XFX R9 290 Double Dissipation VRM Situation

[suzerain_00]

What's poppin forum. I've been meaning to build my little sister a new system for a while now, and this past black friday I ordered all the parts. I did however decide to look to the used market for an R9 290, as the performance figures I've seen are fairly respectable and the cards can be had for cheap. So I travelled out an hour from where I live, picked up an XFX R9 290 Double Dissipation for $130 (CAD) and took it home. The guy I picked it up from had mentioned that he had flashed an R9 290X bios on to the card as this model has dual bios.

 

On first glance, the card seemed to be just what I was looking for. I ran Fortnite, Forza and GTA 5 as well as did some encoding on the GPU and everything seemed pretty swell, although the card does exhibit some coil whine. On further inspection, when I ran furmark the card would run at ≈ 950 MHz, but would do these weird semi rapid dips down to ≈ 300 MHz. I thought that that was a little strange, but the card ran games fine, so I thought it was probably okay.

 

The card was a little dusty when I got it, so I decided to disassemble it to clean it up. Looking at the VRMs/VRM thermal pads, I noticed something a little concerning. There was one mosfet that seemed to be off kilter, and it's position on the thermal pad was discolored as compared to the rest of the contact with the other mosfets. To that point, looking at the cards thermals in GPU-Z, "VRM Temperature 2" under Furmark stress test reads up to 130°C.

 

Do you think that one of the mosfets is dead? Is it possible that the coil whine I'm hearing is due to this potentially dead mosfet?

 

This is my first time buying a graphics card used, and the first time I've fully disassembled one, so I would like to hear from someone who is a little more experienced to get their perspective, if at all possible. I've provided to images as well to give a little more context, but if you require more detailed ones, I can try to provide them.

 

Thank you so much everyone!

[Raudi_]

That honestly looks like a broken joint

[suzerain_00]

1 minute ago, KingCry said:

That honestly looks like a broken joint

Broken joint? Could you elaborate what that is for me, I'm unfamiliar with the term.

[Raudi_]

1 minute ago, ExodusR said:

Broken joint? Could you elaborate what that is for me, I'm unfamiliar with the term.

So there is a solder joint on that looks like it was broken during shipping or it isn't broken and has been just shorting out.

[Fasauceome]

Have you bought some thermal pads to replace the old ones and see if the temps get better? 

[suzerain_00]

2 hours ago, KingCry said:

So there is a solder joint on that looks like it was broken during shipping or it isn't broken and has been just shorting out.

I see, I'd imagine that there isn't anything I could do about it unless I hate an acute skill set in soldering.

[suzerain_00]

2 hours ago, fasauceome said:

Have you bought some thermal pads to replace the old ones and see if the temps get better? 

I haven't, although the thermal pads are still in stellar condition. The other VRM readout in GPU-Z has only ever read up to 70°C

[suzerain_00]

UPDATE

 

I took a look at Buildzoid's R9 290x breakdown, and it appears that my card has essentially this exact same PCB. 

He mentions that the inductors of this particular PCB are known for exhibiting a noticeable amount of coil whine, which would explain the coil whine I'm experiencing with my card, as the two share the same inductors.

 

The "weird semi rapid dips down to ≈ 300 MHz" are most likely being caused by a VRM overheating problem when running Furmark. In Buildzoid's video, he mentions that the 290x's VRMs are fairly powerful, but will also throttle if they encounter a VRM overheating situation. I believe that explains why under Furmark stress, my card experiences those dips in core frequency. If you look at the XFX R9 290 DD's VRM "cooling solution" you'll find that it's just a flat, thin metal plate that sits atop the mosfets and memory. While this has proven satisfactory in gaming and encoding scenarios in my testing, it is inadequate for high stress, high load scenarios like Furmark (or perhaps mining if that's more your speed.). For my use case, I believe the VRM cooling solution to be fine for the time being, however in the future I may take the plate off and replace it with some small heatsinks on the mosfets and memory. It's an incredibly odd design, and frankly I'd wager that removing the plate entirely and just letting the air from the card's fans hit breeze over the mosfets, may even benefit VRM temps. For anyone who is going to really tax a card with this PCB, I recommend you modify the VRM cooling solution, although given the card's age I suppose that isn't news.

 

After doing this bit of research, I've concluded that my card is probably fine.

 

As a side note for folks looking to buy a used R9 290, I'd encourage you to find a PCB breakdown of the card before you are looking at, because if you get a get a card with whiny inductors like I did, you're most likely going to be stuck with it. 

 

Otherwise, I think this thread can come to a close. Thank you everybody!