Need insight on r9 380 MOSFETs (1) + GTX 950 memory related electronics (2)

[chola95]

Hello, i have 2 separate mysteries that can be summed up in one thread, and i'm pretty new to electronics so i figured you guys could help me out.

1) Asus Strix r9 380 2gb - during Furmark stress test VRM Temperature in hwinfo64 goes up to 130 C when it throttles for it to cool off, during normal gameplay it stays under 100-110 C. Undervolting helps. Changing VRM thermal pads didn't really help (although new pads are trash quality, blue silicone ones). Adrenaline drivers have weird blue/yellow line artifacting in CSGO (undervolting/underclocking doesn't help here, also VRM temps are normal), while older Crimson drivers work normally.
Upon opening i saw that MOSFETs on this card seem like they have their pins soldered together (pic).
Now, i am an amateur but i know in theory how to fix that with a soldering iron and flux, but my question is - can this be the cause of high VRM temps and instability with newer drivers?

2) MSI GTX 950 Gaming X 2gb - card boots and shows picture with mobile artifacting (pic 1), driver seems to install successfully but Device Manager shows error (Code 43), and i forgot to check GPU-Z to see if it shows 2gb memory or something else.
On the back of the card it seems like 1 component is missing and someone did some sloppy soldering work near a memory chip (pic 2). On the internet (pic 3) i saw that it seems to be an SMD capacitor.

A few questions here:

• Is it possible to fix artifacting by soldering missing component here? (Can missing cap cause moving artifacts? Is the PCB not too damaged for this procedure?)
• Is the memory chip damaged? (i see discoloration probably from soldering iron, but don't know if it's transistors are damaged or only the protective layer)
• Is there a way for me to find out which exact component (and it's value) is missing here? (maybe by measuring something else on this card; the present SMD right beside the one missing)

Thanks everyone in advance!

 

[RazorWind]

Hello, i have 2 separate mysteries that can be summed up in one thread, and i'm pretty new to electronics so i figured you guys could help me out.

1) Asus Strix r9 380 2gb - during Furmark stress test VRM Temperature in hwinfo64 goes up to 130 C when it throttles for it to cool off, during normal gameplay it stays under 100-110 C. Undervolting helps. Changing VRM thermal pads didn't really help (although new pads are trash quality, blue silicone ones). Adrenaline drivers have weird blue/yellow line artifacting in CSGO (undervolting/underclocking doesn't help here, also VRM temps are normal), while older Crimson drivers work normally.
Upon opening i saw that MOSFETs on this card seem like they have their pins soldered together (pic).
Now, i am an amateur but i know in theory how to fix that with a soldering iron and flux, but my question is - can this be the cause of high VRM temps and instability with newer drivers?

View attachment 419450

Short answer: No.

Long answer: Those three pins are actually all one pad on that type of transistor. What might fix your overheating problem is high quality thermal pads AND a better heatsink AND better airflow over them. Or a water jacket. That would work too.

2) MSI GTX 950 Gaming X 2gb - card boots and shows picture with mobile artifacting (pic 1), driver seems to install successfully but Device Manager shows error (Code 43), and i forgot to check GPU-Z to see if it shows 2gb memory or something else.
On the back of the card it seems like 1 component is missing and someone did some sloppy soldering work near a memory chip (pic 2). On the internet (pic 3) i saw that it seems to be an SMD capacitor.

View attachment 419452 View attachment 419453
View attachment 419454

A few questions here:

• Is it possible to fix artifacting by soldering missing component here? (Can missing cap cause moving artifacts? Is the PCB not too damaged for this procedure?)
• Is the memory chip damaged? (i see discoloration probably from soldering iron, but don't know if it's transistors are damaged or only the protective layer)
• Is there a way for me to find out which exact component (and it's value) is missing here? (maybe by measuring something else on this card; the present SMD right beside the one missing)

Thanks everyone in advance!

1. "Artifact" is not a verb. A card does not "artifact." It presents visual artifacts in the output image. Answering your question, yes, the missing capacitor could potentially cause the card to produce visual artifacts in the image.
2. It could be, but probably not. That's probably just flux residue. Only way to tell is to fix the missing cap and test again.
3. Best way is to remove the corresponding part from an identical card. Another possibility is to guess based on the voltage of the circuit and physical size. In that last picture, there is a row of three MLCs to the left of the bottom memory IC. Check and see if those are connected to the same circuit as your missing cap. If so, unsolder (with hot air, not a soldering iron) the one on the far left, order two more caps of that size and capacitance and 2V rating, and solder them on there. The capacitance doesn't necessarily have to be exact for those noise filtering caps. It just has to be enough.

 

[chola95]

First of all, thanks for a thorough reply, that's exactly what i needed to hear!

Those three pins are actually all one pad on that type of transistor. What might fix your overheating problem is high quality thermal pads AND a better heatsink AND better airflow over them. Or a water jacket. That would work too.

Yeah i figured that out after asking on some other forums and looking at some bare-bones PCB pics on the Internet, didn't really know that was a thing (like i said, i'm new to electronics). Sadly, the only fiscally viable thing i can do with this card is putting better pads on the small VRM heatsink or undervolting, or just leaving it as it is since in real life workloads it doesn't really overheat.

1. "Artifact" is not a verb. A card does not "artifact." It presents visual artifacts in the output image.

I understand that, but English is far from my first tongue, it's a tech jargon in my native language. But i'm glad you corrected that, it shows you're serious about answering precisely

2. It could be, but probably not. That's probably just flux residue. Only way to tell is to fix the missing cap and test again.
3. Best way is to remove the corresponding part from an identical card. Another possibility is to guess based on the voltage of the circuit and physical size. In that last picture, there is a row of three MLCs to the left of the bottom memory IC. Check and see if those are connected to the same circuit as your missing cap. If so, unsolder (with hot air, not a soldering iron) the one on the far left, order two more caps of that size and capacitance and 2V rating, and solder them on there. The capacitance doesn't necessarily have to be exact for those noise filtering caps. It just has to be enough.

For starters i took it to an experienced electronics repairman and he told me pretty much the same story as you did, but he added that soldering another SMD in the place of the missing one is probably not possible since there's nothing to solder it to (one pad is ripped out as i understood). He also mentioned that the memory chip looks like it was tampered with. He gave me a deal: 0$ if he doesn't fix it, and 30$ if he does, that's pretty reasonable considering prices right now (at least in my country). I don't doubt him, had pretty good experience so far with the guy.
Anyways, as far as measuring things and fixing them myself, i will definitely use it as a hobby/training wheels card while measuring things you said, and other tutorials i stumble upon online.