Question I bought a broken r9 390x for cheap has the 6 pin connector grounded

[Richez]

I bought a broken asus strix r9 390 for 30 bucks on ebay. The card itself is in great physical shape but doesnt boot when in pc. When there is no pcie power plugged in it boots blind, when the 8 pin pcie connector is plugged in it does the same, But if i plug just the 6-pin pcie connector in the psu trips. The fans in the pc just turn on for about half a second. Im fairly good at soldering on pcb's and i have a rework station if it is something i can replace. I could use a hand identifying the issue with the card

 

[Richez]

Now at this point, you should be supplying low voltage (~.8v) to that left pad with safe current (~1A) from your lab psu and checking what gets heated via your thermal imaging device.

From here you are blind.
I would check those capacitors, one might be shorted out.
Then (if not shorted) the ceramic capacitors (6) next to them off and check.
Then (if not shorted) the MOSFET (3) pairs further left.
Also, the 7805 (left side on the photo) gets shot a lot on those cards, check it for shorts with a diode ring.

https://imgur.com/a/TcmF702

Both Capacitors to the far right read as a short the 7805 is clean and the ceramics are reading as a short

 

[vov4ik_il]

Possible issue? Mosfet looks like either corrosion or burnt top mosfet of the ones you specified looks damaged

https://imgur.com/a/mdoNX6l

View: https://imgur.com/gallery/mdoNX6l

Most likely. Graphics cards often get overclocked and run hot. People watch the core temperatures but are not aware of or disregard the fact that VRM is not monitored and is not designed to run higher (than design) currents for long periods of time. But the sole fact that there are heat signs is not sufficient. It might have happened after a capacitor failure too. Let's get there first with our troubleshooting. Are all the capacitors OK?

 

[vov4ik_il]

Both Capacitors to the far right read as a short the 7805 is clean and the ceramics are reading as a short

You have to take them off the board to test.

 

[Richez]

You have to take them off the board to test.

Which would i have to take off specify in a picture to which need to be taken off

 

[vov4ik_il]

Every single capacitor you check should be off the board.
Alternatively, you can take off the MOSFETs first and then check the choke pad for having a short instead, but MOSFETs are trickier to reseat, so it is easier to check capacitors first.
Now after you have found the beat-up ones, you will likely have to change the respective controller as when FETs get shot they likely to damage it and you have no scope to verify they are still operational. If you only change the MOSFETs and leave a bad controller, the new ones will burn too or something else will get cooked.

 

[Richez]

Every single capacitor you check should be off the board.
Alternatively, you can take off the MOSFETs first and then check the choke pad for having a short instead, but MOSFETs are trickier to reseat, so it is easier to check capacitors first.
Now after you have found the beat-up ones, you will likely have to change the respective controller as when FETs get shot they likely to damage it and you have no scope to verify they are still operational. If you only change the MOSFETs and leave a bad controller, the new ones will burn too or something else will get cooked.

im assuming the easier alternative is to get a thermal camera and a bench psu i can get a good thermal camera as i am an autmotive mechanic and enjoy working with electrical i have a thermal camera at the shop im currently laid off from i would still need a psu but that could be easy what would you recommend? i wouldnt want to have to remove each cap and have to replace them on the board
just to test them

 

[vov4ik_il]

im assuming the easier alternative is to get a thermal camera and a bench psu i can get a good thermal camera as i am an autmotive mechanic and enjoy working with electrical i have a thermal camera at the shop im currently laid off from i would still need a psu but that could be easy what would you recommend? i wouldnt want to have to remove each cap and have to replace them on the board
just to test them

If you are not comfortable with taking off those capacitors (in the circled areas) you still can take off the MOSFETs (clean the pads well) and then check if the short resistance is gone.

For PSU, you would want to look for 0~20V, 0~30A if you plan on doing it repeatedly. Good ones are expensive. I used to have an older model of this one at my old work, it is stable and precise.
I do not think you should get an expensive unit at this point, +/-5% accuracy will be OK... read reviews and check what is available in your area. 12V@30A is needed to power those cards solely off your PSU and monitor current while testing.

 

[Richez]

If you are not comfortable with taking off those capacitors (in the circled areas) you still can take off the MOSFETs (clean the pads well) and then check if the short resistance is gone.

For PSU, you would want to look for 0~20V, 0~30A if you plan on doing it repeatedly. Good ones are expensive. I used to have an older model of this one at my old work, it is stable and precise.
I do not think you should get an expensive unit at this point, +/-5% accuracy will be OK... read reviews and check what is available in your area. 12V@30A is needed to power those cards solely off your PSU and monitor current while testing.

sorry I still have basic knowledge on the circuit on this card which ones would the mosfets be

 

[vov4ik_il]

sorry I still have basic knowledge on the circuit on this card which ones would the mosfets be

Oh... Those are the MOSFET pairs with integrated driver. Datasheet. Spares.

https://imgur.com/cJi0ypd

P.S Hope you have BGA experience, as I said, they are trickier than SMD but still can be done without matrixes.

 

[Richez]

i have replaced something similar with success i will need to get more flux and tape though i am going to get the thermal camera tomorrow and will look for a psu to power the circuit so we can hopefully narrow it down in a much simpler way i dont want to just start taking things off of the card just yet if there is an easier way to identify the issue. I will continue with this and thank you for all of your help you are probably the best person to respond to this thread!

 

[Richez]

I bought some tape, solder, and solder paste as well to make my life easier. Also bought a new soldering station as the one i have now was kind of lacking in power and really only meant for basic soldering. I still have my hot air rework station as well. These should all be in about a week from now i should be able to continue then 👍

PS: just to make sure, the card wasnt recognized by the motherboard I'm assuming as it only displayed out of the motherboards on-board video. Would the card need to be supplied power from both the 6 pin and 8 pin for that to happen or is it tied into the occuring disgnostics

 

[vov4ik_il]

Would the card need to be supplied power from both the 6 pin and 8 pin for that to happen or is it tied into the occuring disgnostics

The card is missing at least 3 power phases on the core. There might be more issues but we have to get rid of the short-circuit first.
The card takes power from both connectors and the PCIe slot. You should get a riser for some tests. This way you will be able to run it without risking your motherboard to the point it starts all phases.

 

[Richez]

The card is missing at least 3 power phases on the core. There might be more issues but we have to get rid of the short-circuit first.
The card takes power from both connectors and the PCIe slot. You should get a riser for some tests. This way you will be able to run it without risking your motherboard to the point it starts all phases.

just found this as well. The ground pad is ripped off also. Only a sliver of ground left as well maybe just MAYBE salvagable also that cap isnt looking to happy

https://imgur.com/a/uFaNyC5

View: https://imgur.com/gallery/uFaNyC5

 

[vov4ik_il]

just found this as well. The ground pad is ripped off also. Only a sliver of ground left as well maybe just MAYBE salvagable also that cap isnt looking to happy

https://imgur.com/a/uFaNyC5

View: https://imgur.com/gallery/uFaNyC5

Well, this is bad.
It is likely an input fuse pad... and nearby cracked capacitor that might have caused this mess... also there is likely a missing driver nearby. This short circuit might be in between the board layers.
You can safely take off all those MOSFETs, the cracked capacitor and check for shorts. Also look for similar logic board to find out what else is missing. It does look like an unsuccessful repair or a donor card

 

[Richez]

Yea this might be out of my expertise for repair sadly i can continue and try to make it work but the chances are low it would be up to you to decide if i continue with the repairs or instead keep this as a donor

Well, this is bad.
It is likely an input fuse pad... and nearby cracked capacitor that might have caused this mess... also there is likely a missing driver nearby. This short circuit might be in between the board layers.
You can safely take off all those MOSFETs, the cracked capacitor and check for shorts. Also look for similar logic board to find out what else is missing. It does look like an unsuccessful repair or a donor card

I ended up getting bored and couldnt sleep until i scanned over the card entirely one more time and sadly i found this mess i still wont be able to do anything until the items i ordered arrive in the mail

 

[Richez]

Ok i wiped away the broken cap with my soldering iron the pads are still there for the cap but i would need very steady hands and much better equipment to replace that. I was using my phones camera to get a better view of what i was doing making sure not to remove anything else while i was in there. Both the exposed pad and the pad for the broken cap are showing 97 ohms the ripped off pad looks to be a ground i wonder if it is safe to run a jumper wire from another ground? That seems like it would only turn into a nightmare....

 

[vov4ik_il]

Both the exposed pad and the pad for the broken cap are showing 97 ohms the ripped off pad looks to be a ground i wonder if it is safe to run a jumper wire from another ground? That seems like it would only turn into a nightmare....

My concerns are:

• missing elements (see photo)
• possible inter-layer short-circuit

With missing elements, we can only guess what is supposed to be there. It looks like it is missing a PWM controller in those pads.

1. Take a close photo of those 4 so the markings are readable (or just annotate them on the image).
Do you have another card like this from your other thread? Or is it the same one?

2. Take the 3 MOSFET ICs off at the bottom of the Core row. We will make some testing when they are away from the pads.

3. Pls post a good picture of the whole back. We need to find out which arrangement is used in other phases, that might give me a clue.

4. It should be a modified diagram of the typical usage cases from the manual.

https://imgur.com/a/05Jl41F

 

[Richez]

My concerns are:

• missing elements (see photo)
• possible inter-layer short-circuit

With missing elements, we can only guess what is supposed to be there. It looks like it is missing a PWM controller in those pads.

1. Take a close photo of those 4 so the markings are readable (or just annotate them on the image).
Do you have another card like this from your other thread? Or is it the same one?

2. Take the 3 MOSFET ICs off at the bottom of the Core row. We will make some testing when they are away from the pads.

3. Pls post a good picture of the whole back. We need to find out which arrangement is used in other phases, that might give me a clue.

4. It should be a modified diagram of the typical usage cases from the manual.

https://imgur.com/a/05Jl41F

there is no labeling of what i have to take a picture of lol im assuming you meant to mark the picture? also like i said i wont be able to do any BGA/SMD soldering until my Flux and other items come in and who knows this card might just be a loss considering the damage. But like i said in a previous post right now is the point where we should consider if this is a good time to stop or we just keep going. My skills are not all that amazing and I don't believe i could solder a cap on that small I'll see when stuff comes in

 

[vov4ik_il]

there is no labeling of what i have to take a picture of lol im assuming you meant to mark the picture? also like i said i wont be able to do any BGA/SMD soldering until my Flux and other items come in and who knows this card might just be a loss considering the damage. But like i said in a previous post right now is the point where we should consider if this is a good time to stop or we just keep going. My skills are not all that amazing and I don't believe i could solder a cap on that small I'll see when stuff comes in

In terms of time spent and money invested, this card is a loss to even start with
The r9 390x are fairly cheap, hot and power hungry. For me it is more of a fun to solve a puzzle... If you do it with some sort of economical profit in mind - you shouldn’t continue.

I was asking for a picture of the whole back of the PCB so I can see how the upper phases are wired. I maybe missed it in your previous posts.

I also asked to take a good photo of the components I marked in that previous post so I can see what is used there (highlighted by red squares, enlarge the pictures I posted to see)

 

[Richez]

I took a picture of that one vrm as well
which looked to be burnt but i think it actually just had flux on it i cleaned off the residue and you can see balls of solder on the ends. That would probably be the first mosfet i will remove as it looks like someone has already messed with it. Also the idea witb this card was to just see if i could get it to display a picture but chances of that are pretty slim now that i found the issue on the back of the card. Hell if i got it working properly it wouldve ended up as my main card my pc has a very high quality psu and also high wattage and would end up pairing well with my other hardware

https://imgur.com/a/YQuD8Yg

View: https://imgur.com/a/YQuD8Yg

 

[vov4ik_il]

Well,
It is not over yet, but it will cost another card to even buy the lab PSU you are going to need...

The good news is that I found a picture of the back of a similar card and there is nothing there on that empty 10-pin pad nearby. But it is low quality and hard to see what else is there.

The step-downs marked as AMCF are NB671LBGQ (Datasheet) and are not very expensive I think you will need to get at least one.
The IR3555s are not very expensive either, you will need 8 in the worst case. You might need the controller too - Digi+ ASP1300 (rebranded IR3567B).


The pad can be repaired, but if there is an inter-layer short circuit - this card is a junk.

So far I see no issues with memory phases so those might be OK but do not start. It might be due to something wrong with the up1641P (can't find it anywhere but here) memory power controller or just the power-up sequence does not get there. The memory ICs showing expected resistance.
The problem would be to find out the value of the shot capacitor on the back (will have to take off and measure one like that from another memory chip... now, you need a tool to measure ceramic capacitors). The ripped off part is probably another capacitor, but again, the only way to find out for sure is to check one of the other memory IC wirings or get a working card of the same type. Worst case, in addition to the parts in the back, you will have to get a memory IC upfront (HYNIX GDDR5 H5GQ2H24MFR-T2C ) and it takes a matrix to seat this guy properly.

The sad part is - only after getting all the power phases working, we will find out if the core chip is OK.
Will only know better when you take most of them off for which you need tools and skills.

If you are doing all this just to get a working card, you better off buying one. I can send you a working 1060 for say the same 150$.
If it is for joy - let's keep going when you get the tools, there are lots to do 🙃

 

[Richez]

Well,
It is not over yet, but it will cost another card to even buy the lab PSU you are going to need...

The good news is that I found a picture of the back of a similar card and there is nothing there on that empty 10-pin pad nearby. But it is low quality and hard to see what else is there.

The step-downs marked as AMCF are NB671LBGQ (Datasheet) and are not very expensive I think you will need to get at least one.
The IR3555s are not very expensive either, you will need 8 in the worst case. You might need the controller too - Digi+ ASP1300 (rebranded IR3567B).


The pad can be repaired, but if there is an inter-layer short circuit - this card is a junk.

So far I see no issues with memory phases so those might be OK but do not start. It might be due to something wrong with the up1641P (can't find it anywhere but here) memory power controller or just the power-up sequence does not get there. The memory ICs showing expected resistance.
The problem would be to find out the value of the shot capacitor on the back (will have to take off and measure one like that from another memory chip... now, you need a tool to measure ceramic capacitors). The ripped off part is probably another capacitor, but again, the only way to find out for sure is to check one of the other memory IC wirings or get a working card of the same type. Worst case, in addition to the parts in the back, you will have to get a memory IC upfront (HYNIX GDDR5 H5GQ2H24MFR-T2C ) and it takes a matrix to seat this guy properly.

The sad part is - only after getting all the power phases working, we will find out if the core chip is OK.
Will only know better when you take most of them off for which you need tools and skills.

If you are doing all this just to get a working card, you better off buying one. I can send you a working 1060 for say the same 150$.
If it is for joy - let's keep going when you get the tools, there are lots to do 🙃

I removed that broken ceramic capacitor it crumbled apart as soon as i took it off and i dont know where the remains of it ended up but a good friend of mine is using an R9 290 as his daily card so im going there tomorrow to compare the two side by side apparently the cards are identical but i will find out tomorrow and take good pictures of everything and the back of the card

 

[vov4ik_il]

I removed that broken ceramic capacitor it crumbled apart as soon as i took it off and i dont know where the remains of it ended up but a good friend of mine is using an R9 290 as his daily card so im going there tomorrow to compare the two side by side apparently the cards are identical but i will find out tomorrow and take good pictures of everything and the back of the card

Sounds good

 

[Richez]

Sounds good

if it is that mosfet then i should be able to remove it and clean the pads just need to be careful when installing it should heat the board slowly and then ramp the heat up to melting temp dont know the melting temp of the solder on the board. also this is really just a project to get some more understanding of how such circuits work and get some more experience with smd/bga soldering, im sure if its a inter-layer short then ill just keep the card around for parts because i may end up buying another that just has a simpler issue. All in all I just want to get more experience and i always want to understand how things work lol. Im always curious on how things work and always want to know more. Before i continue working on the card i need to wait for supplies to show up maybe next week because of the virus sadly been waiting a while already.

 

[vov4ik_il]

if it is that mosfet then i should be able to remove it and clean the pads just need to be careful when installing it should heat the board slowly and then ramp the heat up to melting temp dont know the melting temp of the solder on the board. also this is really just a project to get some more understanding of how such circuits work and get some more experience with smd/bga soldering

If that is a mosfet, you will have to get it ordered and replace it. If you do not want to wait, get a handful upfront

For temps, use anything below 350C, the lower the safer. I would preheat to a 100C.
For VRM working principle, you can check this nice article.