RX 470 Mining back from the dead - zombie edition

[katapaltes]

So here's the TL;DR question, but I'll put the details below: If I removed one of four inductors/chokes from my GPU's VRM, should the core now be running at its minimum of 300Mhz instead of 1206Mhz? I was hoping I could run using three of the four power phases if I underclocked/undervolted the GPU.

Details: I bought a Sapphire RX 470 8GB mining edition (DVI-only) on eBay months ago. It had been used for mining and had a mining BIOS on it. The white sticker on the fan shroud says it's a "11256-57." The card turned out to be DOA, with both fuses for the 8-pin PCIE power plug dead. I contacted the seller and he gave me a refund and told me to keep the card. Thus began my adventure.

The behavior with PCIE power connector plugged in: Computer wouldn't fire up at all (because of PSU's over-current protection?).
The behavior with PCIE power connector not plugged in: Computer started just fine but only had the integrated graphics.
Impedance testing with multimeter indicated a short.

I know little about electronics, but wanted to learn. At this point I've bought and used for this project: A cheap hot air rework station, .5mm Chipquik solder, flux, solder wick, Littelfuses from China, and Fast Chip removal alloy. I've spent more on the card than it's worth, but it's all about life-long learning so it's worth it. Oh, and I bought another used Sapphire RX 470 4GB mining card (model 11256-58) from eBay so that I could compare voltage, impedance, etc. between the two cards. I replaced its mining BIOS with 193398.rom from TechPowerUp and it works great.

After watching hours and hours of videos (some of them repeatedly; some of them in Russian) and reading forums, I decided that at least one DrMOS was dead and I was pretty sure which one it was. It was then that my nightmare began. I could not for the life of me remove the damn FDMF3035 DrMOS ("integrated MOSFET plus driver power stage") with my $50 hot air station and limited skills. In fact, I accidentally dislodged a couple tiny passive components with the hot air while the DrMOS would never budge. I even tried using a heated frying pan and removed the card's bracket to help with the heat sink effect.

I read somewhere that I could disable one phase entirely and that the card should run without it. I saw that removing the through-hole capacitors wouldn't help in troubleshooting and that I'd have to remove the SMT choke. Again, hot air didn't do the trick (for me) so I bought the $20 tube of Fast Chip and HOLY CRAP that stuff works. In the photo below, you can see the dislodged choke, two tiny passive components hanging on for dear life, and the real culprit, a scratched-up FDMF3035 DrMOS that I couldn't remove.

With the choke removed, all impedance testing was suddenly perfect.  Woo-hoo!  I installed the card in my PC and got fan spin, and then powered it up and Windows launched. I then flashed with the closest BIOS I could find from TPU (209467.rom) as the installed mining BIOS displayed weird in GPU-Z. So to my question above about running on three out of four phases, is there any hack where I don't have to replace the DrMOS? Is the only path forward to replace that stubborn IC?  I have ordered five of them from China and they'll take a month to arrive, but I wouldn't mind leaving the card as-is if I could run it at say 75% performance.

[TheMG]

What is causing the difficulty in desoldering that chip is the fact that two of the three large pads underneath the chip are likely via-stitched to a large ground plane, which doubles as a heatsink.

For stuff like that, you pretty much need to preheat the entire PCB. This reduces the temperature delta between the PCB ground plane and the hot air tool. Lower temperature delta means the ground plane (heatsink) sucks less heat away from the component you're trying to desolder.

The same technique would also be necessary for soldering of the replacement component.

[katapaltes]

Thanks, TMG.  I had tried the "skillet method" in my attempt at desoldering.  Although I really have no idea how hot I got the PCB because my multimeter doesn't have a temperature sensor.  I would not have thought of heating it up for the *soldering* of the replacement component, so I really appreciate your bringing that up.  If I'm somehow able to remove that IC, I'll be sure to do that.

Too bad I can't get to those pads underneath, or I could use the Fast Chip alloy on them.  Now I'm wondering if there's a destructive way to safely remove that chip without harming the rest of the PCB.  If I could remove enough of the chip, I could get to those pads...

My other option is to just hand the card and the DrMOS to a friend.  She has access at her workplace to some skilled technicians that do soldering, failure analysis of ICs, etc., and has volunteered to ask them for a favor. 

[magic]

I would be cautious ordering stuff from China, it may be dodgy.

The VRM phases typically work in round-robin fashion and alternately inject pulses of current into the GPU and its capacitors. If one phase is missing, there will be no corresponding pulse and the voltage may sag more than usual during that time. IOW, there will be more ripple and more risk of instability. At least that's the theory, I have never tried.

Another issue is potential overheating of the remaining phases, but if you reduce power consumption by 25% that should not be a problem.

[katapaltes]

Magic, after I read your reply, Google popped an article on counterfeit chips into my feed.  Interesting stuff, and I will check the components to the best of my ability when I receive them.
http://defcon.no/electronics/fake-electronic-components-a-hazard-of-online-shopping/
https://hackaday.com/2018/01/15/is-that-part-a-counterfeit-here-are-a-few-pointers/

[magic]

In capitalist America Google searches you

Sometimes there are obvious signs, sometimes there are not, sometimes it doesn't work, sometimes it's a similar lower performance part, sometimes you are lucky. Every time you wait for delivery. Why not get it from Mouser or Digikey?

[Rasz]

Fix small smd components, resistor on the left looks like its shorting to nearby capacitor

Cheapest thermocouple capable multimeters start at ~$30

or you can buy dedicated preheater components, temp controller+SSR, for <$20
https://www.ebay.com/itm/Digital-PID-Temperature-Controller-100-240VAC-40A-SSR-K-Thermocouple-Sensor/163413482647
https://www.ebay.com/itm/REX-C100-LCD-PID-Temperature-Controller-with-Solid-State-Relay-K-Thermocouple-/164279459365

300W Infrared Ceramic Heating Plates are ~$5 on alliexpress, or raid goodwill for an old infrared space heater.

or $3 dedicated thermocouple sensor breakout board for arduino. Most likely counterfeit MAX part, meaning it might be couple degrees less accurate, no big deal.
https://www.ebay.com/itm/Competitive-New-MAX31855-Module-K-Type-Thermocouple-Sensor-for-Arduino/192077472612
https://www.ebay.com/itm/MAX31855-Temperature-Development-Module-K-Type-Thermocouple-Sensor-Arduino-US/353159694218

You can also preheat from the bottom with another hotair, or even paint remover, just dont blast it directly at the board. 100-150C on the surface of PCB is ok.
Use metal foil/kapton to shield surrounding elements when using hotair.
You should probably start by practicing on some scrap laptop/gpu boards, anything modern from recycle bin would do.

[katapaltes]

Why not get it from Mouser or Digikey?

From Mouser or Digikey, the part is $2 and the shipping is $8.  From China, the parts were five for $5 with cheap shipping.  Of course if the parts are fake it's false economy - especially when you factor in my (or more so someone else's) time.  The wait doesn't bother me and I need to stretch out this project while stuck at home!  : )  I also didn't know how common counterfeit components are when I ordered them.  The 10A Littelfuses I got from China seem to work, but then again fuses might be the easiest type of component to counterfeit.

I've read a little about how power phases in a VRM work and I've seen explanation about the alternating pulses.  I wouldn't try and run the card at full power with a missing phase.  I was going to run it at reduced power and speed in Crossfire configuration with the other RX 470 I bought for this project.

[katapaltes]

Hi Rasz,
I removed three small SMD components that I had accidentally dislodged with hot air, including the resistor that was shorting to the capacitor.  With the power phase disabled I had assumed that the potential short did not matter.  That was a bad assumption on my part, but removing them didn't help and the GPU clock still won't run above 300Mhz.

Why did I remove them instead of fix them?  I did try, but I couldn't move them back in place.  I simply don't have sufficient skill with such small components.  I taped each of the tiny components to pieces of paper and wrote down their location on the PCB in case I ever want to try and replace them.  With the phase still disabled I assume they aren't needed, and I don't know that I'll ever really have the skills to replace that stubborn DrMOS anyway.  If I ever get truly brave (or truly bored) I will pull the DrMOS and components out of the drawer and try to re/place them.  I'll purchase kapton tape or use foil for any hot air work as you advise.

In the meantime, I'll try and figure out how to run this card at 50 or 75% power using utilities like OverdriveNTool.  I'm not having much luck at that; Power State 0 seems to be hard-stuck at 300Mhz due to the missing phase, which isn't really a surprise. 

[katapaltes]

So just to update/close the thread, I used a BIOS editor (PolarisBiosEditor-1.7.2) to set Power State 0 to 926Mhz, roughly 3/4 of the rated speed of the card.  (Software can be used to change seven of the eight power states, but not Power State 0.)  It's not as fast as 1206Mhz, but the card is now usable and completely stable with the one power phase missing.   With the BIOS modded, I had to run atikmdag-patcher against the video card driver or the driver wouldn't recognize the card.  I also slightly undervolted the GPU core to make things a little easier on the remaining power phases.

Thank you guys for your help and consideration.  I've learned a lot from this crash course.