R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

[sigmeyer]

So i got this graphics card in order to try and fix it.
I plugged it into my personal computer just so to see how it would react and poof i saw alot of smoke come out ... luckly it didnt damage anything in my system.
Next i took it apart and started taking off some mosfets, 2 mosfets shorted other 2 were fine with a resistance across them around 350 ohms in diode mode , which i think is perfectly fine.
Now , my stupidity comes in , and i dont check for more shorts on the 12v psu connecter ,needless to say i plug it in again , just to see if i got rid of the short , and for the first time in my life i saw flames come out of the computer case. Luckly again , it didnt damage anything in my system. I took it out and i found that the 8pin connecter is shorted with 2 ohms across ground , the 6 pin connecter is shorted with 38 ohms , which i dont know if thats normal or not...
I took off some caps to see if they were shorted , but no , they were fine.
So now im looking for some pointers , what do you guys suggest i take a look at ... quite a few ceramic capacitors are showing as complete short (around the connecter site) , should i just start taking things off?
Oh and by the way , quite a dramatic crater was created where that mosfet exploded... So any suggestions? Here are some pictures.[/IMG][/IMG]
So thats where im at.

[piernov]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

This is dead. It burnt through several layers. Probably needs a hole through the PCB to get rid of all the carbonized material that's conductive.
The 2 lower phases are destroyed and it's impossible to rebuild them in my opinion. I see that there are 2 empty phases so there's a small chance that it's possible to populate these instead, but I wouldn't count on it. In any case the damage could be affecting something else than just the bottom 2 phases (even the 3rd one is a bit damaged).
GPU itself could well be dead.

[sigmeyer]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

Would it be difficult to rebuild those upper phases? also is this carbonated area causing the short of the pcie psu connecter? I have seen videos of people doing the rebuild... it can be quite tricky, but cant i just use the driver from the first 2 destroyed phases on the upper ones? assuming they are not damaged? If not im just gonna keep it for parts in case another r390 comes along

[sigmeyer]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

Im thinking of driling this 2 first phases out ..just to see if the short goes away. Should i just drill where those 4 mosfets are or should i go a bit in a larger circle/area?

[momaka]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

This board is beyond repair.

Original issue was likely caused by either a ceramic cap going shorted somewhere or the GPU chip itself shorting out... after which, causing MOSFETs to short out too.

Just put it in your junk bin for parts.

You can re-use many of the ceramic caps from the board, after you remove and test each one (has to be tested out of the PCB for proper capacitance and ESR, as well as not show short-circuit.)

Other things you can re-use from this board are possibly the RAM chips and maybe some of the MOSFETs... though in all honesty, there's a good chance both the RAM and GPU chip got damaged when the MOSFETs shorted, so maybe not best to use these unless you have nothing else and want to experiment.

The PCB itself can also be useful to learn BGA or test BGA profiles if you have a rework station.

And BIOS chip should be saved too, in case you run into another R9 390 that's been flashed with wrong BIOS or BIOS with very aggressive settings and not letting the card boot anymore.

But really, this video card is quite dead and not repairable - and this is coming from me (I rarely give up on anything, so if I think this is done, it probably really is, just ask anyone else around here or see some of the crap I have repaired.)

[sigmeyer]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

Thanks momaka , yeah , i was thinking that too , sadly... I Am actually experimenting with some homemade BGA removal ways...I got plenty of ps3 boards with damaged GPU's and CPU's... So i am trying to manufacture a homemade BGA Rework station. I will post a Thread on this website as to what im planning to do .Its gonna take a lot of experimenting i am aware.. I was successful at desoldering a couple of GPU's from Dead ps3 boards but the boards warped I am guessing from un even distribution of the heat... I need to find a better suitable PRE heater... I am also aware of TEmperature graphs and how the temperature should act when desoldering BGA chips... So thats what im gonna do in the near future... Anyways sorry for derailing of topic .. i seem to do that alot hehe.. Take care all , thanks for inputs

[momaka]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

I Am actually experimenting with some homemade BGA removal ways... So i am trying to manufacture a homemade BGA Rework station. I will post a Thread on this website as to what im planning to do .Its gonna take a lot of experimenting i am aware.. I was successful at desoldering a couple of GPU's from Dead ps3 boards but the boards warped I am guessing from un even distribution of the heat... I need to find a better suitable PRE heater... I am also aware of TEmperature graphs and how the temperature should act when desoldering BGA chips... So thats what im gonna do in the near future...

When I used to work a console repair shop some years ago, we always had problems with PS3 boards warping - even when we used custom metal support stencils.

Then we found that the key to reduce/minimize warping is to use point-source heaters of equal power, so that the top and bottom under the chip you're working on get heated evenly. A large pre-heater wasn't needed too much (and often when set to too high was the main reason why boards warped... so a powerful bottom pre-heater is not the best for PS3 boards IME.)

Nowadays, I don't do BGA work (professionally) anymore. However, when I do need to remove a large BGA or SMD component from a board (like the PS3 Ethernet chip I sent you), then I put the board on a grille on my gas stove. With a heat gun or hot air (depending on how large of a chip I'm working on), I try to match the heat output of that to the bottom (gas burner) heater so that the board heats evenly from both sides. And so far, I haven't warped or burned any boards with this method.

I got plenty of ps3 boards with damaged GPU's and CPU's...

Not to sound like a broken record (I believe I've probably said this in another thread of yours), but if any of those boards have NEC/Tokin caps, chances are those are bad, and not the GPU or CPU.

When I worked in that same console repair shop, our success rate with PS3 reflows and reballs was very very low compared to Xbox 360s. From approx. every 10 PS3s with YLOD, only 1 to 2 would get revived by reflow/reball. It had nothing to do with the station we were using (we went through a few and got down the BGA rework very well.) We just didn't know about the defective NEC/Tokin caps at the time, and looking back now, I'm pretty sure that was the issue with many of them. If you're not convinced, I believe there are a few older Toshiba laptop threads here on BCN that also used those NEC/Tokin caps, and pretty much everyone that's had an issue with these laptops and replaced these caps got a working laptop again.

So just giving you a heads up before you start removing GPUs and CPUs from those PS3 that it may be worthwhile to first check the NEC/Tokin caps.

Anyways sorry for derailing of topic .. i seem to do that alot hehe.. Take care all , thanks for inputs

Hey, no worries.
It is your own thread after all, so feel free to take it in any direction you like.
Of course, a separate thread about your rework station would be better and likely get more interest/help/suggestions from people.

[sigmeyer]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

I know about tokin caps , I have recapped and fixed like 6 ps3 with bad caps. But i have just recently got mailed a few boards with clearly scratched CPU's , from failed deliding process. So thats why im looking in the direction of reballing... Im hoping i can do it on PS3 boards... I was thinking of getting an even heater ,get it to around 120C , and hit it with the Heat gun. I will post a thread about this when i get to it. Hopefully i can do a few boards without them warping , then i will try to do a proper reball. Today i saw a video of some indian guy who seemed to have some knowledge of how to test a GPU chip with multimeter , but it was in indian so i dont know what he was saying. Maybe if you have some knowledge about testing BGA chips on ps3 might come in handy in future. The GPU i desoldered seemed to test fine , according to that indian dude. Anyways this is a big step for me , so far i have only done mosfet /cap replacements , the bga replacement is a whole another realm for me... This hobby of mine is slowly getting out of hand

[momaka]

Re: R9 390 8GB gygabite Shorted 12v PSU connecter fire and flames

Today i saw a video of some indian guy who seemed to have some knowledge of how to test a GPU chip with multimeter , but it was in indian so i dont know what he was saying. Maybe if you have some knowledge about testing BGA chips on ps3 might come in handy in future. The GPU i desoldered seemed to test fine , according to that indian dude.

There are a ton of videos online how to "test" a CPU/GPU with multimeter, and all I have to say about those is they are mostly a waste of time. The only thing you can use a multimeter for on a CPU or GPU is to make sure its power pins aren't shorted - i.e. verify that V_core is NOT showing 0 Ohms (or close) resistance to ground. But for that, you need a good multimeter with good probes. Cheap multimeters tend to show a few Ohms even with leads touching each other, so they aren't that great for testing chip resistance. Non-shorted chips should show at least 1 Ohm of resistance. Any less is suspicious.

But either way, this "test" is very crude, because different CPUs and GPUs tend to have different resistances. For example, it is normal for an ATI Radeon HD4800 series GPU to measure 1.5 to 4 Ohms (typical), depending on room temperature. Likewise, many CPUs will usually show resistance in the 10's of Ohms. However, the resistance is NOT an indication itself if the chip is actually working properly or not. Rather, the resistance measurement is just a very basic test to make sure the GPU or CPU didn't become short-circuited from the reflowing/re-heating/re-balling process, which can happen sometimes, particularly on sensitive chips. Anyone who claims on these videos that the resistance measurement can be used to tell if a chip is good or bad probably doesn't really understand how the electronics inside work at all. Unfortunately, there is a lot of crap on YouTube and other video websites that passes as "technical information". Not to say there aren't any good technical videos... but you just have to look a little more to find them. Obviously Louis Rossmann and EEVLOG are some of the more popular guys that actually know their stuff, among others.