Sapphire HD 7870 possible short

[Agent24]

Re: Sapphire HD 7870 possible short

I think you have the VRM sections mixed up. The GPU takes a lot more power than the RAM, so it will have more power phases.

A short between Drain and GND on the lower FET could be a shorted FET (quite likely) BUT - it could also be anything in parallel with one of those FETs, like an output filter capacitor (I'd be wary of those SMD output capacitors), some IC, or even the GPU itself (that is unlikely though).

Isolating the FETs for testing by removing the coils should work, but you will need a good hot air station to do that as they will suck up a lot of heat... might pay to inject current instead.

If you want to try injecting current, you don't need a huge amount. Just enough to heat the bad part a little. You can put isopropyl alcohol on them all and see on which one(s) it evaporates the fastest. Get a current limited supply set to something less than the full rail voltage, and limit it to 100-200mA or something to start with. You can turn up the current a bit if you need to but if you just whack something capable of several amps straight on there, you may burn up the faulty part so bad it cooks the board and then you'll be screwed.

If you have a very sensitive meter, you can also measure mV drop across each FET\Capacitor and see where the drop is lowest - that should be your shorted part. Or do the same thing with Ohms and power off. That will require a very good meter though.

In any case - do not exceed whatever voltage would be on the rail during normal operation, otherwise you risk blowing the GPU\RAM or whatever else is powered from that rail.

Also, in the interests of avoiding ESD damage, it may pay to keep the card off carpet....
And please upload your photos to the forum instead of elsewhere.

[momaka]

Re: Sapphire HD 7870 possible short

I tried the card in several computers and none of them boot up (one of the PCs had a loudspeaker and beeped a GPU error, so it's certain that the card is faulty).

When you say that none of the computers boot up, do you mean that the fans in those PCs won't spin at all with the video card installed?
- If YES, that IS indeed indicative of a short-circuit.
- If NO, then you are looking at a different problem here.

The three ribbed heatsinks on the left side of the card should be the inductors for GPU Vcc

Those are very likely for the RAM Vdd.

The 8 sapphire logo ribbed heatsinks on the right side of the card should be the inductors for RAM Vcc

Nope.
As Agent24 also noted in his post above, the GPU core takes a lot more power than the RAM. So the VRM with those 8 heatsinks (actually inductors?) to the right of the GPU is for GPU Vcc (or GPU V_core, as some may call it.) The RAM only uses about 10-20 Watts usually.

There are three FETs on the right side of the card for each RAM inductor, the bigger FETs are IRF6725, while the smaller ones are IRF6721
...
From this we can conclude that the smaller 6721 fets are the upper ones, and 6725 are the lower ones.

That seems correct.

Since FETs usually fail in such a manner that they short the drain and source pins, i concluded from the short circuit between 6725 drain and GND that this is indeed a blown lower mosfet

Not quite.

That the lower MOSFETs read shorted is normal (at least in 90% of the cases I've seen). This is because GPU Vcc (and sometimes RAM Vdd) naturally has a very low resistance to ground on just about any video card that was made in the last 15 years. The Source pins of those "lower" MOSFETs are always connected to ground and the Drain pins are connected to GPU Vcc.

Therefore, what you are likely seeing is the low resistance of the GPU chip and not a short-circuited lower MOSFET. Of course, this does not guarantee that you don't have a shorted lower MOSFET. Unfortunately, shorted lower MOSFETs can be tricky sometimes. In many cases, usually the PSU short-circuit protection will trip and so the fans in the computer will spin only for a split second and then stop. If this happens, then YES, you definitely do have a short-circuit. BUT, it is also possible for the VRM controller to detect the short-circuit and simply not run the VRM. This will allow the computer to turn ON without tripping the PSU short-circuit protection. If that is the case, then you need to verify if GPU Vcc and RAM Vdd voltages are present. Take your multimeter, set it to DC voltage, and measure the voltage between one of the GPU Vcc coils and ground. Do the same with RAM Vdd as well.

So with that said, please confirm the symptoms that your video card has - i.e. whether the computer fans can or cannot spin with the video card installed in the computer. And if they can spin, check the voltages on GPU Vcc and RAM Vdd. Report back what you get.

Other than that, I don't see any short-circuits between PCI-E 12V rail on external connector and ground according to your measurements, so at least we can conclude that the GPU Vcc upper MOSFETs are not shorted. As for the RAM Vdd upper MOSFETs, I can't say yet. You need to find if those have their drains connected to the external PCI-E 12V connector or the 12V rail from the PCI-E slot (on that short section of the PCI-E slot of the video card, various pins will be connected to 12V rail, 5V rail, 3.3V rail, and ground).

Furthermore, if my diagnosis is correct, I was wondering if there is any easier way of finding out which fet is faulty apart from desoldering every one of them until the short is gone. I was thinking of maybe rather removing the 8 sapphire chokes so as to remove the connections between fets and the remaining parts of the board

If removing the chokes/inductors/coils is easier for you, then YES, you can do that. But personally, I always find it easier to remove the MOSFETs. I do this with two strong soldering irons (50 Watts or more recommeded and big chisel tips).

... and then sending some current through the fets to actually blow the faulty one to the point where its identifiable by naked eye (heat, smoke..?).

Won't work in many cases (almost all I've encountered).
Semiconductors (diodes, MOSFETs, BJTs) can often become short-circuited to such a point, that it would take a silly amount of current to blow then up. When I say 'silly amount', I'm talking on the order of 10-20 Amps or more. I've had shorted MOSFETs repeatedly trip the OCP (over-current protection) on the 12V rail on one of my PSUs. And that PSU is capable of 16 Amps on the 12V rail. So unless you use a welder with DC output (which would be silly ... unless you want to celebrate New Year's Eve fireworks early), I don't think you would be able to push enough current to blow a shorted FET.

But i must admit, it seems a very risky strategy, since i dont know what voltage to use, how the controllers for the FETs (which are on the backside of the card ) work and if there are any other connections to the board that bypass the chokes etc..

Good thinking

Yes, even with the inductors removed, there are still other paths the current can go through and potentially cause damage (pardon the pun here ). In particular, some MOSFETs will also develop a short-circuit between Gate and Drain or Gate and Source (and yet sometimes, all three as well). A high-enough voltage going back through the Gate can damage the MOSFET driver IC (if there is one) and/or even the VRM controller IC - and those are NOT fun to replace .

So no, DO NOT try to burn out the MOSFETs with a current source. "Risky strategy" is a mild, but still accurate way to put it .

If you want to try injecting current, you don't need a huge amount. Just enough to heat the bad part a little. You can put isopropyl alcohol on them all and see on which one(s) it evaporates the fastest.

Oh yes, you DO
Even a tiny shorted ceramic cap can take upwards of 1 Amp to heat up when it blows.

Also, I wouldn't suggest anyone to put a flammable substance on live electronic equipment. IMO, that's a fire display show waiting to happen.

Plus, how would you even make sure the alcohol layer is even everywhere? From personal observation, I know that whenever I clean anything with alcohol, it never dries all at once - some parts take loner and some shorted. So I don't think this would be a very precise way to find which part heats up.

[Agent24]

Re: Sapphire HD 7870 possible short

Yes, even with the inductors removed, there are still other paths the current can go through and potentially cause damage (pardon the pun here ). In particular, some MOSFETs will also develop a short-circuit between Gate and Drain or Gate and Source (and yet sometimes, all three as well). A high-enough voltage going back through the Gate can damage the MOSFET driver IC (if there is one) and/or even the VRM controller IC - and those are NOT fun to replace

Good point about shorts to Gate. I didn't think of that in this situation, for some reason!

Alcohol and external power does seem to work. In one video by Louis Rossmann, he just poured it on and the shorted part stuck out like a sore thumb because it dried off very fast while everything else stayed wet. Nothing went up in flames. The shorted IC in question was drawing about 400mA.

[Anzezaf]

Re: Sapphire HD 7870 possible short

Thanks to everyone for the input and correcting me on the power supplies

@momaka

The fans in the computers spin normally.
I now measured the GPU and RAM voltages but i get nothing. 0 volts on RAM inductors, and weirdly consistent 0.03V on GPU inductors... The GPU itself does get hot though, pretty fast at that. How the hell is this possible, 0V yet the GPU is getting power..
It's a slim chance... but could the 0 volts be due to the fact that the card doesn't boot properly in the few seconds i had to work with until the GPU heated up (the cooler was removed to allow for measuring)?

Moving on, the RAM "upper" fets are weirdly connected - one (the upper one on the picture) leads to the external connector and the lower one on the picture is connected to the PCIe slot ..??

And for the final thought, I did some investigating and am now positive (no pun inteded ) that a short circuit occured. 3.3 or 5V was applied to the ground .

EDIT: I now also attached a pic of the backside of the card. The assemblies to the left of the choke housings are (I presume) the fet drivers, but more importantly, each of the capacitors to the right (circled in red) "report" a short circuit, with the same damned resistance as measured before - 0.6 ohm .

[Agent24]

Re: Sapphire HD 7870 possible short

0.03v is not 0v. And if the GPU gets hot it's definitely getting power. You may find if you measured the GPU core rail with an oscilloscope, it is actually sending out pulses of DC - quite possibly from the VRM controller trying to drive a short - which means it's constantly starting up, detecting overcurrent, shutting down, and then trying to start again. In this situation the GPU won't get a lot of power, but might get enough to heat up.

That's one theory anyway. Maybe momaka has more of an idea about those symptoms.

[Anzezaf]

Re: Sapphire HD 7870 possible short

Hey guys, a quick bump.

I really really dont want to be ungrateful for all the advice already given, but I am kinda in a hurry to get this sorted ASAP, if even possible.

[Anzezaf]

Re: Sapphire HD 7870 possible short

I also wanted to check the buck/phase controller (the large IC on the far right side of the card, near the fan headers) for any shorts, but I cannot make any sense of the pinout in the datasheet - can't find any grounds. The IC is CHL8228G and here's the datasheet. https://cdn.badcaps-static.com/pdfs/...17738a7af4.pdf

[Agent24]

Re: Sapphire HD 7870 possible short

The ground is connected to the pad under the chip.

QFNs (and various other similar SMT packages) have a large pad underneath (see attached image) and this pad is often used for ground or power connections. Sometimes it's NC (Not connected) though. It's also used for physical support - soldering to the PCB - and often as a thermal interface between the IC and PCB, and the PCB is used as the heatsink.

[momaka]

Re: Sapphire HD 7870 possible short

Sorry for the late reply. Been sick lately (again).

The fans in the computers spin normally.

So your video card doesn't have a short-circuit then.

I now measured the GPU and RAM voltages but i get nothing. 0 volts on RAM inductors, and weirdly consistent 0.03V on GPU inductors... The GPU itself does get hot though, pretty fast at that. How the hell is this possible, 0V yet the GPU is getting power..
It's a slim chance... but could the 0 volts be due to the fact that the card doesn't boot properly in the few seconds i had to work with until the GPU heated up (the cooler was removed to allow for measuring)?

Okay, the text there in bold is worrysome.

Do NOT ever run a processor or video card without a heatsink, for any reason. I've seen people online recommend this as a method to see if the CPU/GPU turns ON, but I think it is something very silly to try. The GPU core may be detecting that there is no heatsink (due to overheating) and thus shutting down the VRM controllers. Therefore, you CANNOT take voltage measurements with the heatsink off.

So put the heatsink back on the GPU (with proper thermal compound) and measure the voltages again. The GPU and RAM voltages should be present on the back of the card as well. If you're not sure how to find them, then you can always solder some wires to the output inductors, and then put the GPU heatsink back on, and take voltage measurements from the wires.

Moving on, the RAM "upper" fets are weirdly connected - one (the upper one on the picture) leads to the external connector and the lower one on the picture is connected to the PCIe slot ..??

Yes, that is possible.
Looking closely at your pictures, there may actually be two VRMs in what we called the "RAM VRM". It is hard to tell, though. For troubleshooting, we really need good focused pictures of both sides of the video card, taken straight from above.

And for the final thought, I did some investigating and am now positive (no pun inteded ) that a short circuit occured. 3.3 or 5V was applied to the ground .

How do you know or what symptoms did your PC/PSU/video card show?

EDIT: I now also attached a pic of the backside of the card. The assemblies to the left of the choke housings are (I presume) the fet drivers, but more importantly, each of the capacitors to the right (circled in red) "report" a short circuit, with the same damned resistance as measured before - 0.6 ohm .

That's because they are in parallel with the GPU core, so they will show the same resistance. Nothing to worry about here either.

And if the GPU gets hot it's definitely getting power.

Exactly.

Which is why I recommend that you take the voltage measurements again, but this time with the heatsink on.

I also wanted to check the buck/phase controller (the large IC on the far right side of the card, near the fan headers) for any shorts, but I cannot make any sense of the pinout in the datasheet - can't find any grounds.

I'll have a look at the datasheet.
But first, check those voltage measurements again.

Also, check resistances between 12V rail and 3.3V rail to ground on the PCI-E slot. Here is a pinout of a PCI-E slot:
http://cfile3.uf.tistory.com/image/1...4A793AA1229E75

On that picture (top side), "+12V" is the 12V rail and "VCC3" is the 3.3V rail. So check between those and ground with the video card removed from the computer. I doubt you will find a short-circuit, but just in case.

[Anzezaf]

Re: Sapphire HD 7870 possible short

Okay, the text there in bold is worrysome.

Do NOT ever run a processor or video card without a heatsink, for any reason. I've seen people online recommend this as a method to see if the CPU/GPU turns ON, but I think it is something very silly to try. The GPU core may be detecting that there is no heatsink (due to overheating) and thus shutting down the VRM controllers. Therefore, you CANNOT take voltage measurements with the heatsink off.

Didn't know that running it without the heatsink is such a serious matter . I only ran it for a few seconds every time though, so this shouldnt be a problem.

So put the heatsink back on the GPU (with proper thermal compound) and measure the voltages again. The GPU and RAM voltages should be present on the back of the card as well. If you're not sure how to find them, then you can always solder some wires to the output inductors, and then put the GPU heatsink back on, and take voltage measurements from the wires.

Right, I reassembled everything except the backplate which i left out. I guessed the backplate doesnt matter that much , since it doesn't provide any direct functionality, is grounded as a whole and it would only get in the way of measuring the voltage on the chokes. The voltages measured are the same as before . 0 volts on RAM chokes and 0.02 / 0.03V on the GPU chokes. Which could make sense because of the low resistance of the rail itself, the voltage "drops". Maybe? No?
I should also say that when I was measuring for shorts on the PCIe slot connector I found none, but i arrived at an interesting fact nonetheless. When I intentionally measured a short (both voltmeter probes on the same pin just for the heck of it), it measured 0.6 ohm (I have a cheap-o multimeter so it can't be very precise anyway). Could this actually mean the drains of the GPU Vcc MOSFET(S) are really shorted to the GND, because of the same 0.6ohm resistance measured a few posts back?



Looking closely at your pictures, there may actually be two VRMs in what we called the "RAM VRM". It is hard to tell, though. For troubleshooting, we really need good focused pictures of both sides of the video card, taken straight from above.

I tried my best with the pictures, even used a tripod stand , but it is hard to take good ones nonetheless. Are you interested in any specific area? It would be easier to take pics closer up. Ask away for any IC model numbers, if you need any. It's the least I can do.

How do you know or what symptoms did your PC/PSU/video card show?

Here's the full story. The card we're diagnosing now was in my PC all along. I recently bought another card (R9 290)to also be used for mining next to the existing one. I installed the new 290 , and connected the PCIe external connectors. Here is where I made the error - I managed to connect a normal physically compatible connector which wasnt ELECTRICALLY compatible. It was wired differently, with 12V, GND, 5V and 3.3V pins instead of the normal one which only has 12v and GND. I don't remember exactly, but as I said, either 5V or 3.3V was connected to the cards'ground. I of course didn't notice the error when connecting, since who the hell checks the voltages before connecting anyway right? If its fits, it should work, just as with any other connectors/ cables in PCs.
Well then, after connecting the cables i ran the computer, and it went like this: A small (electrical) pop, and fans stopped turning instantly after starting (it all happened in lets say half a second). I then quickly switched the PSU off, and after smelling something burned, disconnected the PCIe external cables on the new 290 card, but kept it inside the mobo slot. Then I tried running the computer again, but this time there was a second electrical pop, which was a bit louder (nothing extraordinary though) and this is the time I think also my original card got damaged. (Looking back at this, the first short must have damaged the motherboard). I then disconnected everything, removed both cards, and began testing them in other PCs, only to found out they're broken. The bad news is also that the motherboard was faulty, and I had to replace it. At least the PSU is still okay (measured all outputs). So that's it, I actually have two cards with (as far as i briefly compared) similar symptoms.



Also, check resistances between 12V rail and 3.3V rail to ground on the PCI-E slot. Here is a pinout of a PCI-E slot:

Yeah, again as I said, no shorts there.

.

[Agent24]

Re: Sapphire HD 7870 possible short

because of the low resistance of the rail itself, the voltage "drops". Maybe? No?

Yes, in fact. Look up Ohm's law. The lower the resistance something has, the less voltage across it. A dead short across a rail basically pulls the entire thing to ground so the voltage will be very low, but the current (if not limited) very high.

I managed to connect a normal physically compatible connector which wasnt ELECTRICALLY compatible. It was wired differently, with 12V, GND, 5V and 3.3V pins instead of the normal one which only has 12v and GND. I don't remember exactly, but as I said, either 5V or 3.3V was connected to the cards'ground.

That is probably more likely to cause damage to the GPU core and other ICs on the card, than just a MOSFET shorting out.

A small (electrical) pop, and fans stopped turning instantly after starting (it all happened in lets say half a second). I then quickly switched the PSU off, and after smelling something burned, disconnected the PCIe external cables on the new 290 card, but kept it inside the mobo slot. Then I tried running the computer again, but this time there was a second electrical pop, which was a bit louder (nothing extraordinary though) and this is the time I think also my original card got damaged. (Looking back at this, the first short must have damaged the motherboard). I then disconnected everything, removed both cards, and began testing them in other PCs, only to found out they're broken. The bad news is also that the motherboard was faulty, and I had to replace it.

If you managed to kill the motherboard and another card as well... it's not looking good, IMO.

[keeney123]

Re: Sapphire HD 7870 possible short

Hey guys, a quick bump.

I really really dont want to be ungrateful for all the advice already given, but I am kinda in a hurry to get this sorted ASAP, if even possible.

With electrical devices you do not want to be in a hurry. That is when one causes errors.

[Anzezaf]

Re: Sapphire HD 7870 possible short

Small bump. Having nothing better to do, I "mapped" some IC's on the board. Figured it might help.

1. 78M05G 500mA voltage regulator
2. PL25LD010 SPI flash
3. SL1601 clock generator
4. CHiL CHL8228G multi phase buck controller
5. Bank of CHL8510 gate drivers
6. PL25LD010 SPI flash again
7. MC74VHCT125A quad bus buffer
8. pair of CHL8510 gate drivers
9. CHiL CHL8212 Multi-Phase Buck Controller

[momaka]

Re: Sapphire HD 7870 possible short

Didn't know that running it without the heatsink is such a serious matter . I only ran it for a few seconds every time though, so this shouldnt be a problem.

Nope, even 1 second is too much. Sure, the newer GPUs and CPUs have protections to turn OFF / throttle down when they go over temperature, but they can still get damaged. So you should never run any PC component without a heatsink if it was designed to have one originally.

I should also say that when I was measuring for shorts on the PCIe slot connector I found none, but i arrived at an interesting fact nonetheless. When I intentionally measured a short (both voltmeter probes on the same pin just for the heck of it), it measured 0.6 ohm (I have a cheap-o multimeter so it can't be very precise anyway). Could this actually mean the drains of the GPU Vcc MOSFET(S) are really shorted to the GND, because of the same 0.6ohm resistance measured a few posts back?

Yes, it could be that (shorted lower MOSFETs).
Or it could be that the GPU or RAM actually shorted from the reverse polarity.
Or it could be that the Tantalum caps shorted (and/or possibly some of the polymers as well). I know that Tantalum caps cannot withstand any reverse voltage.

Also, 0.6 Ohms with shorted leads on your multimeter is not that bad. I have some cheap meters that show 2-3 Ohms.

I tried my best with the pictures, even used a tripod stand

Your pictures are good enough now. Thank you for posting them.

Here's the full story. The card we're diagnosing now was in my PC all along. I recently bought another card (R9 290)to also be used for mining next to the existing one. I installed the new 290 , and connected the PCIe external connectors. Here is where I made the error - I managed to connect a normal physically compatible connector which wasnt ELECTRICALLY compatible. It was wired differently, with 12V, GND, 5V and 3.3V pins instead of the normal one which only has 12v and GND. I don't remember exactly, but as I said, either 5V or 3.3V was connected to the cards'ground. I of course didn't notice the error when connecting, since who the hell checks the voltages before connecting anyway right? If its fits, it should work, just as with any other connectors/ cables in PCs.
Well then, after connecting the cables i ran the computer, and it went like this: A small (electrical) pop, and fans stopped turning instantly after starting (it all happened in lets say half a second). I then quickly switched the PSU off, and after smelling something burned, disconnected the PCIe external cables on the new 290 card, but kept it inside the mobo slot. Then I tried running the computer again, but this time there was a second electrical pop, which was a bit louder (nothing extraordinary though) and this is the time I think also my original card got damaged. (Looking back at this, the first short must have damaged the motherboard). I then disconnected everything, removed both cards, and began testing them in other PCs, only to found out they're broken. The bad news is also that the motherboard was faulty, and I had to replace it. At least the PSU is still okay (measured all outputs). So that's it, I actually have two cards with (as far as i briefly compared) similar symptoms.

Well that paints a much better picture now . Should have stated this in the beginning.

Burned electronics smell and popping is never a good sign. It means there is definitely something burned on the card... possibly even multiple items. Therefore, I would say it is time to heat up the iron.

I see some Tantalum caps on your video card. Those are the black rectangles with "470 30up6" written on them. Tantalum caps will not tolerate any reverse voltage at all, no matter how brief. They always short-circuit when they go bad, so I suggest your start by removing those. Once removed, test them with your multimeter resistance setting - they should show open-circuit. If YES, we move to the next item: GPU VCC and RAM lower MOSFETs. Don't remove them one by one - remove all at once and test each out of circuit. With those MOSFETs removed, check the resistance between GPU Vcc and ground again. Also check between RAM Vdd and ground.

Let me know what results you get and we will continue from there.

Small bump. Having nothing better to do, I "mapped" some IC's on the board. Figured it might help.

1. 78M05G 500mA voltage regulator
2. PL25LD010 SPI flash
3. SL1601 clock generator
4. CHiL CHL8228G multi phase buck controller
5. Bank of CHL8510 gate drivers
6. PL25LD010 SPI flash again
7. MC74VHCT125A quad bus buffer
8. pair of CHL8510 gate drivers
9. CHiL CHL8212 Multi-Phase Buck Controller

Do any of these have missing pieces, bumps, cracks, or blisters on top of them? I don't see any on the pictures, but double-check me on that. If any of those chips do have those symptoms, they are more than likely bad as well.

[Anzezaf]

Re: Sapphire HD 7870 possible short

Right, thanks for taking the time again.

Therefore, I would say it is time to heat up the iron.

Now, I don't have an SMD hot air soldering station, and it's quite expensive to buy one, so I'm wondering if it's safe to desolder the FETs with a normal iron applied to the case? Since there's no plastic to be seen, I think this could actually be possible. But, on the other hand though, there might be plastic filler inside the case itself, which could present a problem. (By plastic i mean the bakelite or some kind of similar material that IC's are encased in)

I see some Tantalum caps on your video card. Those are the black rectangles with "470 30up6" written on them.

Do I take every 470 tantalum off the board, or just the 30up6 ones? They're on both sides of the board .

Do any of these have missing pieces, bumps, cracks, or blisters on top of them? I don't see any on the pictures, but double-check me on that. If any of those chips do have those symptoms, they are more than likely bad as well.

I can't see anything out of order with the ICs, sadly.

[Agent24]

Re: Sapphire HD 7870 possible short

A hot air tool is highly recommended for this kind of thing, in fact if it turns out you need to replace any VRM controller or such, it's practically necessary, in my opinion.

You can get pretty cheap ones on eBay from China. They aren't awesome, but they will work.

[momaka]

Re: Sapphire HD 7870 possible short

Now, I don't have an SMD hot air soldering station, and it's quite expensive to buy one, so I'm wondering if it's safe to desolder the FETs with a normal iron applied to the case? Since there's no plastic to be seen, I think this could actually be possible. But, on the other hand though, there might be plastic filler inside the case itself, which could present a problem.

To be honest, I haven't worked with newer FETs like that, so I can't say if you really need hot air. I see the MOSFETs have two pins on two side, so you may be able to remove them by flooding their pins with solder and using two high-power irons (60W or more). At least that's the method I use with TO-252 case MOSFETs, and it works okay even on very tough lead-free motherboard (I have a 70W soldering station and a 30W regular iron). You can also pre-heat the board a little bit with a hair dryer or heat gun (or in the oven too, if you are very careful) to help you remove parts more easily.

I *don't* suggest you try to remove the MOSFETs by heating the case. They most likely do have some plastic-like insulating material inside, so the temperature transfer from the case to the legs of the MOSFETs probably won't be very good.,

Do I take every 470 tantalum off the board, or just the 30up6 ones? They're on both sides of the board .

Probably better to remove every Tantulum - or at least every one that shows a short-circuit across its leads. That doesn't necessarily mean the Tantulum cap is shorted, but you can't know until you remove it from the board and check again.

I can't see anything out of order with the ICs, sadly.

Well, they could have survived, so let's leave them for now then.

[Anzezaf]

Re: Sapphire HD 7870 possible short

I tried removing the 470 capacitors with a 30W iron on one side and a 100W on the other and they wouldn't budge. The board is incredibly efficient at cooling so I'm waiting for a friend to lend me some kind of hot air fan, I might be able to something with that. I should get it this week and hopefully I will be able to report back during the weekend

[momaka]

Re: Sapphire HD 7870 possible short

Sounds good.
While waiting, you can give it another try. But this time make sure you apply flux to the capacitor leads before trying to desolder them (if you didn't already). This should help with the heat transfer from the iron to the board. As for flux, Rosin or Rosin Mildly Activated (RMA) should both work. Just *avoid* Rosin Activated (RA), as that may be too aggressive and corrode traces on the board.

[keeney123]

Re: Sapphire HD 7870 possible short

They also make solder that operates at a much lower temperature than even leaded solder. This would then mix with the solder on the board and prolong the length of time the solder would stay hot.