Resoldering missing capacitors GTX 670, Need your help.

[PartyWire]

I need someone's help that has a gigabyte gtx 760 windforce 3x 2gb by hand, and a multimeter to measure the capacitance of 3 capacitors.

So recently I had gotten a gtx 760 for free, it was broken to some extent. I know it's not a GPU problem because it shows the boot logo of my Asus MoBo and any other MoBo I tried it on. But it gets stuck there. I inspected the PCB and found that 2 capacitors above the PCIe slot have been torn off (pic 1). I game my best to resolder them with my soldering station but no luck, still gets stuck. Most likely because of 2 problems. Near the GPU die on the back (pic 2), there had been another capacitor broken off. Which I didn't see prior. And that the capacitances were ways below what it was probably expecting. The others measure around 250nF - 310nF, whilst this measured a 4nF.

 

So if anyone has any schematic I could look at to fix this gtx 760 by gigabyte or if you have a multimeter and the same card, maybe you can do a few measurements of these three. I would be very thankful.

 

P.S. I know it isn't really worth fixing this card, but I'm just doing it so I can practice my soldering and maybe to my surprise I can get it to work once again.

[Levent]

Why are you even bothering with trying to solder caps? dont get me wrong I am no professional but capacitors are used for smoothing voltage output and are not exactly mission critical. Your issue sounds like GPU core itself.

[PartyWire]

3 minutes ago, BlueScope819 said:

I don't think that's the problem. If the graphics card works at all then there isn't a problem with the board. It looks like from the pictures that the resistors are for the 3.3 volt power coming from the PCIe slot which means that if they were truly broken then the card would not work at all. Find a review of the card with PCB pictures and I bet that those are missing in that as well.

No, I'm pretty sure that they were broken off before, because of 2 reasons. There was visible that it had something there but it was missing, solder was not rounded down like anywhere else, it was a bit spiked like something was removed by force. Second, is that I already looked at pictures of the back of this card, and the few i managed to find had those 3 capacitors.

 

Also, the area near the die, it had its capacitor still in one of its solder pads but hanging on with its life. After I noticed it, I took it off but on further inspection of the capacitor, it had its one end broken off and it was still stuck in the solder where it sat.

[PartyWire]

6 minutes ago, Levent said:

Why are you even bothering with trying to solder caps? dont get me wrong I am no professional but capacitors are used for smoothing voltage output and are not exactly mission critical. Your issue sounds like GPU core itself.

I've had such a problem before. I had a hd 7950 that had its aux power area ridden of a few capacitors. And due to that it didn't work. I let that be fixed by someone else, after they replaced the cap it worked again. Looked janky but it worked.

[PartyWire]

1 minute ago, BlueScope819 said:

If the card can boot up at all then that probably means there isn't a problem with any of the power rails, which is what you would be trying to fix. It seems to me like there is a problem with the GPU BIOS.

I forgot to add that before I started resoldering. Them I placed it into my pc, I saw a screen I had never seen before. Even though I had connect all the required power cables to the card.

[PartyWire]

Even if it is something with the Bios of the card. I'd still just in case love to resolder those caps. So that I won't have any future issues or any hiccups when using this card

[Levent]

1 minute ago, PartyWire said:

Even if it is something with the Bios of the card. I'd still just in case love to resolder those caps. So that I won't have any future issues or any hiccups when using this card

Try to get the card to work first. I broken off caps of my previous cards before and they all ran perfectly fine. Picture you posted suggest there is something wrong with VRMs or power rails.

[PartyWire]

4 minutes ago, Levent said:

Try to get the card to work first. I broken off caps of my previous cards before and they all ran perfectly fine. Picture you posted suggest there is something wrong with VRMs or power rails.

In about 9 or 10 hours I can try to flash the card to a bios from Techpowerup. For that time being maybe someone can tell me the capacitance of them or has a schematic.

[PartyWire]

21 minutes ago, BlueScope819 said:

Well, you should have said something. It seems like in fact this is a problem with your power rails.

Exactly why I want to try to fix the capacitors

[mariushm]

C826 and C827 have to be there, otherwise the pci-e line won't work.

You can see there's 2 capacitors on each pci-e lane, so if there's no capacitors there the card will have transmission errors on that pci-e lane.

You can desolder one other capacitor from another pci-e lane, measure it and then put it back.... all those capacitors lined up vertically and grouped in two will have same specs ... probably 10nF or 100nF or something like that, but measure them.

 

As for that other capacitor, no idea what value it has, and it could be important or needed for a voltage regulation circuit to work. a 330-470nF would be a relatively safe value to experiment.

Chances are those capacitors are not your only problem though, if you get that no power message with the extra power cable plugged in. I'd look into seeing if fuses are dead near the 6/8 pin connector, or see if diodes are open

[PartyWire]

12 minutes ago, mariushm said:

C826 and C827 have to be there, otherwise the pci-e line won't work.

You can see there's 2 capacitors on each pci-e lane, so if there's no capacitors there the card will have transmission errors on that pci-e lane.

You can desolder one other capacitor from another pci-e lane, measure it and then put it back.... all those capacitors lined up vertically and grouped in two will have same specs ... probably 10nF or 100nF or something like that, but measure them.

 

As for that other capacitor, no idea what value it has, and it could be important or needed for a voltage regulation circuit to work. a 330-470nF would be a relatively safe value to experiment.

Chances are those capacitors are not your only problem though, if you get that no power message with the extra power cable plugged in. I'd look into seeing if fuses are dead near the 6/8 pin connector, or see if diodes are open

I measured the PCIe lane ones, pairs usually measure around 250 - 310nF. They varied on groups for some reason.

 

Is there any difference if I just put a higher capacity capacitor in place of the old one?

 

And once I get home I'll take a closer inspection of the PCB

 

[PartyWire]

1 hour ago, mariushm said:

C826 and C827 have to be there, otherwise the pci-e line won't work.

You can see there's 2 capacitors on each pci-e lane, so if there's no capacitors there the card will have transmission errors on that pci-e lane.

You can desolder one other capacitor from another pci-e lane, measure it and then put it back.... all those capacitors lined up vertically and grouped in two will have same specs ... probably 10nF or 100nF or something like that, but measure them.

 

As for that other capacitor, no idea what value it has, and it could be important or needed for a voltage regulation circuit to work. a 330-470nF would be a relatively safe value to experiment.

Chances are those capacitors are not your only problem though, if you get that no power message with the extra power cable plugged in. I'd look into seeing if fuses are dead near the 6/8 pin connector, or see if diodes are open

So I got my brother to take pictures of the card and I saw a few things, most were just empty soldering pads but there was this one orange tube, I'm guessing it's either a diode or fuse, probably more like a fuse. And I haven't really dealt with SMD circuits that much, but it seems broken to me. What do you think?

[mariushm]

7 hours ago, PartyWire said:

I measured the PCIe lane ones, pairs usually measure around 250 - 310nF. They varied on groups for some reason.

 

Is there any difference if I just put a higher capacity capacitor in place of the old one?

Those capacitors MUST be matched with the other capacitors that are on the traces going to the pci-e connector ( a mismatch can cause data transmission errors) 

You can't measure the capacitors reliably while they're on the circuit board, you should desolder the capacitor and measure it outside. 

No idea how well it would work if capacitance is incorrect AT THAT LOCATION. 

 

6 hours ago, PartyWire said:

So I got my brother to take pictures of the card and I saw a few things, most were just empty soldering pads but there was this one orange tube, I'm guessing it's either a diode or fuse, probably more like a fuse. And I haven't really dealt with SMD circuits that much, but it seems broken to me. What do you think?

That's a fast diode, nothing broken or bad about it, that's how the package looks. You can test it with a multimeter in diode mode, should show some voltage with the meter leads one way, and nothing with the leads the other way.

It's not a fuse. 

 

Fuses will have the letter F by the component and they're usually white or ceramic and sometimes with some kind of green lines on them.