AMD Radeon RX 5700 XT Sapphire Nitro - troubleshooting and repair

[for.software3.45]

I am trying to save RX 5700 XT Sapphire Nitro. Computer was normally shut down in the evening and next day graphics card was not responding - no post.

I have done some troubleshooting but I've come to dead end.
What do we know so far:
- LEDs light up and cycle trough colours as normal
- fans light up but don't spin
- all fuses are OK
- no shorts
- no visual damage to PCB or components
- GPU is not heating up
- 12V (from 3 connectors) - present
- 5V (from 2 regulators) - present
- 3,3V (from PCIe connector) - present
- No 1,8V
- No 0,75V
- No 0,9V (Vcore and/or Vmem)

Now since there is no 1,8V of course there won't be anything going on until that is present. This is where I've hit the dead end. I can not identify any of voltage regulators and since I can not do that, I can not get pinouts. If I don't have pinout then I can not see what is going on with regulator etc.

Hopefully someone here will be able to help me. Either with identifying regulators and therefore pinouts or maybe just with other advises.

I have prepared some graphics for front and back of PCB so one could easily understand current situation.

Red arrows - 12V is present at these points
Green arrows - 5V is present at these points
Orange arrows - 3,3V is present at these points
Blue arrows - 0V is present at these points
Yellow squares with numbers - unknown voltage regulators that I need help with

On marked "unknown" voltage regulators these writings are present:
1.) 7162 0NVZ
2.) FFK 153
3.) FFK 153
4.) Nuvoton NUC028LAN - probably only used for LED and fan control
5.) AREK 930

If someone could help me with identifying these it would help me a lot.
Also I have no idea which one of these produces what voltage, so that would also help.

Any help will be greatly appreciated.

[foshland]

I think one of the 3 voltage regulators on the top are the cause, you should have voltage on that rails

[for.software3.45]

Originally Posted by foshland

I think one of the 3 voltage regulators on the top are the cause, you should have voltage on that rails

I don't think this is the case. These regulators on the top are for Vmem + VDCCI (I think) and will only come "ON" once 1,8V is present. These NCP302155 are being controlled by NCP81022 (blue square).
Vdd on NCP81022 is present, however there is no "chip enable" signal and that should be present for controller to operate. I presume "chip enable" is not there because something is wrong earlier in the power sequence.

[dicky96]

1.8V is usually an LDO, not a buck regulator - if that helps

It is usually near the PCIe connector where the 3.3V come in. I suggest you trace from PCIe 3.3V and see where it goes to in that area

What are the two 8 pin SOIC chips near the PCIe connector? Is one of them a uPxxxx? If so that is probably the 1.8V regulator

[for.software3.45]

Originally Posted by dicky96

1.8V is usually an LDO, not a buck regulator - if that helps

It is usually near the PCIe connector where the 3.3V come in. I suggest you trace from PCIe 3.3V and see where it goes to in that area

What are the two 8 pin SOIC chips near the PCIe connector?

Now I do believe that I had pointed out pretty much all of the chips & possible regulators that are on PCB aside that two 8-pin ones. Those are bios chips (identical). Card has dual bios ergo two chips.
I did not see any RX 5000 series repair video, but on RX 6000 series the 1,8V seems to be produced with buck regulator. However I will try to find where the 3,3V rail goes.

dicky96, believe it or not I would not be here if it wasn't for your 2h+ video on graphics card diagnose (watched it till the end). Thank you.

[for.software3.45]

Not a whole lot of progress but better something than none at all.
Since I've got nothing to work with I started comparing working and non-working "unknown regulators" with the same designation (numbers 2 & 3 from the first post, both FFK 153). The working one is for 5V rail on the top of the card (#3) and the non-working (#2) one was for unknown voltage source near PCIe connector. Since only the output voltage is suppose to be different I started comparing them pin-to-pin.

Here is what I found.
On the non working one there was one pin without signal (apart the output one of course). That pin was connected to 10kOhm resistor which had 3.3V on one side (the one not connected to FFK 153) and 0V on side connected to the pin. This 0V side was also connected to another line which I traced back only to find out it comes from the back of the PCB - from another "unknown regulator" AREK930 (#5 in first post). This must be "power good" from AREK to enable the FFK 153. Because of this I believe that AREK930 is the one that produces 1,8V on this card.

If I had pinout for this "AREK 930" regulator things would be alot easier. The unit itself has 16-pins. 12 in DIL configuration and 2×2 on the sides (photo attached in first post).
Hopefully someone will recognize this element.

[piernov]

https://www.badcaps.net/forum/showpo...85&postcount=3

ARE: MPS NB685AGQ

[for.software3.45]

Originally Posted by piernov

ARE: MPS NB685AGQ

Thank you. You did save me a lot of trouble and at the same time completely negate what I came up with in my previous post.

I did some measurements and results can be seen in attachment.
I have colored good results in green and red where different results should be present. Where I was not sure if result is ok or not, I've left them black.
It can be clearly seen that pin EN2 on NB685A is not getting any signal, so chip is not enabling/starting.

Line that I thought was "chip enable" for FFK 153 is actually going the other way - it enables NB685A (AREK). So apparently I was wrong and this FFK 153 should produce 1,8V.
Again I have no idea what the pinout for FFK 153 is and hopefully someone recognizes this one too.

However I went ahead any made some additional measurements on "non-working FFK 153". Input voltage (12V) is present, supply voltage (3,3V) is present, however there is no "power good" (probably) that would enable next regulator in sequence and also no output voltage.

After that I rechecked resistance measurements across the board and found out that there is actually short to ground from output of this "FFK 153". I didn't notice this one before. However all fuses are still ok.

Here are all the common current resistance measurements of my card:
- SoC: around 2 Ohm
- Vcore: around 0,1 Ohm
- Memory: 2×55 Ohm
- VDDCI: 10 Ohm
- Rail with NB685A (on the back of the card): 1,5 Ohm
- Rail with "FFK 153" (near PCIe connector): 0 Ohm
- all 12V rails: kOhms

I have compared this to some data from working RX 5600 XT (similar card) I found on the web:
- SoC: 2,4 Ohm
- Vcore: 0,36 Ohm
- Memory: 55 Ohm
- VDDCI: 13 Ohm
- all 12V rails: kOhms
- 1,8V rail: 575 Ohm
- 0,75V rail: 10 Ohm

Now clearly on my card there are two resistances that stand out: 1,8V rail and 0,75V rail both read significantly lower than working RX 5600 XT.
My Ohm-meter is not really accurate on low readings so anything under 1 Ohm is more or less guessing.

Apparently something is shorting one of the rails (or maybe both). I will have to investigate further but at this moment I can not see any damaged or discolored components and nothing is heating up (to the touch at least, I do not have thermal camera).
Ideas welcome.

[dicky96]

@for.software3.45
I've just looked at an ASUS RX 5700 XT and the video is set to publish at 2pm GMT today on my channel. See if that helps you. My board has a different layout to yours, but if I can take any voltage measurements to help you (all voltages are present and correctr on this card), let me know.

It's great to hear I have encouraged someone to start GPU repair. That's brought a smile to my face that will last all day

On the bad side, usually if 1.8V and 0.7V (PEX) are both reading abnormally low resistance to ground, you have a shorted GPU core. I also mentioned this and explained it in the video you enjoyed.

[for.software3.45]

@dicky96
I will definitely watch mentioned video and thank you for being prepared to help me out.

Unfortunately I did hear this in your video and I am afraid that might just be the case here. I am not really sure why or how didn't I notice this when I've done initial resistance test.
Then again I am surprised that all of fuses are still intact.

I've prepared photo with dots (points of resistance check to ground) of different colors.
- Green colored dots are the ones that actually should have 0 Ohm resistance to GND.
- Red colored dots are the ones that should not have 0 Ohm resistance to GND but they do.
- Blue colored dots are reading around 2 Ohms, which is the same as I get on output of NB685A on the back of the card.

So either one of elements is shorting both outputs (from FFK 153 & NB685A) to GND or the core itself is shorting this two to ground.
I will try to lift one side of these capacitors up and see if anything changes. Hopefully it will, but chances are probably pretty low.

[for.software3.45]

Just to update situation on this one. Two 0 Ohm resistors were desoldered and the lower part was not shorted anymore. Since there is nothing else on the upper part of photo from post #10, I am pretty sure GPU itself is faulty.

Thank you all for help, but this one is going to the parts bin.

@dicky96:
I have watched your video and I can only wish it was that easy in my case.

[prettyfury]

Searched for FFK IC in Aliexpress. Result was MPS NB691GG.
Couldn't find the DataSheet Online.
FFK - This IC is Similar to MPS MP2329 - QFN 11 (2mmx2mm).
Please Find the Attachement of Pin Out and Pin Function.

I have a Sapphire Nitro Plus RX 5700 XT - GPU Heats up, Fan turns on for a fixed time but the Graphics Card is not detecting.