Sapphire Radeon R9 390 Nitro Chip issue.

[shojo]

Hello guys.
I am trying to remove a chip from this card with no luck.
I tried preheating the board to 180 degrees(Celsius). Tried with 350 degrees (Celsius),400,450 and it doesn't want to come off. Tried different flux,tried different air speed,tried tinning the feet as good as I could and nothing.If you have any ideas please tell me what to do.
Any help will be appreciated.
Thanks in advance.

How the chip looks before heating :

https://postimg.cc/Z9htBQQP

After heating:

https://postimg.cc/cKcjKLwZ

Top side of the chip:

https://postimg.cc/XZn106hP

Bottom side of the chip :

https://postimg.cc/v4TD2482

[diif]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

Ideally you need a preheater when working on graphics cards, something like this https://www.aliexpress.com/item/1005002869922306.html
It's the large ground and power planes that are causing the issue not the pins you are tinning.
Use a decent hot air station , high temp, low air speed.

[foshland]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

And cover the capacitors with aluminun tape or they will pop because of heat

[Bozog]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

The output stage is presumably blown, so could have welded itself to the PCB, so you may have a hard time getting it off with hot air without destroying the pads underneath. You can grind them off with a dremel if done carefully.

[shojo]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

Ideally you need a preheater when working on graphics cards, something like this https://www.aliexpress.com/item/1005002869922306.html
It's the large ground and power planes that are causing the issue not the pins you are tinning.
Use a decent hot air station , high temp, low air speed.

My hot air station has a preheating plate.I wrote in the post I tried preheating the board with no success...😔

[shojo]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

The output stage is presumably blown, so could have welded itself to the PCB, so you may have a hard time getting it off with hot air without destroying the pads underneath. You can grind them off with a dremel if done carefully.

What...Never thought of that.Damn this is gonna be tricky if that's the case.Ok, gonna try grinding the chip.
If it has welded itself to the board does that mean that there is gonna be more damage than the chip itself or it's uncertain?

[foshland]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

Yes, it usually has blown the pad underneath the chip. Depending of your luck it will be better or worst, in worst case, when removed the chip, look for short in there and other rails and try to start it without fixing that channel, at last you will know if the card is more damaged or if it works

Have luck

[shojo]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

Yes, it usually has blown the pad underneath the chip. Depending of your luck it will be better or worst, in worst case, when removed the chip, look for short in there and other rails and try to start it without fixing that channel, at last you will know if the card is more damaged or if it works

Have luck

Thanks for the tips.
The card had no power before I changed the 8A fuse near the pci-e.After that I could turn on the PC,but had no display.Then measured with the multimeter and it turned out it was the chip,maybe something else too, but started with that chip.
Damn I hate complicated shit 😁.I would never thought off the welding scenario.
Thank you.

[DynaxSC]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

I would try once again before grinding. Set preheater higher gradually even to 240 degrees C, and hot air gun to even 450 degrees C (take off all plastics from the card). You need also some time, up to 5-7 minutes to heat, you have near the mosfets also the coils which take away a lot of heat making it diffucult to heat the mosfets fast. Also the board has internal heat dissipation layers (usually a ground layer), so its necessary to heat the board accordingly, as it distributes the heat quite good. Factory tin melts at 254 C, so it is really a lot of heat and some time necessary to make the tin liquid. You have to have a suitable flux, resistant to such temparatures, most of them you can buy are only for leaded tin, melting at some 224 C. I use Kester TSF-6502-E, this is a professional industry flux, but it may be difficult to buy, it is also very active, suitable for difficult pads, eg corroded or so. I buy it in Poland at RENEX, a company selling machines for electronics industry, they have a page you can buy it, but I'm not sure if they send outside Poland.
Then before soldering again, clean devices and board pads from all the non-leaded tin, and use leaded tin, this makes it easier to solder again.
.
BTW, an easy trick to check what is bad is taking off all the coils, then you can measure where the short is, on VRM side or on GPU/memory side, more often it's the GPU in my experience, as VRM's have over current protection. If VRM is shorted, then usually the transistors and often board are burnt.

[shojo]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

I would try once again before grinding. Set preheater higher gradually even to 240 degrees C, and hot air gun to even 450 degrees C (take off all plastics from the card). You need also some time, up to 5-7 minutes to heat, you have near the mosfets also the coils which take away a lot of heat making it diffucult to heat the mosfets fast. Also the board has internal heat dissipation layers (usually a ground layer), so its necessary to heat the board accordingly, as it distributes the heat quite good. Factory tin melts at 254 C, so it is really a lot of heat and some time necessary to make the tin liquid. You have to have a suitable flux, resistant to such temparatures, most of them you can buy are only for leaded tin, melting at some 224 C. I use Kester TSF-6502-E, this is a professional industry flux, but it may be difficult to buy, it is also very active, suitable for difficult pads, eg corroded or so. I buy it in Poland at RENEX, a company selling machines for electronics industry, they have a page you can buy it, but I'm not sure if they send outside Poland.
Then before soldering again, clean devices and board pads from all the non-leaded tin, and use leaded tin, this makes it easier to solder again.
.
BTW, an easy trick to check what is bad is taking off all the coils, then you can measure where the short is, on VRM side or on GPU/memory side, more often it's the GPU in my experience, as VRM's have over current protection. If VRM is shorted, then usually the transistors and often board are burnt.

The flux I use is Amtech nc-559-asm.
It was recommended by a friend for this kind of work.
I am going to give it a try with the temps you gave and if It doesn't work, I am gonna try and grind it off the board.

If I manage to successfully change the chip and the card doesn't work, I will try the coil removal and will do further measurements.
Thanks a lot for the reply and the tips.

[DynaxSC]

Re: Sapphire Radeon R9 390 Nitro Chip issue.

The GPU can have a very low resistance, I don't know this GPU, but eg. GTX1070 has average 0,13 ohm, when measured with a special low resistance meter (having 4 cables). Your GPU probably will have higher resistance, but if you wan't to measure the GPU resistance with a normal meter, then you have to deduct from the result the resistance of the cables, as it is usually somewhere in the range of the GPU resistance or even higher. So measure the resistance on the GPU, then short the measuring cables, and measure the cable resistance, and deduct the first from the second result. if it is below 0,1 ohm, then probably PGU is dead. If somewher of 0,3 ohms then it's OK.

The memory resistance is much higher, eg ca 60-120 ohms on GTX1070 (8GB), dependent on memories, Samsung memories have the lower resistances of ca 60 ohms. Microns and Hynix's will have ca 100-120. But your card can be somewhat different.