Gigabyte RTX 2070 WINDFORCE OC 8G (GV-N2070WF2OC-8GD) - missing PEX voltage

Card like in the topic. Here's matching boardview I posted recently:
schemat-gigabyte-gtx-1660-ti-windforce-oc-6g-gv-n166twf2oc-6gd-t71608.html

I have replaced U13 (GS9216) because of low resistance on L504. After that the resistance looks healthy at around 13ohms.
Unfortunatelly U13 still does not get enable signal on pin 2. I have traced as far as D505, it does not get any input on pin 1, hence not producing anything on pin 3, U505 does not get input on pin 1 and U506 does not get input on pin 1. All other voltages are present on each chip in this area, especially 3.3V. I think it can have something to do with the 12V plug detection but I could not figure it out beyond D505.
The card is missing one memory chip, but I don't think it should affect the PEX power on.

Hello!

Specify the voltages to ground on all coils.

matic написав:Hello!

Specify the voltages to ground on all coils.

Hi matic!

So the voltages are:
L18 - 12V OK
L1 - 5V OK
6 VCore coils - 0,8V OK
L503 - 1,8V OK
L504 - PEX voltage missing
L21 - 12V
L16/L17- VMem missing (because no PEX)

I'm sure that restoring PEX voltage will make the card work. The resistance on memory coils is 28ohms, so pretty healthy for Samsung memory.

I also think that the problem is related to the missing THERM_OVERT* signal on D505. The pin has low resistance, around 120ohms. Is it normal because it connects to the GPU? Or that's the fault? There are some similar cases related to THERM_OVERT* and a guide on RepairWiki, with GTX 1070 as an example. But I started checking all the diodes and nfets that connect to it and cannot figure it out.

Measure the voltages to ground on pins 1, 2, 6, 7, 9, 20, 21, 23 of chip U13.

matic написав:Measure the voltages to ground on pins 1, 2, 6, 7, 9, 20, 21, 23 of chip U13.

Hi matic!

The voltages are as below:
1 - 3.3V
2 - 0V
6 - 12V
7 - 12V
8,9 - 12V
20 - 0V
21 - 5V
23 - 0V

Also, I have compared U13 with U15 because these are the same chips. Pin 6 on U13 has 12V while on U15 it only has 0.85V. R243 next to U13 has 12V on pin2 and 12V on pin1. Compared to R291 next to U15 and it has 12V on pin2 but only 0.85V on pin1. But it's strange, both resistors have correct resistances, as shown on the boardview. R243-261k and R291- 255k.

Do you think R243 can be faulty anyway?

check 1v8_main_powergood, i had same issue, mosfet was broken and pulled that signal to 0.6v

ankheg написав:The voltages are as below:
1 - 3.3V
20 - 0V

Are these two measurements correct?

- Measure the voltages to ground on all pins of chips U505, U506.

Are these two measurements correct?

pin 20 yes 0V
Pin 1- I measured a few days ago and it was going to 3.3 at startup and then going back to 0. Yesterday when I checked again, 3.3V was staying on longer, so I though it's fine. Today, when powered on for the first time, it started at around 2.5V, went up quickly to 3.3 and stayed on quite long. Now, after powering on it either goes to 3.3 for a moment, sometimes 1-2 seconds, sometimes longer even like 20seconds. But sometimes it just goes to 0.02V straight away. So pretty random...

U505
1 - 0,3V
2 - 1.8V
3 - gnd
4 - 0V
5 - 3.3V

U506
1 - 0V
2 - 3.3V
3 - gnd
4 - 0V
5 - 3.3V

So pin 1 on U505 seems low

@panovchina: Do you mean PS_1V8_MAIN_PGOOD? It's fine for me, 1.8V. Also I don't see it connecting to any mosfet ...

Measure the voltage on capacitor C1304.
If there is 0V, measure the voltage to ground on pin 1 of resistor R635 and directly on capacitor C2195.

matic написав:Measure the voltage on capacitor C1304.

Stable 3.3V on C1304, even after a minute.

R635 - 3.3V on pin 1, 680mV on pin 2, resistance is fine 10k
C2195 - stable 5V

Also:
R739, pin 1 3.3V, pin2 315mV, resistance is fine 10k

R747, pin1 3.3V, pin2 15mV, resistance ok 10k

Low resistance on pin1 of D505 - 120ohms

Please, post a clear picture of chips U505/U506 area of the board with a nearby surroundings and with a visible markings on diodes D505, D507.

Both diodes have 76t marking

Was this board spilled? The solder joints on diode D505 seems dark.
Check the resistors R739, R747 (both should be 10kΩ).

I don't know the story of the card. Bought it as faulty over a year ago. I attempted to fix it several times, each time doing some small progress and now I'm stuck. D505 was replaced and soldering isn't pretty, too much solder below the diode. After taking the photo I replaced it again with another one from a donor, nothing changed. R739, R747 are both fine with 10k as mentioned in my previous post.

Remove the diode D505 from the board and remeasure the voltages to ground on pins 1, 2, 4 of chips U505, U506.

I removed D505 and 3.3V voltage appeared on pin 4 of both U505 and U506. All the way to pin 1 of R260. R260 did not have 17.4k as per boardview and no voltage on pin 2 so I replaced it as well. I also replaced U13 once again, with another chip from a donor. And soldered one missing memory chip. Now I have 1V PEX voltage and 1.35V on memory. But the card still does not give a picture. Will test later this week if it's even detected . Removing D505 helped, but it was there for a reason. So what does it mean? What is D505 for?

The diode D505 works as OR gate and disables the card in case of thermal protection (overheating) or in case there is something wrong with the 1.8V power rail.

- Measure the voltages to ground on all pads on the place of diode D505 (with the diode removed from the board).
- Measure the voltages to ground on all coils.

Diode D505 pinout:

Without D505 the voltages are:
Pin 3 - 3.3V
Pin 2 - 3.3V
Pin 1 - 0V

Coils:
L21 -12V
L18 - 12V
L1 - 5V
L503 - 1.8V
L504 - 1V
L16,L17 -1.35V
GPU (L5-L14) - 0.8V

Without D505 the card gets detected under Linux and Windows(error43) as...RTX2080....
So I checked the marking on the core and it is TU104-400A-A (so looks like it really is RTX2080)
But RTX2070 is TU106-400A-A1. No signs of soldering/BGA/flux. Maybe done and cleaned well?
Original BIOS is identical as the one for RTX2070 from Techpowerup. Both say the model is N2070WF2-8GD, same as the backplate sticker.
I tried flashing some other Gigabyte RTX2080 BIOS but nothing changed.

I removed Q6,R61,R131 - no change, power draw before and after around 0.6A
After that I removed D505 - PEX and Vmem appeared, card gets detected, power draw 1.1A

Removed also R747, no change

Maybe the pad under the GPU is pulling THERM_OVERT to ground?

Do you have any information about the history of this card?

Remove the resistor R131 from the board and remeasure the voltage to ground on pad 1 on the place of diode D505.
If you still get 0V, also remove the transistor Q6 and check the voltage again.