[SOLVED] XFX RX-570P45D RX-57045DDT VA2 RX570 AMD 4GB No power on all rails

[taxxin]

Hello everyone I have an exciting new problem today, I just picked up a RX570 on auction here, the seller says that it cant start.

Testing:

Firstly tested for shorts all resistances seem okay, no obvious shorts present. However the Vcore was quite low (0.7 ohms) ive written the resistance readings on the board in white marker.

Symptoms: When powered on it appears that all the 12v comes into the card just fine. It powers on but the fans do not start spinning, (this could be normal because AMD cards are known to turn off fans when temperature is low). It continues to run for about 10 to 20 seconds and the whole computer will reset and start up again.

Checking with a multi-meter it looks like no power is coming into the VRAM VCORE and the other power circuit near the 12v Connector.

I noticed that some of the pins on a controller chip look dark or dirty it is closest to the PCIE slot.

Initial guess is that it is one of the smart-ass power controllers doing a check up, finding some kind of fault and then telling the card to shut down. Im not sure where to start everything on the board looks reasonably nice and tidy, most likely it is detecting a short from a faulty mosfet then running power protections on it. I guess my next step might have to be checking the power controller datasheets and see if they are getting all the power they need. It would be quite handy if they had error codes or diagnostic capabilities. It would need a good explanation as to why all of the power rails are not functioning at all including the FAN.

what would you do next? I dont have a fancy ESR meter or a hypersensitive ohm meter. Also if anyone can get me the SCHEMS for this I would be grateful!

[petehall347]

hard to see the pic but looks messy on topside of u900

[dicky96]

Originally Posted by taxxin

I dont have a fancy ESR meter.....

There is nothing particularly fancy about my ESR meter!! .

I just looked on ebay or Ali (cant' even rememebr which) for an ESR meter about 6 years ago and that is what came up lol

0.6R on a GPU Vcore seems quite reasonable to me

[bensatrio8]

you can see here for a reference, because it is the same card.
I also on my way repairing same card, it is from HIS but has the same exact layout with XFX.

[taxxin]

Originally Posted by dicky96

There is nothing particularly fancy about my ESR meter!! .

I just looked on ebay or Ali (cant' even rememebr which) for an ESR meter about 6 years ago and that is what came up lol

0.6R on a GPU Vcore seems quite reasonable to me

hhahaahahahah, yes thank you for introducing me to the ESR meter, i had no real understanding of that tool until i watched your gpu short video explaining how the A/C generated waves are affected by the inductors.

[taxxin]

Originally Posted by bensatrio8

you can see here for a reference, because it is the same card.
I also on my way repairing same card, it is from HIS but has the same exact layout with XFX.

Thank you for this video, I will check it out and see. I guess the key will be in understanding how the power controllers function. I also noticed that this particular card does not have a soic8 bios chip, instead it has a dual bios switch, not sure where that data is stored.

[bensatrio8]

Originally Posted by taxxin

Thank you for this video, I will check it out and see. I guess the key will be in understanding how the power controllers function. I also noticed that this particular card does not have a soic8 bios chip, instead it has a dual bios switch, not sure where that data is stored.

It have 2 bios chip in the back, near the apl5932 chip.

[taxxin]

Originally Posted by bensatrio8

It have 2 bios chip in the back, near the apl5932 chip.

Oh yes you are right, I couldn't read the laser etching before but now i see they look like 256kb chips

[stj]

for laser etch'd chips, get a colour pencil with a light colour like yellow.
rub it across the chip in several directions - then wipe it off with a tissue or rag.
it will fill the etched part with bright colour that is easy to read.

[taxxin]

Originally Posted by stj

for laser etch'd chips, get a colour pencil with a light colour like yellow.
rub it across the chip in several directions - then wipe it off with a tissue or rag.
it will fill the etched part with bright colour that is easy to read.

Thank you for the tip, works a treat!

[taxxin]

Okay so ive got a small breakthrough, its this "High Input Voltage 8A PWM Converter WithAdj.SoftStart" APW8713 which is the first rail to power up and i believe supplies power to the other controllers . I have checked all the components around it and it seems that this part is faulty. because none of the other power rails including this one is working.

It just so happens to be the dirty looking thing with corroded pins near the PCIE slot. Anyways i have ordered a new part locally, hopefully it should fix the problem with this card.

[taxxin]

update:

I Replaced the Buck converter and still the same, So anyways i decided to look through the datasheet and check the particulars, In short ive found that the POWER OK pin reads ungodly high resistance like 1.5m Ohm which according to the typical circuit chart should be only 100k ohms. This is probably why the POK is not receiving sufficient voltage when the 5V VIN is coming, therefore the chip does not perform its start up functions.

Gonna try to find this resistor and do something about it

[taxxin]

Okay I was wrong, the reason why it was so high was because there was no pull up resistor! it was it either knocked off or never came with one, so anyways I decided to pull one off my trusty scrap lenovo board and tried it. I got about 97K ohms reading.

I threw it all back together again and tested it, same thing, so i decided to test it out and found that there is 5V coming out of the LDO nearby but there is no 5V coming into the VIN of the chip, i probed the capacitors corresponding with that power line and there was no voltage at all? this probably explains why its not working.

Also on a side note i found some capacitors at the pcie slot knocked off (they look like they go straight to the GPU DIE), some even pulled the trace pads up with it, spent some time replacing the missing ones and adding some new traces with some thin gauge wire. what a pain in the ass.

[taxxin]

Messed up, I checked the schematics and it shows that the BUCK controller POWER OK is connected to the VCC not VIN must be used in a different configuration as stated in the datasheet.

So im gonna have to take that resistor out and keep on investigating these missing voltages.

[bensatrio8]

CMIIW but when you see the power sequence page in that pdf, it is stated that APW8713 is connected with the GS7103. So, there is a relation between them or maybe check for busted component around them.

[taxxin]

Originally Posted by bensatrio8

CMIIW but when you see the power sequence page in that pdf, it is stated that APW8713 is connected with the GS7103. So, there is a relation between them or maybe check for busted component around them.

Thank you for the tip, Ive been trying to read that PDF for several hours last night, its not very clear about the EN and the 0.8v EN which is also 21 in red. Its so hard for me to know which circuits are connected by what. I dont know what all these numbers mean? are they the order of power sequence? Its probably not 0.8v its probably 1.8v,

But i am yet to check, I need to setup my test motherboard and set it up with the PCIE extender cable that i got, just like exactly how dicky did to diagnose his 9800gt card. because at the moment i cant check the card inside my computer its hard to get underneath.

[taxxin]

Okay guys im stuck here and need some help with these diagrams,

The buck controller chip is getting 3.3v to ENABLE, VIN, also 12v Coming into the AIN (ANALOGUE INTERNAL SUPPLY) (mine doesnt have that mine is just not connected) but nothing is coming in or out of the VCC 5v pin, The schematics make it look as if its supposed to output the 5v?? the datasheet makes it look like it should have 5V coming into it?

Could it be the chip that is cooked or shorted?

[dicky96]

That APW8713 datasheet seems very confusing. In the description of the pins it says connect Vcc pin to ground!!!??!

Vcc Supply Voltage Input Pin for Control Circuitry, Connect +5V from the VCC pin to the GND pin. Decoupling at least 1μF of a MLCC capacitor from the VCC pin to the AGND pin.

But in the typical applications shown on page 12 and 13 it shows Vcc is clearly an input voltage

In the block diagram on page 11 it is clear Vin is the supply to the built in high side FET, but it isn't clear what Vcc is - internal voltage or an input

So the answer is - I don't know

[taxxin]

Originally Posted by dicky96

That APW8713 datasheet seems very confusing. In the description of the pins it says connect Vcc pin to ground!!!??!

Vcc Supply Voltage Input Pin for Control Circuitry, Connect +5V from the VCC pin to the GND pin. Decoupling at least 1μF of a MLCC capacitor from the VCC pin to the AGND pin.

But in the typical applications shown on page 12 and 13 it shows Vcc is clearly an input voltage

In the block diagram on page 11 it is clear Vin is the supply to the built in high side FET, but it isn't clear what Vcc is - internal voltage or an input

So the answer is - I don't know

AHHAAHA, thank you at least for the reassurance that its actually confusing and that its not just me whos stupid.

[SMDFlea]

The RX570 schematic doesn`t have an APW8713 it has a GS9238, the apw pin 7 is NC whereas the GS has pin 7 - Supply Input for internal analog circuitry.

The 2 aren`t pin to pin compatible.Did someone replace the GS9238 with the APW8713 ?

GS9238 datasheet https://datasheet.lcsc.com/szlcsc/20...-R_C532950.pdf
.