Gigabyte Aorus 1080ti - card not initialising #2

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Powered off, powered on - no image

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Checked ram straps on back of card, all appear fine (100k resistors x 3)

RAMCFG = 00010 (same config as other card)

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Also checked booting plugged into DVI port (this switches control to the other bios chip) - same story, no image.

I think its a poor connection under the core...

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Cleaned back of card, mainly caps on back of core and caps on back of ram chips. Also cleaned the PCIE connector as looked a bit dirty.

Powered up with DVI connected, booted fine with image but device manager said problems. Flashed the DVI bios chip (can only be done if cable plugged into that port).

Rebooted and all fine. Powered off completely, and back on again - did this several times and still ok.

Booted back into MODS/MATS with DVI connected - fine.

Ran mods test - PASS

Running MATS test again now (512MB) - all seems fine and can see progress on screen now as primary card.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Ok, powered down and moved the cooler about a bit. Powered up again and no image.

Reflow needed

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Right then.

Did a ghetto style reflow on the gpu. Fashioned a stand out of a sturdy box with a hole cut in the top, hot air below preheating the card to 100c. Used some NC-559 flux around the edges of the chip (copy I believe from Amazon, not had any issue with it).

Tried to follow a reflow profile as close as possible, let the board soak at 100c then used hot air on the top (no nozzzle attached) to bring the chip up to 220c (Quick 861DW set to 380c / 70 air)

Let it set there for 120secs (moving the heat around, but actually focussing more on the bottom edge of the chip as this is where the lines from the bios come in).

The upped the temperature and the air speed to bring the chip to 250c for 90 seconds.

All the flux got sucked under the chip and evaporated, very little clean up to do.

Let the card cool and popped in the test bench, fired up first go with a picture.

Need to run some more tests, powering up and down and moving thw card around. Will report back, but looks ok so far and the card hasn't been killed.

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Right then.

Did a ghetto style reflow on the gpu. Fashioned a stand out of a sturdy box with a hole cut in the top, hot air below preheating the card to 100c. Used some NC-559 flux around the edges of the chip (copy I believe from Amazon, not had any issue with it).

Tried to follow a reflow profile as close as possible, let the board soak at 100c then used hot air on the top (no nozzzle attached) to bring the chip up to 220c (Quick 861DW set to 380c / 70 air)

Let it set there for 120secs (moving the heat around, but actually focussing more on the bottom edge of the chip as this is where the lines from the bios come in).

The upped the temperature and the air speed to bring the chip to 250c for 90 seconds.

All the flux got sucked under the chip and evaporated, very little clean up to do.

Let the card cool and popped in the test bench, fired up first go with a picture.

Need to run some more tests, powering up and down and moving thw card around. Will report back, but looks ok so far and the card hasn't been killed.

Thanks for the updates. Let us know how is goes.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Tested a little and although much better, still not rock solid.

Ran MATS again and the F1 bank is also showing a few errors that vary when heatsink is jiggled (it did this when first tested, but went after heating the chip). Will pull that chip and hopefully reball it.

In the meantime I have been cobbling together a BGA reflow station using:

800w heat plate
Quick 861DW with a modified 45x45mm BGA nozzle
Stand to accommodate the gun from above.

Spent about £60 so far

Would like to introduce a PC410 PID thermal controller (https://www.aliexpress.com/item/Digi...2f352e0eAfvX5R) , but don't want to butcher the heat gun.
Might get an IR head from aliexpress and use that in the future.

https://www.aliexpress.com/item/-/10...2f352e0eAfvX5R

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Actually works pretty good as-is, just need to keep an eye on time/temps. Was cautious with the hot air speed as didn't want to blow components off the top of the board. Thermocouple on bottom of PCB so not directly in air flow.

Practised on an old 512mb Radeon card (GPU chip much smaller, but wanted to get a feel for it). Chip came free at 205c measured from under PCB. Means the components on the bottom wouldn't come loose with unleaded solder. Pre-heater set to 180c, hot air at 400c / 85 air.

No popcorning as let heat below soak into board first (board reached 110c with just preheat as air gap between hotplate and pcb)

Cleaned up the board and chip nicely afterwards. Kept the pre-heater on while cleaning the board to aid the solder wick from not ripping pads.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Moved back onto the 1080ti - used same method to remove F1 ram bank (without BGA nozzle, just straight out of the gun). Came off really clean, all pads perfect on chip and board.

Reballed chip, put on too much flux - went to heat and all the balls moved - doing it again

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Ok, chip reballed and reinstalled on board.

My "reflow station" made this a lot easier and more controlled. Have both hands free and having controlled heat top and bottom of board made it a much better experience.

Popped the card into the test bench (without heatsink) and powered on. Card worked straight away

Will reassemble and see if the issue has gone, hopefully won't have to try and either reflow or reball the gpu core (although I secretly want to try and see if its doable with this set up )

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Card back together - still works.

Running MATS now - 512mb test

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

MATS passed, now in windows running Heaven benchmark

[zefrench]

Re: Gigabyte Aorus 1080ti - card not initialising #2

What kind of 800w hot plate did you use? Feel free to post more picture of the setup showing the thermocouples

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

It's a Persder 946A i got 2nd hand from eBay for £45 + delivery - goes under different brand names. The newer models have a REX C100 PID controller, this is more basic (similar one here https://www.thanksbuyer.com/ansai-94...-station-42399).

The thermocouple is from the multimeter and I taped it directly under the chip on the other side of PCB - reasoning was that it would be close to the balls of the BGA but not being blasted with heat from the hot air so when the temperature read about 15c below the reflow temperature it was about right.

Bought this stand from Amazon (https://www.amazon.co.uk/gp/product/B094CXMKYQ) which works quite nicely to hold the hot air wand - I just needed to extend the pole at the back that the clamp is attached to and make it a bit taller to give more clearance over the hot plate.

Can get some better pics later.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Dug out an old usb microscope and attached it to one of the flexible mounts that came with the stand - not super great, but can position it to see the balls under the chip. Would work for a gpu core, but probably not ram. See attached

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Would be set up a bit like the attached (BGA nozzle is not on straight as just did it quick to illustrate)

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Quick update - switched card to my main PC with watercooling block attached.

Running ok, but occasionally glitching. Ran MATS again several times and erroring on bank E1 sometimes (only like 240 errors).

Used the "reflow station" to get the chip to reflow temperature, gave it a few nudges and it sprang back into place.

Running tests again now and looks like errors are gone (using the memory tests in MODS instead of MATS as they are more thorough and test more items)

I intend to keep this card and sell my daily driver card (1080ti Aorus xtreme) instead just in case there are any minor issues as don't want any returns from ebay.

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Card wouldn't pass a 4GB MATS test before (or a full 11GB one), it does now

[benwaterson]

Re: Gigabyte Aorus 1080ti - card not initialising #2

Been running the card for a few days now, overclocks core and ram pretty good and has been completely rock solid stable. I am marking this as a success

In other news I thought I would see if the "reflow station" could remove a large BGA (ie the GP102 GPU core from card #1)

Seems like it can and pretty cleanly too - had to up the bottom heater plate to 300c and used 385c / 90 air from the top) - chip came off fine.

There is 1 suspect lifted pad on the chip - A52 which is just a ground pin, so guess that's ok.

Still don't know if this chip works fully - any way to test resistances on the pads to determine?