Broken GTX 980M

Discussion in 'Hardware Components and Aftermarket Upgrades' started by Darker01, Nov 11, 2017.

  1. Khenglish

    Khenglish Notebook Evangelist

    Reputations:
    453
    Messages:
    668
    Likes Received:
    400
    Trophy Points:
    76
    The heat gun you selected will do the job fine if it works. I got one and it was DOA. The replacement worked for 1-2 years before the heating element blew. The replacement head for the replacement unit did not heat properly.

    I'm not sure what you mean by "but do you think if I get away with just using the hot air gun to remove the FET directly with sufficient heating of the surrounding area". You only want to use a heat gun. You should not be using an iron at all. You blow hot air directly on the component and board to remove and place a new component. There is nothing in the area that is significantly temperature sensitive that can be damaged by the heat.
     
    woodzstack and Darker01 like this.
  2. Darker01

    Darker01 Notebook Geek

    Reputations:
    31
    Messages:
    92
    Likes Received:
    58
    Trophy Points:
    26
    Thank you. It'll take at least another week for all the parts to arrive. I'll take a look at hot air reworking tutorials in the mean time.
    I'll let you know the results.
     
  3. NGX83

    NGX83 Newbie

    Reputations:
    0
    Messages:
    7
    Likes Received:
    4
    Trophy Points:
    6
    Last edited: Nov 23, 2017
  4. MahmoudDewy

    MahmoudDewy Gaming Laptops Master Race!

    Reputations:
    400
    Messages:
    1,583
    Likes Received:
    646
    Trophy Points:
    131
    That card in the thread you linked is running in my CLEVO machine atm. I wouldn't say adding the MOSFETs helps with power draw or higher clocks (That is still dependent on your chip), it just helps the card to sustain higher loads and not die as fast as it would without the MOSFETs.
     
    NGX83 and Darker01 like this.
  5. Darker01

    Darker01 Notebook Geek

    Reputations:
    31
    Messages:
    92
    Likes Received:
    58
    Trophy Points:
    26
    Happy Thanksgiving everyone. I'm back to provide update regarding the progress of the repair.

    TL,DR: 1 MOSFET is indeed shorted. This one had silicone oil on it where the V_SW and V_IN pins supposed to be. Still need flux and wick to clean the pads before soldering the MOSFETs back on. Might take another week or so.

    The W.E.P. 858D hot air rework station mentioned in one of my posts arrived on Wednesday. Popped it open and found that the thing wasn't put together haphazardly like some other 858D clones. Fuse's present and was hooked up correctly for the most part. Both the chassis and the metal casing on the heat gun was properly and securely grounded. There was a loose piece of broken plastic inside the heat gun case, and I'm not really sure there that came from. I guess it's a good thing I opened everything up to check. I was a bit worried about some strange magic smoke coming out of the heating element, but it turned out that I had a screw stuck in between the add-on tip and the heat gun's mouth.

    Flux is bound to arrive on Friday or Saturday, so I decided to start removing the MOSFETs and checking which one shorted. The 858D didn't explode, which was nice. This was my first time working with surface mount components, so needless to say it took a lot of trial and error to remove all 3 MOSFETs with the last one being the culprit of the short. Pictures are here.

    Only 1 MOSFET has V_IN and V_SW pins shorted to ground. Removing that one rid the short between the power pads altogether. The other 2 MOSFETs and the brand new ones did not have shorted pins, which is great I suppose. @Khenglish was right about 1 MOSFET being the issue. Nevertheless, I noticed that this shorted MOSFET had a noticeable amount of silicone oil on the package where the V_IN and V_SW pins supposed to be. Could the factory default thermal pads be the culprit?

    I think I'll clean up the pads and apply leaded solder to them before soldering the MOSFETs back. Not sure when the wick I ordered nearly 2 weeks ago from China is going to arrive.

    I'm not gonna OC the GTX 980M even if I managed to get the thing working again. I'm the kind of person that wants his things to last. That said, I'll still fill out all 6 MOSFET pads just to spread the heat.
     
  6. Khenglish

    Khenglish Notebook Evangelist

    Reputations:
    453
    Messages:
    668
    Likes Received:
    400
    Trophy Points:
    76
    Good to hear it was just a blown FET. The core should be ok.

    Too bad Radioshack no longer exists for flux. You don't need very good flux for soldering FETs, so you'd just spend a couple bucks and not have to wait.

    Btw, usually it's best to just get flux from a USA Ebay source. Amtech 4300 is usually the go to solder. Lacks nasty chemicals which sometimes show up in the China solders.

    Don't try soldering without flux. Heat transfer from the FET to the pcb will be terrible, so you could easily overheat and kill the FET.
     
    Darker01, woodzstack and Vasudev like this.
  7. Darker01

    Darker01 Notebook Geek

    Reputations:
    31
    Messages:
    92
    Likes Received:
    58
    Trophy Points:
    26
    Thank you. I already ordered Amtech NC-559-V2 from here since it should be genuine. This is the flux that's going to arrive on Friday.
    I sure hope that I didn't damage the 2 functional FETs pulled from the PCB. I was still getting used to the hot air station while removing those.

    EDIT: Apparently I can still request more of the CSD87350Q5D MOSFETs. I guess I don't have to worry about reusing the original FETs. Knowing how fast TI ships things I think I'll resume the project on Monday or so.
     
    Last edited: Nov 24, 2017
    Vasudev likes this.
  8. Khenglish

    Khenglish Notebook Evangelist

    Reputations:
    453
    Messages:
    668
    Likes Received:
    400
    Trophy Points:
    76
    If you put on new FETs on the original FET's pads then you'd need to add solder to the board. I'd just reuse the originals if their solder is still in the correct spots.
     
  9. NGX83

    NGX83 Newbie

    Reputations:
    0
    Messages:
    7
    Likes Received:
    4
    Trophy Points:
    6
    Ok. For now, with only 3 FETs, i'm pretty sure my gpu can't pull more than 130w without some blackscreens.


    Ok no problem. I'm waiting you to see how it's gooing with adding this FETs.
     
  10. Darker01

    Darker01 Notebook Geek

    Reputations:
    31
    Messages:
    92
    Likes Received:
    58
    Trophy Points:
    26
    Hello everyone.

    Flux arrived on Friday as expected, but the wick didn't. Regardless, I decided to proceed anyway with make-shift wick from a spare composite video cable. The wick wasn't perfect, but it helped getting rid of the extra solder on the center pads after I added leaded solder. I went through this to make soldering the MOSFETs easier since leaded solder melts at a lower temperature than the lead-free solder on the board.

    To the point: I added the MOSFETs, checked to see if the pins made contact, repositioned a few MOSFETs, rid excess solder with the soldering iron, wiped nearly all of the leftover flux off with IPA, dried the card with hot air, replaced the crummy thermal pads with Thermal Grizzly Minus Pad 8, installed the card, and booted the laptop.

    Laptop booted.

    There were a lot of things that could have killed the card for good during the past 5 months ranging from physical damage to ESD. It still amazes me that me with my lack of expertise in electronics and my janky setup somehow managed to get the card repaired. Overheating MOSFETs, stripping pads off of the PCB, burning surrounding components, blowing small capacitors into oblivion, jamming the soldering iron tip where it shouldn't be, not drying the card well enough, not pressing down the MOSFETs to squeeze out excess solder, giving myself 2nd degree burn, etc... were concerns that troubled me up to the point of booting the laptop up. I was prepared to be disappointed, but I guess setting the expectations low made seeing the laptop boots after 5 months all the more satisfying.

    After using DDU, I installed the driver. That got GPU-Z to recognize the card, and the specs looked about the same as other CLEVO GTX 980M cards.
    The next step was to check if the card is stable.I ran Heaven benchmark for about 5 minutes, and the card drew ~95-104W during the entire time. I didn't notice any artifact or anything unusual on the screen. At this point I decided to stop and have some food since I have been working non-stop for about 6 hours.

    I'll do more stability testing on the card in the near future, probably tomorrow. I'll occasionally post follow-ups test results here after that. I have yet to decide whether or not I want to sell this laptop to recuperate the cost of the Sager NP9870-S. As far as I concern the GTX 980M won't be accepted at the CLEVO repair center with its torn serial number and tampered PCB.

    Anyways, I believe thank-yous are in order. This repair wouldn't be possible without @Khenglish 's expertise with MXM GPU modification. His diagnostics was spot-on, and through that I saved quite a lot of $ by repairing the board myself (858D ~$50, soldering station $35 off of Craigslist, flux ~$25, FETs were free samples).
    I would also want to thank @Danishblunt and @woodzstack for suggesting replacement thermal pads. The Thermal Grizzly Minus Pad 8 is much more robust than the stock pads. Couldn't find anything to replace the pad on the row of MOSFETs though.
    As for everyone else, thank you for staying with me for the ride. It's one hell of an adventure going from knowing nothing about what caused my GTX 980M to fail to burning the card with Heaven benchmark.
     
    Last edited: Nov 25, 2017
Loading...

Share This Page