
Sent from my ONEPLUS A5000 using Tapatalk -
You should use some electrical insulation tape when applying liquid metal like @mossesmos said, pretty sure some transistors are in contact with the liquid metal.
Take a screenshot of the contact surface of the heatsink too if you can. -
Sent from my ONEPLUS A5000 using Tapatalk -
-
Falkentyne Notebook Prophet
Ok first of all, that is WAYYYY Too much liquid metal to be used without any sort of protection whatsoever! WAYY too much. If you have full protection, it's okay to use a bit more than necessary, since then it can't go anywhere, but the fact that you see pools like that mean you used too much. Now I do not know whether it leaked out and got shorted on something...did you see ANY sign of ANY liquid metal balls that were *OFF* of the GPU, and anywhere around the little SMD contacts? Any at all? Because I would hate to have you waste time on that and then you find that some completely unrelated component wound up dying, which just happens to be one of those red herrings, basically.
In any event, the right way to rep for liquid metal is basically the following, which can be used safely IF YOUR HEATSINK HAS A GOOD FIT ON THE CHIP(S), meaning full complete even mount and balance. Avoid ANY LM on any badly fitting heatsink, PERIOD.
1) EITHER:
a) 3 coats of transparent nail polish, cellulouse based (look in the ingredients) over the SMD resistors--the OCD way is to do a coat over, wait 5 minutes, do another coat, wait 5 minutes, then do a third. The "I'm in a hurry" way is do a complete coat, then start the second coat right away, then the third coat. I've seen no case where the hurry way failed to insulate.
or b) Scotch Super 33+ or Kapton tape (Do NOT use super 88 tape, too thick), and tape completely over the resistors and SMD's, all the way up to the edge of the CPU or GPU surface and make sure there are no gaps in the tape or "air holes.". This does the job well, but some people swear by nail polish. And using cheap, bad tape might wind up losing adhesion and thus losing protection.
You do not need do both tape and nail polish. That's obsessive. Just do the nail polish if you feel that way.
For BGA CPUs and delids, make sure you also nail polish any exposed circle traces or marks too, even if there doesn't seem to be something on top of it.
2) Then, Foam Dam trick (Free insurance) courtesy of @tijgert . Smart airline pilot who was the first to think of this.
The foam dam trick is *NOT* EVER to be used in substitution of #1. Do NOT EVER use it in place of nail polish or tape. The purpose of the foam dam is to stop LM runoff from escaping the entire CPU or GPU area and getting onto the MXM or mainboard PCB, where it will then for sure turn into a conductive ball of doom, going where it pleases and taking out what it pleases. Foam dams will trap any spare LM runoff, so it has absolutely nowhere to go. The foam dam trick requires that you get foam that is dense enough to fully block everything when compressed down, AND is compressible and light enough to compress to literally MICROMETERS. Basically it has to be compressible to the width of a few paper sheets, because you don't want resistance from foam causing heatsink pressure issues and thus higher temps! Finding such foam isn't easy, but a hardware or hobby craft store should have something. Some people suggested possibly thermal pads, but they have too much thickness to be used for something like that and will just add resistance, and would be worthless for foaming away LGA processors too.
Mr Fox has a video of the foam dams he used early in this video here:
Here is a good picture of foam dams at work on LGA CPU's and BGA GPU's together. That Liquid Metal ain't going anywhere.
3) then you usually want to 'paint' the heatsink side with a coating also (especially since copper heatsinks will absorb and react with some of the LM, so its good to have a coating on that for good measure right away. Even for the Intel IHS underside (For IHS delids) putting a coat of LM on the underside of the IHS has given people lower temps than just applying it on the CPU die.
Doing all of that work should stop LM for ever being at fault for a laptop going boom.Ashtrix, ll_r1d0_ll and TheDantee like this. -
Sent from my ONEPLUS A5000 using Tapatalk -
Sent from my SM-G935F using TapatalkTheDantee likes this. -
With the both cards having the tape around the transistors during a benchmark using 100% off the GPU the screen just blacks out
Sent from my ONEPLUS A5000 using Tapatalk -
Would like to remind everyone helping... That the gpus have been swapped so this is not an issue with the GPU but an issue with the gpus being in slot 1... I have cleaned the heatsink and still Nothing... Applied tape to transistors (raised idle temp from 30 to 45... But worth it if it's makes the pc work properly) and no it's not a temp issue it crashed at 81C last time.... Cards run fine on desktop and browsing web but when u use them for a game or benchmark they die... Any ideas?
Sent from my ONEPLUS A5000 using Tapatalk -
Falkentyne Notebook Prophet
Remove one of the two video cards and use it as a single card only system, then overclock the card like usual and run the battery of tests then switch cards. This will of course require repasting; never re-use old paste when its mounted to a copper heatsink. Also one reason for high temps with LM is from not completely and fully cleaning the heatsink of any dried or oxidized LM remnants after removing.
If there's a problem with the slot itself, buy some Deoxit D5 and clean the slot contacts with a spray of that stuff. As far as I know, it does not have to be completely dried up before using the device again but don't quote me on that.
If even that fails, you could have a blown VRM on the mainboard that delivers power to the MXM...and that's bad.Ashtrix, Papusan, Pete Light and 1 other person like this. -
Sent from my ONEPLUS A5000 using Tapatalk -
-
Falkentyne Notebook Prophet
Yeah I know right? Usually you hear about blown VRM's on the video cards themselves. But again if SWITCHING the cards doesn't help..
I STILL say that ONE CARD has to be removed and the system should be run as single card only. I do NOT mean keep both cards installed and disable one, I mean physically REMOVE a card. Then overclock the card to the max stable and bench it. This itself will eliminate problems with the actual primary slot and its power delivery system, and that's important.TheDantee likes this. -
Sent from my ONEPLUS A5000 using Tapatalk -
Maybe next time you pull it apart, have a smell of the motherboard - follow your nose (literally) to find the area where the bad smell is coming from? Sounds dodgy but it can help - I've done more times than I care to remember
But based on the description, sounds like either a VRM or blown cap issue.
-
Sent from my ONEPLUS A5000 using Tapatalk -
Probably a component on the motherboard.
-
-
-
Sent from my iPhone using Tapatalk -
Sent from my ONEPLUS A5000 using Tapatalk -
Sent from my ONEPLUS A5000 using Tapatalk -
Testing what works and what doesn’t and will be parting out the computer. Using that money to build a desktop because the laptop scene is screwed.
Ashtrix likes this. -
Sent from my ONEPLUS A5000 using Tapatalk -
-
Sent from my ONEPLUS A5000 using Tapatalk -
I can almost assure you, that your board is burning a mosfet near the DC jack, because I’ve had two Alienware 18 boards do this and a couple more Alienware 17 boards do this as well. Not robust systems at all, m17x/M18x were much better built.
-
Sent from my ONEPLUS A5000 using Tapatalk -
Ashtrix likes this.
-
Sent from my ONEPLUS A5000 using TapatalkLast edited: Jan 15, 2018 -
-
Sent from my ONEPLUS A5000 using Tapatalk -
-
-
-
Sent from my iPhone using Tapatalk -
-
Sent from my iPhone using Tapatalk -
Falkentyne likes this.
-
TheDantee likes this.
Burning Smell
Discussion in 'Alienware 18 and M18x' started by TheDantee, Jan 13, 2018.