Maker Pro
Maker Pro

need help with Powercolor R9 390 PCS+

Chris_top

Mar 29, 2022
3
Joined
Mar 29, 2022
Messages
3
Hello everyone,

as the title says: I have a big problem with my GPU. After a short circuit ( ATX Powersupply damage ... ) my GPU stopped working.
I try to repair it.
I did find one clear defect, which I will eliminate first. The IC with the name U350, 8 pins is burned out. It is no longer possible to read the top writing on IC.
So I hope someone maybe has this GPU and is willing to help, find out what IC it is.

please see attached picture
 

Attachments

  • IMG_20220329_162541 (2).jpg
    IMG_20220329_162541 (2).jpg
    272.5 KB · Views: 6

kellys_eye

Jun 25, 2010
6,514
Joined
Jun 25, 2010
Messages
6,514
The device is probably a MOSFET and can be determined by careful inspection of the PCB tracks (post some pics) - however - such failures are never an isolated case and more usually point to other failures elsewhere in the system. If replacing that part fixes it then count yourself lucky but I suspect the fault is going to be terminal and a new GPU will be required.
 

Chris_top

Mar 29, 2022
3
Joined
Mar 29, 2022
Messages
3
thank you very much for your reply, I completely understand, that the chance, this is causing the problem, is very low.
More likely the main problem caused U350 to burn.
But since this is not very difficult and also cheap to replace, let's give it a chance.

finding out what kind of IC/Mosfet it is, by inspection of PCB tracks another level of expertise, which I cant match :)
Do you really think you could do so?
on the PCB there is no wiring visible. it seems that everything is in layers inside the PCB for this area. I made another picture. I also did some checking, on where the pins go on the PCI Connector.
RED is 0 Ohm
GREEN is 40 Ohm

I think the only way to come further is when maybe someone with the same GPU could tell me the IC model or make a picture of a working IC U350.
 

Attachments

  • pcb-min.jpg
    pcb-min.jpg
    432.2 KB · Views: 5

kellys_eye

Jun 25, 2010
6,514
Joined
Jun 25, 2010
Messages
6,514
There are a couple of clues that we can gather from the way the board is populated.

a. The device isn't a buck or boost converter as there isn't a suitable inductor close to the device.
b. Board markings indicate a 5V source on one side and 1.8V on t'other side.
c. The inference is that the device regulates the 5V (from the 78M05 close to it) down to 1.8V i.e. is a linear regulator.
d. We need to search for 8-pin SOIC linear regulators with a fixed 1.8V output.....

A couple of tests may confirm this assumption:

1. Check for continuity from the pin marked +5V to one of the pins on the 78M05 - this is the 78M05 input pin.
2. Check for continuity from one of the other pins on the 78M05 to ground (0V) - this is the 78M05 ground pin.
3. Check for continuity between the output of the 78M05 to the input of the blown chip i.e. is there a continuity path from one device to the other - this is the 78M05 output pin.
4. Check that the pin marked 1.8V has a continuity path back to the blown chip.

Note that you can get the actual pinout of the 78M05 device as a datasheet online.

If you can 'prove' 1-4 above then we can infer the type of chip that has blown and, maybe, do a parametric search for a similar or even identical device that we can then 'prove' (yet again) by comparing the connections (traces) on the blown device to the actual functionality of the pins on the potential replacement device.
 

Chris_top

Mar 29, 2022
3
Joined
Mar 29, 2022
Messages
3
Hello,
I did the recommended things, here are the results:
the 78M05 seems not to be connected anyhow to the blown chip: connected to 5v, the middle pin is not connected at all, pad is ground, the backplate is ground, of course.
Blown chip pin 1,3,4,5 are connected to 1,8 V
 

Attachments

  • new-min.jpg
    new-min.jpg
    512.2 KB · Views: 3
Top