ASUS DUAL GTX1070 (white)- a hopeless case ???

[DynaxSC]

Hi,

I have a very stubborn card from ASUS, I believe I made almost everything what can be done, no success:

• Card has all voltages, resistances seem to be OK, but is not recognized in Bios, neither Mats, nor Windows (mesurements on attached picture) - has Micron memories.
• Monitor not reacting, even no backlight going on
• Mats test not possiblle, card not visible
• PEX_RST* - The signal is OK, checked with an oscilloscope, it goes Low for a short period when I do board reset, then is High again
• VRAM PGOOD? It is not "anded" with PEX_RST_BUF* (no logic gate, only R0 Ω) -> ASUS is going the simplest way on this card, several logics are simplified
• Crystal 27MHz - works, has 0,9V twice
• PCIe capacitors - all OK and not shorted, not missing
• PCI data and address lanes diode mode check - all OK, not shorted, no missing, all same levels
• Reading Bios - YES, it works, the BIOS is read shortly during boot, level changes are visible on pins 1,2,5,6, then silence forever
• Uploaded another BIOS from TECHPOWERUP (86.04.50.00.AS3), then newest from ASUS site (86.04.50.00.AS12) - did not help
• Some additional measurements:
- LOWPWR_MODE_Q1 / GPIO6_NVVDD_PSI * -> High
- GPIO1_GC6_FB_EN -> Low, active, but not connected anywhere, maybe it should go to GC6_FB_EN_CONN_A32 (have only a similar schematics here, so can't verify this), maybe a resistor is missing ?
- GC6_FB_EN_Q * -> OK Low
- 1V8_MAIN_PGOOD -> OK High
- GPIO4_GC6_1V8_MAIN_EN (on D504) -> High - OK
- NV3V3 -> 3.3V OK
- GPIO2_GC6_GPU_WAKE * -> remains Low -> probably not used on this card ?
- 3V3_Aux -> OK, present
- 3V3_F -> OK, present
- FBA_VREF_Q -> OK 1.08V on Q11 and Q8
- 1V8_GPC_SP_VID_PLL -> OK
- PEX_SMCLK -> not connected anywhere (also on boardview) ?
- PEX_SMDAT -> not connected anywhere (also on boardview) ?

• GPU not becoming too hot, memories too

This card is really a big questionmark for me, everything seems to be OK, but it's not working. I'm shortly before giving up.

Anybody any ideas ???

[manu76]

Re: ASUS DUAL GTX1070 (white)- a hopeless case ???

Exactly the same thing happens to me, but with a 1060 6gb and they tell me that it is most likely the bad gpu.

[sinu0us]

Re: ASUS DUAL GTX1070 (white)- a hopeless case ???

Hi,

did you ever manage to get the card working again? I have the exact same gpu, except it boots fine... for a while, until it just seemingly at random times stops video output and maxes fans to 100.

I've only done basic diagnostics; cleaned, repasted, changed vrm pads, passed mats test, good temps.

I have a multimeter and probably the same schematic to go off as you do, but this is all new to me so don't really know what to look for :/

[DynaxSC]

Re: ASUS DUAL GTX1070 (white)- a hopeless case ???

Hi, unfortunately not, I gave it up, too much time lost. Generally my observation with this card is that the ASUS graphic boards are designed by accountants, as they are from design point of view as simple as possible, so eg. no buffers protecting GPU, etc. I much more like MSI's and ZOTAC's, as they seem really much better designed, better protecting the GPU's, and with much more efficient and quiet cooling systems and better PCB quality. So thats why I do not invest my time in ASUS grahic cards repairs any more. I also do not like EVGA cards, they have these very poor black PCB's (tending to burn with electric arcs when mosfets are getting burned and get very hot), weak cooling, and all the cards I bough for repair had broken GPU's, so VRM design is also bad, not protecting the GPU in case of an fatal high-side mosfet failure (12V on VCORE). Long time ago EVGA was a high quality card, I still have an EVGA GTX270 working, but today, as I said, accountants are the designers.

But if you want to try to repair your card, you need to check what can be done, eg checking the address and PCI lanes with reversed diode mode, and the GPU VCORE resistance, for this core it should be slightly below 0,2 ohms, values under 0,12 ohms mean usually a broken GPU - but to measure this very low resistance you need to have a special low resistance meter working with 4 wires, no conventional meter, no matter how expensive, can measure exactly such low resistance, and you get errors of several hundred percent. Also check all supply voltages if present (meaning contignously and stable present - you need to use a scope to check and see this) and if BIOS chip and contents are healthy and chip beeing read by the GPU. Details have been described on this forum several times, so just search for it, read some 10-20 posts, and you'll know what can be done. Generally this knowledge can be applied to any type of cards, so also for yours. Also your type of error is very different to my card, so the reason will most probably be completely different. In your case I would see 5 options: 1. some bad contact under the GPU or some memory chip - need to do reballing or at least reflow with an "aggressive" solder fluid, 2. some memory chip is "at the end of life" - passing simple and short mats test must not necessarilly discover an error - try to run much longer tests and possibly some load or 3-D tests also possible to do with mats (mats manual is available in the net), 3. GPU is at "end of life", 4. some VRM chip is not stable working when getting hotter (I had such a real case), 5. some other small part, eg. an MLCC cap or even electrolytic cap getting a short from time to time with rising temperature or a decalibrated resistor (low ohms smd resistors under 100 ohm can loose with time their values due to long high load time, even if the load is within the design parameters) - this type or error means long time for checking all the possible parts, check first all the 2,2 and lower ohms resistors till 0 ohms bridges, they are most probable to have decalibrated, especially the serial resistors on VCC lines supplying VRM's and other chips on the board (low ohms resistors you can measure usually in circuit).