PC Graphics

[colesdav]

Hello

I purchased a second hand Powercolor R9 Fury X and a component of the back of the card has blown or was previously modified.
I cannot post pictures of the card at the moment. I cannot teardown the GPU to look to see if there is damage to the other side of the PCB yet.

I would have purchased a new card ... but was unable to. I invested in R9 Nanos / Fijii based cards as RX Vega release was so late.
RX Vega 64 cards are overpriced or > 2 slots high and cannot work with R9 Nanos in Crossfire.

The R9 FuryX card failed within 1 day of use. I had just started testing the card with default clocks, fan speed and power target.
I was running Windows 10 64bit, i7-4790K at default clocks, 32GB Ram, Primary GPU = Powercolor R9 FuryX, Secondary GPU = PowerColor R9 Nano. PSU = Corsair AX1200i. 
I was running Adrenalin 18.3.4. driver and Wattman Fan controls seemed to be working fine.
I tested DisplayPort 1 and 2. I ran 3D Mark  Firestrike and Timespy Demos (not benchmarks), and a few hours of some AAA games.
Card temps seemed fine @ ~60'C during testing.
Card crossfired O.K. with a PowerColor R9 Nano.
I powered the PC down. I connected the third DisplayPort Output to the monitor and I booted up the PC.
The Red Radeon Logo lit up on the R9 Fury X but the Red and Blue tachlights did not power up, the fan did not spin and I got no display output.

Could someone with an R9 Fury X to help me identify the component on the back of an R9 FuryX card and it's function if possible please?

Here is a picture of a teardown of a functional R9 Fury X (not the card I purchased) with the component of interest circled in red named B200.

I think the card is totally beyond repair. The damage at the red area is bad. Either someone had previously modded the card to temporarily replace a blown component or the component shown had failed due to overuse.

I have already tried opening an AMD Support Case (they cannot help), and contacted PowerColor Support (No response).

However if I could identify the component it might help me work out what happened/ went wrong with the card.

[elstaci]

IMO, the only way to find out is by locating the Electrical schematics for that card. Then you can find out either by the symbol or reference code what the component does and its purpose.

If it was just that one component and no other traces or components were destroyed inside the PCB (Some PCB boards have more then one layer of circuits) and you have the proper equipment to solder and desolder the component. Find yourself another bad Fury card and try to replace it from that card.

BTW, Some warranties are transferable to new owner if bought used and if still under Warranty. I know that Asus does.

EDIT: I didn't realize it was you colesdav​. I guess I am preaching to the preacher. You probably are aware of everything I said.

I did google R9 Fury electrical Schematics and found a website that seems to have it but you need to sign up and it seems like it isn't free even though its say it is free. Anyway, if interested here is the link: Nitro Wiring Diagram - free download wiring diagrams

[colesdav]

Hi,

Thanks.

I did open a support case with AMD. I believe that the R9 Fury X PCB are all reference designs so I thought they might know.  They told me to contact PowerColor.
PowerColor Support seems to be to return the card to the place where it was purchased. I have no idea where that was. The person who sold it to me is not responding to that question, and did not include the original receipt in the box.

I do not think the electrical schematics will be available easily.

I think the PCB is multilayer and it looks like it has been punched through. I would need to look at the card PCB with a microscope to determine if any tracks in that area look damaged. I do not know if some damage was present before I connected the card and ran it in my PC.
I should have removed the 4 screws on the card backplate and checked the PCB before I powered the card up, but the external condition of the card looked great, and even if I did that I have no way to see the other side of the GPU PCB.

It might be possible to measure across the component with a multimeter, as long as the card is disconnected from the power supply and tell what it's value is. For example if it is a resistor connected to ground or perhaps a diode. It might be possible to tell what the component is.

I have been working on a proper SMD Solder Rework station, but I really was not thinking of using it to work on this GPU....

RE: Find yourself another bad Fury card and try to replace it from that card.

I might be able to get hold of one, but unlikely. Most I see for sale seem claim to be working, have had one previous owner, and never been used for mining ....

RE: BTW, Some warranties are transferable to new owner if bought used and if still under Warranty. I know that Asus does.
I think it depends on where the card was sold. I contacted PowerColor, and Sapphire about this, they tell me Warranties are not transferable at all.

I  will think about your suggestion of looking for a burnt out for parts only card (like the one I bought ...).

Oh well, looks like this R9 Fury X is one for the bin or or maybe to be sent to Buildzoid ... he *might* be able to make use of it for something.

It's a real pity the card failed because I was really impressed by the GPU Clock Stability and the Temps compared to an R9 Nano. The AIO cooling solution looks good, and I assume it is ~ the same as on the RX Vega 64 Liquid Edition. I did not overclock the card but I estimate it might have been able to hit maybe a 5-7% overclock from 1050MHz  so maybe 1175MHz just based on the temps I was seeing versus an R9 Nano. 

Cheers.

[colesdav]

Hi, the schematics site you pointed me to seems to refer to a 2008  Dodge Nitro Radio Wiring Diagram, and my Norton Internet Security definitely does not like the link.

Thanks anyhow, it made me laugh.

[elstaci]

I don't blame you. I was wary of logging on also. I didn't get any warnings from my Mcafee Website safety app. nor from WOT which is why I posted the link. If I would have gotten a  Warning I would never post the link.

A couple of Users recently here in the Forum had Sapphire GPU cards. One was a R9 Fury and the other, I believe, was a R3900 card (My sapphire r390 just die after drivers update). Both cards suddenly died after either updating the AMD driver or playing a game. Both computers shut down by itself.  After restarting, no signal from the card. Both cards seemed to have burned out.

Found out Sapphire has a two year Warranty (From Amazon.com). But you must take it to the place of purchase for them to repair. The User in Europe ( Sudden death of Sapphire R9 Fury? ) took his R9 Fury to the place of purchase. They told him if the card can't be repaired that they will give him a prorated amount to replace the card. They offered to give the User 150 Euros if it couldn't be repaired. The User said he couldn't purchase another similar card like the Fury for 150 Euros.

The reason why I mentioned Asus, it was when I was helping a young User with his Asus Nvidia Card on Nvidia Forum, it seemed like his GPU card went bad. He couldn't afford another card (He sounded like a High School kid) and he purchased the GPU card used from Amazon Marketplace. So I went to Asus and Contacted Asus Support to find out about their Warranty. They have a three year Warranty regardless of who owns the card. As long as you give them the serial Number of the GPU card and it is under Warranty the User can RMA the card. it is covered. They might ask for a Receipt under certain circumstances. Gave the good news to the young User because the card was still under Warranty with less than three years since the original owner bought it. He said he was going to contact Asus. Never heard from him again. So I suspect his card got RMA from Asus Support.

I tried to google the Reference "B" on a PCB board. In Wikipedia it showed the majority References for electronic parts in a PCB board. This link is safe : Reference designator - Wikipedia .

The only close reference was BT for Battery. So I know that wasn't it. Sometimes "B" could stand for BATTERY according to the Wiki.

Good luck, your best bet is to cannibalize the part from another card.

[colesdav]

Hi.

I am certain the card is beyond repair and I do not really want to put it back in my PC.
There is a hole in the the PCB either due to previous modding and the component is completely missing from the board.
I smelt burning when I powered up my PC when the card failed.
That is the only reason I removed the GPU backplate.
There are what look like old burn marks on the PCB, some black material on the board ... and a dark brown mark on the inside of the GPU backplate above the missing component.

When I purchase a used GPU I treat it with great care in case if fails initial testing and has to be returned. I do not overclock, and I do not run anything crazy like Furmark stress testing.
 
I am more interested to determine what the component is and why it may have failed, since I connected the card properly to a fully working and tested PC it should not fail in < 1 full day of testing.

Here are the conditions on the PowerColor Warranty.

Warranty Information (Global) - PowerColor

The bold red uppercase text comes from the PowerColor Website ... not from me. I am not shouting at you.

"The warranty is valid for two (2) years starting from the original purchase date (verified through invoice) and ONLY VALID to the ORIGINAL owner. The warranty cannot be transferred if the product is to be resold or has switched ownership."

Looks like PowerColor might be tired of people having to purchase old GPU's because there are no new ones at reasonable price / good enough Performance/Power as well.

Bye.

[elstaci]

That to me is not a very good Warranty. The Manufacturer should be the one to RMA the card and not the place of Purchase. Suppose the Place of purchase closes or goes out of business. Also having a One year Warranty on AMD Vega cards and any Reference AMD Card that are so expensive is really a joke.

The Warranty also doesn't mention if the card that failed and is RMA through the place of Purchase whether they will replace it if it can't be repaired.

I know that Asus, as far as I know, doesn't care who owns the card as long as you provide a good serial number. Visiontek has a limited Lifetime Warranty. that was over five years ago when I Purchased a HD 7850 from them. Dell, according to their website, sells and uses Visiontek GPU cards.

It could be why their GPU cards are less expensive than normal because the Company doesn't have good support for the product.

[colesdav]

Hi,

The Sapphire Warranties here are similar story. I contacted Sapphire Support about it. The warranty only stays with the original owner provided they have a receipt, not that the PowerColor Card has anything to do with Sapphire. There are a number of Sapphire R9 Fury X that are available and also a Radeon Pro Duo that I was thinking of purchasing. However, I think it it is a high  risk today to buy any high end second hand GPU.  I think R9 FuryX Liquid Cooled Cards are pretty good for Etherium Mining. If you look at the Specification for those FuryX Fans they have a very long MTBF. Any indication of any problems with the fan on any 1-2 year old GPU indicates heavy 24/7 use or mining to me following this experience.

Bye.

[colesdav]

Just had a thought. I own a couple of PowerColor R9 Nanos. I will had a look for a component named B200 on the R9 Nano PCB and maybe it has the same function as B200 on the R9 Fury X. Long shot I know but maybe worth a try?

[elstaci]

Should be the same. Same company manufacturer and same Reference designator. Probably the electronic component is used in more than one type of card. Good idea.

[elstaci]

Found this IEEE Reference PDF. "B" could stand for Blower or BLower Motor. Here is the PDF link: http://electronica.ugr.es/~amroldan/pcb/2007/modulos/temas/ReferenceDesignators.pdf .

It is possible that component has to do with the GPU fan circuitry.

[colesdav]

Thanks.

Nice find.

The component is in an area of the PCB on the opposite side beneath the fan header on the card.
So it could indeed be B for Blower that sounds correct.

Or it could be B for "Blowup" ...

I believe the The R9 Fury X uses a 120mm  3000r.p.m Sythe Gentle Typhoon Fan, specs are here:

Gentle Typhoon 120 mm High RPM: CPU Kühler, Lüfter, Lüftersteuerung von Scythe

I think the exact model is D1225C12B7AP-29 (3,000 rpm)

The PWM fan controller wiring on  the R9 Fury X is:

Black = 0V

Red = +12V

Yellow = Sense (Fans have a Fan Speed circuit which sends back information to the Fan Controller) - measured at 3.75 volts with the fan running at full speed.  Blue = PWM control signal from the R9 Fury X Fan Circuit. Measured between 0.45 and 3.75 Volts depending on Manual Fan Speed control setting in AMD Adrenalin 18.3.4 Wattman.

I measured those values using a high impedance multimeter.

A simplified electrical model for a DC PWM fan should be a resistance (to represent the winding resistance of the coil on the rotor) in series with a "Back e.m.f."  voltage source which represents the voltage generated when the windings move and cut the magnetic flux in the permanent magnet field of the motor.

In other words the PWM fan itself, when connected to the Red and Black wires on the PWM Fan Controller,  looks like this in a simplified electrical model:

+12V(Red)  -------Rwinding------- + Vbackemf -  --- 0V (Black wire).

When the Fan is not spinning (at start up) there is no back e.m.f. voltage generated by the fan, and hence the fan start up current is the maximum worst case condition you should see.  In the above case the start up current would be Istartup = 12v/Rwinding.
So just for example if the winding resistance of the fan were 12.4 ohms then the starting current of the fan would be 0.97 amps stated in the specification for the fan.  The rated current for the fan is 0.22 amps which means the back e.m.f. of the fan when running at 3000 r.p.m. in the specification would be:
(12 - (0.22*12.4)) = (12- 2.73) =  9.27 Volts.

Note in my case the card worked initially but failed at subsequent boot up, in the condition when the fans are not spinning.

Also note, it is not generally a good idea to stop a spinning pwm fan with your hand,  as you will artificially run the fan in the startup current condition.

Getting back to the PWM fan controller on the card, I do not believe that the +12volt supply (Red) or the 0V DC supply lead outputs have any form of current limiter. However is it possible that the B200 component is a fuse?

Other than that I think the Yellow fan  speed sense wire would have to connect to a high input impedance circuit on the R9 Fury X card which should not draw any high current.

The Blue PWM fan control signal from the R9 Fury X  should not have to drive any high current either, it should only be sending a signal to a fan driver power stage on the fan circuit itself, which could be as simple as an N channel pull down power Mosfet (I__I) connected as shown.  

+12V(Red)  -------Rwinding------- + Vbackemf -  --- |___| --- 0V (Black wire).

                                                                                    |

                                                                                   Blue (PWM Control)

Note the pull down Nchannel Mosfet would have an "on resistance" > 0 ohms, so the Rwinding in the earlier calculation would have included the Mosfet on resistance.

Does the above sound reasonable to you?

Thanks.

[elstaci]

LoL, you sound like a Electric/Electronic Engineer. My training is from the Military in the 70's and 2.5 year of Electric/Electronic Tech school in the early 80's and over thirty years of troubleshooting Electro/Mechanical computerized equipment.

I understood basically what you are saying (not all ). If the B200 was a fuse, I would think the Reference name would include a "F" like BF200.  By the picture it has two contacts. It could be some sort of sensor like you mentioned. The card uses Three fans right? That is a lot of current to run three large fans.

The User of the other R9 Fury had the same problem. The computer shut down and when it restarted the card was bad. I am beginning to wonder if something in Windows 10 is causing , at least , Sapphires GPU cards to "blow up".

[colesdav]

Hi,

My PC did not fail to boot. The PC booted fine, which indicates no electrical shorting to me.
The PSU has overcurrent protection trip, if anthing had shorted the PSU supply then the red trip light would have come on and I would have had to reset it before next attempt to boot up.

The card simply died, no fan, no GPU tachs, no display output. The Red Radeon Logo came on.Thats it.

The R9 FuryX has a single fan. I do not think it runs at 3000 r.p.m. though.
The actual operating current on the R9 FuryX would possibly be lower, it depends on the PWM controller duty cycle that is used to control the fan speed.

Thanks again for all of your help.

I will power down this PC, remove the primary PowerColor R9 Nano and see if I can work out if there is a B200 on the PCB and see if I can work out what it is.

Bye.

[colesdav]

Hi,

There is a B200 on the R9 Nanos but it is underneath the cooling shrouds and not accessible.
I would have to remove the cooler and I do not want to remove them at the moment.
Also the Fan on the R9 Nano is completely different to the Fan on the R9 Fury X so there is no guarantee that the component value would be the same.

Some good news - I just received a full refund for the cost of the card so I can now decide if I want to risk purchasing another second hand R9 Fury X
or if I should just keep all of the new and second hand R9 Nano's I own and move on to purchase a new GPU.

Prices of new cards have now dropped again here.

I have seen new two slot high PowerColor RX Vega 56 Red Dragon announced however no 2 slot high Vega 64 Red Dragon and Performance/Price is still an issue.

I see rumor that AMD is working on new GPU for release this year but I cannot really wait for new GPU any longer: 
AMD RX 600 Series GPU Project "Zen" Detailed - Radeon on Steroids to Amp Clock Speeds & Efficiency
I will wait to see what Nvidia announce for next gen GPU's since they seem to be moving forward with support for OpenCL 2.0 recently, which is the main reason why I stayed on AMD cards last year. I need to do more investigation on that, and I also need to check if Windows 8.1 64bit support will still be there on new Nvidia cards.

I have been told I can keep this broken R9 Fury X card but I do not think it is repairable and I have insufficient information to attempt to repair it. I do not want to risk fitting it to my PC's.

If anyone with an R9 Fury X has removed the card from their PC and could take the time to open the backplate on their card by removing four screws, and measure B200 and tell me what that component is I would still be grateful to know.

Also if anyone has any more details about the Fan Controller on the R9 Fury X, or any success with changing the R9 Fury X fan to a different model I would be grateful to get information on that.

I think I have done all of the investigation I can on this for now.

Thanks again for your help.

Bye.

[colesdav]

Hi,

Here is a breakdown of the PowerColor Vega 56 Red Dragon PCB for anyone who may be interested.

PCB Breakdown: RX VEGA Nano / sapphire Pulse / Powercolor Red Dragon / XFX air - YouTube

Bye.

[colesdav]

Hi,

I just purchased and completely QA tested a second hand Sapphire R9 Fury X. The card was tested on ~ 80 of the latest AAA games and benchmarks on Windows 10 64bit with Adrenalin 18.3.4 driver. The results are pretty impressive, especially when paired with my best PowerColor R9 Nano.  

Since I keep the card I removed the backplate by pressing down the Sapphire Sticker which covers the backplate screws over the screw locations, using a tiny craft knife to carefully cut circular hole 'flaps' on the Sapphire Backplate cover, removing the screws and taking the backplate off.

I used a standard multimeter on 200 ohm setting to measure across the B200 component, I also tried the Diode meter across the component.

Here are the results.

B200 is not a diode.

B200 is not a capacitor as it does not register as an open circuit.
B200 measures at 0.3 -0.4 on my multimeter.
Shorting  the multimeter leads I see 0.2 - 0.3 depending on how hard I push the leads together.

It looks like B200 is a fuse on the good Sapphire R9 Fury X card.

This is good news and bad news in terms of looking to repair the failed PowerColor R9 Fury X.
Fuses can be replaced in a normal circuit and the circuit should still work.
This makes it pretty difficult to figure out what value the fuse value should be though ...
I guess I will try to calculate that myself, assuming the fuse is in line with the power supply to the fan.

Wish me luck.

[colesdav · Accepted solution]

After some effort I managed to repair the Blown R9 Fury X Fan Controller.

The GPU is now running fine.
I disassembled the R9 Fury X completely and cleaned the PCB with Isoproyl Alcohol.

I did not touch or replace the thermal paste on the cooler.
I replaced the blown component with a '0' Ohm surface mount resistor rated to blow if the fan windings short out.
I soldered the connection as follows using a battery powered soldeing iron with a very thin tip and some thin strands of AWG wire.
I used a magnifying glass from a "hobby hands" set to solder the smt resistor to the wire strands  and then connect the wires onto the Lowest  Pad and the red fan lead pin:

Solder Connection:
Lowest "B200" pad ---> '0' ohm surface mount resistor--> Blown "B200" Pad (Red Fan lead pin (Fan "VDD"))

The replacemt surface mount resistor runs horizontally from the lower "B200" pad across to the Red Fan Lead pin on the fan header.
It was to difficult to get a connection to the top Blown "B200" Pad.

I just did the soldering to the surface mount resistor by hand because I was just doing an initial test to see if repair was at all possible.

I will replace with a proper through hole to surface mount pcb to through hole adapter if this GPU lasts for more than a month after repair.

Here is a picture of the Blown Fury X in operation- it is running outside my PC case via PCIe2.0 mining adapter.

I connected a Corsair HD 120  RGB  "Pull Fan" driven from my PC Motherboard and controlled by AIsuite3 and Corsair iCUE to the other side of the radiator.
 

The fan speed control is working on the supplied fan that came with the R9 Fury X. It only runs at 1200 r.p.m and is only rated at 0.1 Amp  but I will look for a replacement Fury X Gentle Typhoon 3000 rpm fan in the future which has a rated current of 0.22 amps.  The rated current of the HD 120 RGB fan is 0.3A, so I will not drive it from the R9 Fury X just in case i blow the fan controller "fuse".

Here is the Repaired R9 Fury X running Compubench Ocean Surface Simulation.

In conclusion GPU "repaired". It will be interesting to see how long it lasts though.
I still need to replace the thermal paste, and run more extensive test on the GPU so I will leave it for now.

Thanks for everyone who helped with suggestions on how to fix it.

Bye.