MSI GL75 9 Se Nvidia GPU code 43

[totoFR37]

Hello guys.

I own an MSI GL75 (i7)
Motherboard : 17E51
GPU : Nvidia 2060 RTX

I spent weeks to install and clean the drivers but still have an code 43 and the Nvidia GPU not working.
When using the GPU-z the GPU is recognized but there is no clock frequency.
Seems the problem is from the hardware.

What do you advice me to check ?

Thanks

[totoFR37]

I made some test yesterday night, board on the bench with no power supply on it.

I found some short on :
_ FB VDDQ
_ NVVDD

All the caps of these 2 lines are on short.

From what i understand, that seems to be 2 high power supplies rails for the GPU and the Rams, right ?

I didn't found any short on the 1V8 line.

So what i did is to remove the choke 4, 5, 6 (on NVVDD line) to separate the mosfets from the power line. the short still on the line but there is short at the mosftet output.
I also removed the chock 9 and choke 10 (on FBVDDQ line) and the result is similar.

I feel strange that the 2 power line are in short in the same time.

I didn't check the PEX_VDD

The GPU is the common point between these 2 lines.
The Rams only use the FBVDDQ. So i'm a bit afraid that the GPU is the problem.
What do you think ?

some stuff that i don't understand, is how GPU-z is getting the informations.
If the GPU was dead, how the software can get the reference (BIOS revision not available and no clock frequency also)
But we can see the RTX2060 in the windows manager even with the code 63.
maybe a part of the GPU is still responding and then my problem is maybe only one of the Ram in short ?

How can i check the rams one by one without removing them ?

Please help.
Thanks

[totoFR37]

sorry made a mistake "the short still on the line but there is NO short at the mosftet output".

thanks to guide me.

[Sephir0th]

You should better stop talking in puzzles and instead share clear measurements of clear locations. This includes to explain in detail what value you got from what location exactly. And when you don't feel confident to explain in detail, then share pictures as least

[totoFR37]

Ok sorry it's maybe a bit confuse. Just starting to learn, so i am throwing what i'm finding until i can know enough and connect the data together.

So starting again from Zero.

MSI GL75 9Se have an additionnal Nvidia RTX 2060 GPU.
This GPU is not working.
Using GPU-z we can't see any BIOS or Clock frequency. But the GPU is recognized.




Windows indicate in the device manager " this device have a problem, must be stopped" with a code 43 indicated.
We have first try to update all drivers and GPU bios like it's explain by Nvidia to solve this problem that seem to be really common these days.
After weeks to try every thing we can on the software side, i've decided to open the computer and make few checking.

As i don't really know how the component are working together, i have to start somewhere.
So basically i start now by checking the supply lines and finding short circuts.


I found shorts on all the caps that are connected to FBVDDQ.



this is connected to the Rams.
Removing Choke 9 and choke 10 show me that the short is on the GPU / Capacitor side, not on the Mosftet IC POQ25 and POQ22.




Same finding for NVVDD all caps connected to this output are in short circuts.
So i removed the choke 4, 5, 6 and similar behaviour : the short is not on the IC side U42,43,45, but on the GPU side, that is still connected to all the caps.

I'm not sure that is possible to have caps burned on these 2 lines in the same time.

So i need now to understand where the shorts come from (capacitor ? Ram ? GPU ? )





Seems on the schematics (available on this forum) that there is no link between these 2 supply lines. The only common point is the GPU if i'm not wrong.

But i don't understand how does it work, why GPU z is able to read some information and not some others if the GPU is not working.
Could be a part of the GPU that died and it lock itself the clock ?
have no idea.

I found also a "fuse" line and a start IC on the schematics.
I think that should manage how the GPU and other stuff start all together, but didn't check that at the moment as i found these shorts.

If you can guide my how to go head, i could learn on this interesting problem.

thanks

[Sephir0th]

I got that, but again, share a measured value instead of saying "SHORT", since the Main GPU supplies are (partially) "shorted" per default.

So these are the first lessons for you:

- Never say "short" but share a value in Ohms
- GPU and CPU Core supplies can read below 1 Ohms, respectively 3 Ohms without any problem in completely healthy condition.

So it is actually possible that you're hunting "nothing".

[totoFR37]

Ok sorry I understand, just came back home.

FBVDDQ is 33 ohms(probably not a short then)
NVVDD is 0.4 ohms

I check my multimeter measurement cables and they are 0.2 ohms themselves.

[Sephir0th]

Ok sorry I understand, just came back home.

FBVDDQ is 33 ohms(probably not a short then)
NVVDD is 0.4 ohms

I check my multimeter measurement cables and they are 0.2 ohms themselves.

These are perfectly fine measurements for a recent GPU Core in combination with Samsung VRAM. So yes, this all is no issue at all, or let's say not conclusive. At least not regarding the resistance to GND.

if shorts to GND can be excluded on all the other secondary rails supplying the GPU too, then i guess Mats/Mods is the way to go from here and some serious skills in soldering BGA chips, with at least the size of the VRAM chips.

[totoFR37]

ok thanks for your analyze and advices.
i will first check the other rails to be sure.

I just had a look what is Mats / mods, ok will try that also and let you know.

thank you.

[totoFR37]

made some checking this week-end :

I have the following voltage :
12.0 V
5.14V
3.36V
1.8V

I don't have anything on the 1V8_AON that seems related the PGOOD and FUSE circuits.
the gpu is hot.

when i go the the device manager and switch off the Nvidia, the voltage became 1.25V on the 1V8_AON.









i don't know what to think about these tests.
seems that nothing happen !
is it coming from the mats/ mods revision ?
Or is it because my GPU is not starting ?

Any advice ? thanks guys.

[totoFR37]

I have a tought ... seems that on a Laptop the Vbios is integrated in the same bios chip than the motherboard is.
So i wonder why GPU-z is not finding the Bios revision.
GPU Z is just reading in the Bios chip i suppose.

If the GPU was not starting, then it will not be recognized, but we can see it.
The Vbios is not related to the GPU itself.
It something strange that i don't understand.

If someone can guide me ...
thanks guys

[totoFR37]

More measurements yesterday :

PEX_VDD : 1.06 V
1.2DIMM : 1.25V
PBVDDQ : 12 V
PWR_SRC_NVDD : 1.31V
NVDD : 0.9 V

Still missing some voltage, they seems to be on the other face of the PCB (under) so i don't know how to switch on the computer and make this measurement on the same time.

I think the GPU become hot very quickly, i'm not sure that's normal.

So the tests i want to do are :
_ is the GPU have the autorization to start ?
_ are the RAM have power supplies working properly.
_ is there any short on the GPU input/output that make it become hot.
_ is the bios have been manipulated for overclocking ? (seems there is no way to tune it directly)

[totoFR37]

I just soldered some wire on the back of the motherboard in order to try to check few more voltage.

FP_FUSE : 1.84 V
1V8_ AON : 0V
FBVDDQ : 1.31V

If i understand well :

FBVDDQ is ok that's mean the Vram have power. is 1.31V the correct voltage ?
1V8_AON is used with many GPIO, i'm not sure for what it is used. Some kind of Alert when it's down ?

All the others are working great, PEX VDD, FB-VDD, 1V8, 3V3, NVDD

Please give me some light in this dark tunnel

[Sephir0th]

Without 1V8_AON the GPU shouldn't turn in at all. To what schematics are you referring and where have you measured in detail?

[totoFR37]

Thanks for helping.

I took 1V8_AON ( n°5) and FP_ FUSE (N°1) on U 14.
this is very close to the GPU on the back of the motherboard.

schematics come from the 17E51-Rev 1.0 MSI motherboard.

A lot of things are connected to 1V8_AON. i don't really understand the goal of this signal. Seems that it could "down" by a lot of the GPU I/O in case of problem.

resistance measurement of 1V8_AON is 88kohms to the GND.

[m1ch43lzm]

Maybe check the chip which switches 1_8VSUS to 1V8_AON
U53 SLG7NT402V

[totoFR37]

Hi.

ok on U14 i have all the volatage around, and i have 1V8 AON working.
in fact i didn't check correctly before, and i have the 1V8 working.

So .. seems that i have all voltage working properly.

the GPU is heating as soon as it start and the fan are running full speed also (not always)
so i'm not still sure if i had to go to the conclusion that the GPU is the problem ?