Announcement

Collapse
No announcement yet.

MSI R9 390 Gaming 8GB - short on +MVDD

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    MSI R9 390 Gaming 8GB - short on +MVDD

    Hello,

    Please note I am a noob and all the actions described below are based on the experience gathered on the internet, mainly YT

    MSI R9 390 Gaming 8GB - motherboard model MS-V308 v2.3.

    The card has a short circuit on the entire + MVDD line, starting from the Q703 element ending with the MVREFD outputs under the GPU, but I will describe from the beginning what I did:

    1. At the very start, I measured the resistances on the main power lines, coils and checked the 0 ohm resistors. Everywhere ok, hundreds/thousands of ohms on each, well except the 0 ohm resistors(unfortunately I did not write down the values on the coils ;( ), I remembered 3.5 ohms on the GPU power supply only. Memory power supply - 0.2 ohm. At this point I have abandoned the idea to put the card into the comupter to measure voltage.

    2. I started with desoldering Q703 (bottom key?) Because the short appeared to start there. After desoldering, the short circuit remained and the Q703 turned out to be fine. At this stage I didn't have boardview yet so I ran with the meter here and there but found nothing obvious.

    3. I have desoldered the L700 choke and connected bench psu to the + side of L700. I set 0.8V on the power supply and turned on the power supply to see what will heat up. At 0.8V, the board consumed more than 5A, the power supply itself was overloaded and I did not notice anything heating up. GPU cold, memory cold, all cold on both sides of the board. For the moment it seemed to me that the U2600 memory module got a little warmer than the rest, but in the end I didn't know if it was due to the short or my fingers warming it. I gradually increased the voltage to 1.2V (yes, I know, a stupid idea when already injecting 5A), I still did not feel anything heating up, but after a while I heard a slight click so I turned off the power supply. As it turned out, the EC200 capacitor fell off and the C2623 capacitor was un-soldered on one side (it was unsoldered from the + MVDD side, not GND).

    4. At this point I came to the conclusion that I would desolder each memory module in turn(as they were all shorted), until the short circuit disappears (at this point still 0.2 ohm, and the board, despite falling off 2 elements still took over 5A). After desoldering the last out of 16 modules (U2000), a light appeared in the tunnel, resistance at + MVDD increased to 9.5 ohms measured at L700. Still very low, so I connected the power supply and at 0.8V it was only 0.43A, and at 1.2V 0.53A. There was still nothing hot, and I couldn't find any obviously broken element. At this point I realized that the GPU might have died as the resistance of the capacitors on the GPU was the same as on the L700.

    5. I desoldered the GPU, measured the resistance on the L700 and the short was still there, 9.5 ohms. I measured the capacitors on the GPU and the resistances increased to tenths and hundreds of ohms. After connecting the power supply still nothing is noticeably warm. I have decided to search for a boardview (that is how I found this forum btw ).

    6. After reviewing the boardview I found out that my short circuit is + MVVD (whatever this abbreviation means). So I plugged in the power supply again, this time I reached for the freezer and frozen everything under the GPU. Unfotunately no obvious results, but a few resistors clearly got warmer faster, even though they were still cold to the touch. These resistors are directly connected to the MVREFD pads under the GPU. I started to desolder each resistor on the MVREFD line one by one and measured the resistances after each. After desoldering the last resistor related to MVREFD, the resistance on L700 increased to 490 ohms, which seems to me to be the correct value. I went through the resistors after desoldering (they had low values on the board), the values looked ok with those from boardview with an accuracy of 1 to 5%. I soldered one of them for a test and the resistance automatically dropped to 110 ohm (the soldered resistor is 100ohms), and each GPU pad connected directly to any of the MVREFD resistor is the same value as the resistors. While back populating the resistors, the resistance got lower and lower till it reached back ~9 ohms.

    Sorry for long and boring story but I wanted to illustrate what I did as at this point, I have run out of ideas. Is it possible that the short is in the board itself? Is there anything else I can check? Did I overlook something?

    I have also chekcked the GPU pins voltage drop in PCI-E pairs to GND, all have about 0.87V drop so mayby the GPU is not dead yet.

    I have attached a pic with the current resistance values.
    Can I attach the boardview here as well?
    Attached Files

    #2
    Re: MSI R9 390 Gaming 8GB - short on +MVDD

    The resistor thing makes sense

    It sounds like all these 100 ohm resistors are basically in parallel - and they all feed some load that is approx 9 ohms (which is not a short buy the way)

    So as you unsolder each resistor, you are increasing the value of the remaining resistors in parallel

    As you are measuring from the supply end of the resistors (not the load end) the total resistance you read is going to be something like

    ~9 ohms + (100/the number of parallel resistors)

    This videos should give you all the expected resistances on a R9 390 8Gb. The GPU does not detect but it does power up correctly
    https://www.youtube.com/watch?v=DvPjTfO3CG0 (part 1)
    https://www.youtube.com/watch?v=5ILM2kvm3Vk (part 2)

    This video should give you all the expected resistances for a similar R9 380 with short circuit Vram and Mem Controller
    https://www.youtube.com/watch?v=ncsZsPdXnl8

    I hope that helps.
    Last edited by dicky96; 10-05-2021, 12:58 PM.
    Follow me on YouTube
    ------------------
    Learn Electronics Repair
    https://www.youtube.com/channel/UCFX...R8UZ2vg/videos

    Comment


      #3
      Re: MSI R9 390 Gaming 8GB - short on +MVDD

      Hi Richard,

      Many thanks for your reply and interest. It is nice to read some logical conclusions as I am easily confused with my results due to lack of knowledge

      Originally posted by dicky96 View Post
      The resistor thing makes sense

      It sounds like all these 100 ohm resistors are basically in parallel - and they all feed some load that is approx 9 ohms (which is not a short buy the way)

      So as you unsolder each resistor, you are increasing the value of the remaining resistors in parallel
      Ok, that idea came to my mind once I was poking around with it, but because I have an old, working R9 270 I measured the resistance on the memory and it is about 250 ohms. So this measurement misleaded me a bit thus I dropped the idea that it may be actually correct reading. I did not consider the fact that the 270 may have different resistors on this rail. Also, the IRF and the coil is desoldered, may these elements change the value?

      Originally posted by dicky96 View Post
      As you are measuring from the supply end of the resistors (not the load end) the total resistance you read is going to be something like

      ~9 ohms + (100/the number of parallel resistors)
      I am not sure if I do understand correctly but I think I did check all possibilities but without learning the lesson:

      The resistance on the L700 coil is (depends which spot of GND i put the probe on) ~9 to even 15 ohms, I think 10,5 is most common value. The resistance of the resistors themselves while they are in place on the board is about 34 ohms and ~100ohms while desoldered (as stated in boardview). The resistance on the GPU pads themselves is same as the resistors - 34 ohms each. There are 16 of them total. So 15 ohms would be the exact value.
      By the "load end" you mean GPU pads?

      Does that mean that I am diagnosing a fault that is not actually a fault?

      I now wonder why I have read 9 ohms on some GPU caps while it was still on the board and while off it reads 50+ ohms, this fact also caused me to think the problem is still on board. But maybe, the GPU simply added more parallel caps? or it is simply dead.

      Originally posted by dicky96 View Post
      This videos should give you all the expected resistances on a R9 390 8Gb. The GPU does not detect but it does power up correctly
      https://www.youtube.com/watch?v=DvPjTfO3CG0 (part 1)
      https://www.youtube.com/watch?v=5ILM2kvm3Vk (part 2)

      This video should give you all the expected resistances for a similar R9 380 with short circuit Vram and Mem Controller
      https://www.youtube.com/watch?v=ncsZsPdXnl8

      I hope that helps.
      I have seen the 3rd earlier already and watched the first 2 now. I haven't got a chance to measure voltages as the board was/is(?) shorted and I did not want to put it into the computer in thois shape.
      In the 3rd video, you mention that the resistance on memory PSU should be something about 50 or 100 ohms. Mine is 10 but with all chips desoldered, should it change(go higher) while I solder them back?

      As I am pretty sure that some of the chips will be bad, is it possible to check them with multimeter instead of soldering them bad and checking for short?

      Do you think I should now solder the GPU back as well as the mem chips and if the resisitance will keep at let say 10 ohms, try to power up the card?

      Comment


        #4
        Re: MSI R9 390 Gaming 8GB - short on +MVDD

        I have soldered back the GPU and memory and guess what? It WORKS!
        So it looks the culprit was shorted R47 - 10k resistor... I yet do not know if all is ok as I haven't installed drivers yet and the card is artefacting on uefi screen a bit but I am happy I have managed to make it that far...
        Thank you Richard for pushing me the proper way

        In case I would have memory issue, does anyone have a clue on how to diagnose it on Radeon?

        Comment


          #5
          Re: MSI R9 390 Gaming 8GB - short on +MVDD

          Well done getting it to work at all - even if not fully!

          I haven't had the time to play around with memory diagnostic software so can't answer this question.
          Follow me on YouTube
          ------------------
          Learn Electronics Repair
          https://www.youtube.com/channel/UCFX...R8UZ2vg/videos

          Comment


            #6
            Re: MSI R9 390 Gaming 8GB - short on +MVDD

            Hi everyone!
            I'm looking for a little help with graphics card.
            I'm a noob when it comes to to this and I only managed to fix 3 cards with shorts with voltige injection.
            I have an r9 390
            I had a short on 1.8v rail wich was a bad cap.
            Changed the camp now 212 for resistance.
            No the card will start in the pc with no display.
            No iv checked the vram coil and I'm getting 8.4 for resistance.
            What the resistance on the vram be on a gigabyte r9 390
            I also can find where I can start a new therd on the matter.
            Any help would be greatly appreciated

            Comment


              #7
              Re: MSI R9 390 Gaming 8GB - short on +MVDD

              Mine is ~10 ohm on memory and the card boots up and displays. Those cards seem to be very hot, thus high chance for fried GPU

              Comment


                #8
                Re: MSI R9 390 Gaming 8GB - short on +MVDD

                @floodster
                These videos should help you with the R9 390 - I think I measured just about every voltage rail resistance including Vram

                part 1: https://www.youtube.com/watch?v=DvPjTfO3CG0
                part 2: https://www.youtube.com/watch?v=5ILM2kvm3Vk
                Follow me on YouTube
                ------------------
                Learn Electronics Repair
                https://www.youtube.com/channel/UCFX...R8UZ2vg/videos

                Comment

                Working...
                X