Gigabyte Aorus 1080ti - card not initialising

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Ok, has EN and 12v on the chip pins.

However, no 5v on VCC pin - there are few 0 ohm resistors from 5V to VCC pin 21 (R982, R942, R943, R944) I need to check next - some of these may not be populated though.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Only populated resistor is R944 - 12V_F and it appears ok.

As the other resistors are not there, there is no 5V or 12V connected to VCC pin from what I can see. R944 just connects to pin 7 (AIN)

Datasheet says "Supply Input for analog functions. Bypass VCC to AGND with a 1uF ceramic capacitor. Place the capacitor close to VCC pin.", this must be C1372

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

So looking at any schematic for a 1080ti, there is mention of APW8713 all over the lines going in and out of the 1.8v buck converter.

My guess is that this chip was used in the original design for the Titan X Pascal and the different configurations of the 0 ohm resistors mentioned above was adjusted for the AOZ1268 (also AOZ1267 is used on other boards but this is an 8amp chip, whereas the AOZ1268 is 10amp).

I have found stock of APW8713 in the UK so have ordered one to try.

Main differences are :

1) APW8713 is NC on pin 7 (AIN) - has 12v supply on AOZ1268
2) The VCC pin is used for a 5V input

Moving R944 0hm resistor to the unpopulated R982 will disconnect 12v from AIN and connect 5V to VCC and provide power to the chip.

Otherwise all pins are the same.







Thoughts?

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Not sure if this is an issue, but measured the feedback voltage divider circuit and the 2 resistors measure nothing like the diagram (measured in circuit so might be off).

R3 - should be 12.7k - measured 1.2k
R5 - should be 10k - measured 437ohms

R6 on the TON pin to VIN measures 10k (correct)
R8 in diagram is not populated
R4 present and working

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Removed R5 to test out of circuit and it is 10k, so will rule those out

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Ok, status update:

Replaced the AOZ1268 with APW8713 and moved the 0hm resistor to detach 12v from AIN (pin 7) and connect 5v to VCC (pin 21)

Now have 1.8v output and also have 1v PEX

However, have no VCORE or RAM voltage

Need to check the enable lines and also that the up9511 is soldered in correctly

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Bad news.....

Was probing to see why the enable pin wasn't pulling high on the up9511, found input to the mosfet that pulls the pin high, reading 2v so was making progress.

Suddenly there was a flash of light and blob of white hot solder formed on the board so killed the power (oddly it didn't trip the OCP on the power supply).

Appears one pair of mosfets for the ram have shorted and took out the input resistors on both sides of the board at the same time (12v). Big blob of solder on both sides of the board and vaporised resistors!

See attached schematic of where....

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Removed the mosfet (was Q953 as suspected AON6994). Ordered more.....

When will this end

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising

Please continue updating here. There is great learning from your experience.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Current status is waiting for parts and a thermal imaging camera to arrive.

Repaired the vaporised resistors (x3 - R1631, R1651, R1661) - not the neatest soldering but seems solid

Low ohms on 5V and FBVDD (not a dead short though), more like 7ohms.

Did voltage injection at 1v on FBVDD and 1 ram chip warms up so have removed that for now (could actually have been the issue I was seeing in MATS). Just need to rule out a short in the core itself as that would be game over.

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising

Current status is waiting for parts and a thermal imaging camera to arrive.

Repaired the vaporised resistors (x3 - R1631, R1651, R1661) - not the neatest soldering but seems solid

Low ohms on 5V and FBVDD (not a dead short though), more like 7ohms.

Did voltage injection at 1v on FBVDD and 1 ram chip warms up so have removed that for now (could actually have been the issue I was seeing in MATS). Just need to rule out a short in the core itself as that would be game over.

Did you look at learn electronic repairs video on the ten ways to find a short on a gpu vrm? Maybe there is a technic you can use.

https://youtu.be/WMhpKR9x7Lg

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Did you look at learn electronic repairs video on the ten ways to find a short on a gpu vrm? Maybe there is a technic you can use.

https://youtu.be/WMhpKR9x7Lg

Yup, tech cemetery and northridgefix also

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Decided to reball the removed ram chip because :

1) Never done it before
2) Waiting for parts
3) Have it ready to test if its not actually faulty.

Reballed by hand, no stencil. Think it came out pretty good for a first go...

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Seems to actually be faulty - 1ohm resistance between VDD and VSS

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

So thermal camera arrived so tested FBVDDQ @1v with bench psu (resistance on that rail is 1.6ohms)

Card is drawing 3 amps and there is slight heating of all ram chips and the core, so before and after images attached.

Memory bank D1 was removed, D0 was the unpopulated which is why there is no heat top left above the core.

(The pcie fingers at the bottom are reflecting the IR so they show as a hot area)

I can't see any other areas heating on the card otherwise, front or back.

(I have purchased another card exactly the same in the meantime that is also faulty, so this may just become a parts card now).

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Similar story on the 5v rail (resistance 7ohms) - pulls 0.9a at 2.5v

2 x DRMOS heating (2 already been removed)

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

Yikes, its had an overvoltage on the 5v rail it would seem. I know the FDMF6820a drmos cannot handle more than 6v, similar thing happened to me and when the 3rd drmos popped it took the gpu memory controller with it It'd be a shame to give up now

I think this is actually what has happened in hindsight...

5v overvoltage from MP1475 has taken out most chips that use that rail (up9511, DRMOS) and then also caused a short on the memory side.

12v input and 5v output are right next to each other on that chip, so a short there will send 12v straight across the board.
.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

I am now systematically removing the ram chips one by one to:

1) Find which are good still (if any!) to re-use
2) See if the low ohms on the FBVDDQ rail is gone once all removed

Removed 3 so far and all low ohms on the chips, 1 is 14ohm which is still too low (should be 30-60ohms)

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising

So the ram chips use FBVDDQ directly (no resistors in series), the core has 60ohm resistors in series on pins such as FBD_CMD34 (and then an assumed internal resistance to gnd from there).

Logic dictates that the low ohms on the rail is from either 1 or more ram chips, or some other unseen low ohms failed component.

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising

I am now systematically removing the ram chips one by one to:

1) Find which are good still (if any!) to re-use
2) See if the low ohms on the FBVDDQ rail is gone once all removed

Removed 3 so far and all low ohms on the chips, 1 is 14ohm which is still too low (should be 30-60ohms)

How do you measure the ram chip resistance? What pin to what pin?