gigabyte x870i aorus pro ice stuck on dram light, ddr5 power good voltage 0 when ram stick installed

[Apoco_ lips]

I have a gigabyte x870i board I am attempting to repair. I have a working board as well to reference. The board is stuck on the dram light, wont post and I found the power good pin has 0 volts when powered on with a stick of ram installed. With out the ram it is 3.3v. Another weird issue with the board is when trying to use qflash to update the bios, the qflash light blinks for about 15 seconds then turns off for 30-45 seconds and then turns back on solid for about 3 minutes. My working reference board will flash the qflash led for about 15 seconds, turn off for a few seconds then turn back on and flash for around 4 minutes. On my working board my usb stick with the bios file will flash about 3-4 times per second, on the non working board it flashes about once every two seconds. I am trying to figure out where to go from here. I have traced the power good voltage source. It is just 5vdual_ddr5 connected to a 4.7k resistor and on the other end of the resistor it is connected to the power good pin of the ram socket (ddr5 pin # 147) and a 10k resistor to ground. The power good pin has the same diode and resistance readings on the broken and working boards I have. It seems like its the ram stick pulling the power good pin low. I have tried multiple sticks of ram, all test good and post in the other working board. Also the 5v for the ram is steady at 4.9v, reads the same on both the working and non working board. At the moment I am not sure what is causing this behavior with the ram and the qflash. If you have any idea I would be excited to hear them, thank you

[DynaxSC]

I would bet the problem is with the board, not the rams. PG RAM signal does not get high probably due to non finished ram initialisation. My experience with a few AM5 boards is that such issues have to do with partially damaged chipsets. Repaired already a few of AM5 boards with similar behaviour by exchanging chipsets. My impression is these new AM5 chipsets are quite fragile, running too hot (very small die size, too much thermal load on a very small area). They are much smaller than Intel chipsets, and run also 5-10 C hotter than Intel chipsets, although Intel chipsets are much more complex, are in fact CPU's with internal OS, the so called Management Engine, and much more functionalities. So I would suspect it's the chipset. It would be good to connect a debug code card, eg. an TL631 to the board, to see on which debug code the board hangs. Another very common issue are missing smd parts on the board.

[Apoco_ lips]

Thank you so much for the reply. I actually was able to fix this board yesterday. Turns out the board had a bad chipset like you mentioned. A few resistors and test pads near the chip set showed a short so I replaced that and the shorts were gone but I still had the symptoms mentioned above. I gave up on the board until a week ago and decided to swap the sio and I think its called the ecd chip and the bios chip as well which finally resolved the issues.

I have another x870i aorus pro ice that I am working on now, it had the ecd chip improperly removed and had 5 damaged traces, 3 of which were torn off the board. I got the traces repaired and the ecd chip replaced (the original ecd chip had pins 52 and 53 burned, I believe those pins were connected to rgb headers) all the measurements on the chip seem to be the same as my working test board but this board is cycling between the cpu light and the dram light (my working board on start up lights up the cpu light then dram light then back to the cpu light then back to the dram light and then vga boot. This broken board goes cpu to dram, hangs on dram for about 9 seconds then resets back to cpu and to dram for 9 seconds and constantly repeats that) It also has the PG ram signal at 0v. I have all the heatsinks and covers removed from the board and checked it against the complete working board and I do not see any missing smd components. I will look into getting the TL631 board to see if there is any more information I can get out of this board. Do you have any ideas of where to look next? Im trying to find information on the start up sequence of the board and what self tests/signals it is looking for but I have not found that information yet. Any help is appreciated, thank you

[Apoco_ lips]

I would bet the problem is with the board, not the rams. PG RAM signal does not get high probably due to non finished ram initialisation. My experience with a few AM5 boards is that such issues have to do with partially damaged chipsets. Repaired already a few of AM5 boards with similar behaviour by exchanging chipsets. My impression is these new AM5 chipsets are quite fragile, running too hot (very small die size, too much thermal load on a very small area). They are much smaller than Intel chipsets, and run also 5-10 C hotter than Intel chipsets, although Intel chipsets are much more complex, are in fact CPU's with internal OS, the so called Management Engine, and much more functionalities. So I would suspect it's the chipset. It would be good to connect a debug code card, eg. an TL631 to the board, to see on which debug code the board hangs. Another very common issue are missing smd parts on the board.

Also on the boards you repaired that had a faulty chipset, did they always show a short somewhere on or near the chipset?

[Apoco_ lips]

Just using this thread as a diary I guess, I wanted to rule out a bios issue with this 2nd board I am attempting to repair. The qflash seems to work fine on this board but again it just cycles between cpu and dram light. My plan is to start taking parts off the working reference board to see if I can get it to have this same cycling debug light behavior so I started with removing the bios chip to see what the behavior would be. For some reason removing the bios chip and improperly reinstalling (a few pins not being fully soldered) it caused the atx power on pin to be shorted to the power button positive. So I had 4.4v on the pwrbtsw + that should have only been 3.3. No idea how or why that happened. I studied a boardview of a different gigabyte board view to find where those circuits overlapped and ended up removing the ecd chip and the short went away. I checked the atxpwron and pwrbtsw pins on the chip that were measuring a short on the board and they were no longer shorted. I checked the board and no short. Reinstalled the ecd and the short returned (Carefully inspected for bridging but did not see any) removed the chip again and the short was still on the board. Then I decided to check the atxpwron and pwrbtsw pins on the sio since they were near each other, no visible bridging, it was the factory solder so makes sense. I added leaded solder anyways and wicked away the excess and reinstalled the ecd chip a 2nd time and finally no short. I attempted a qflash before finding this short and that corrupted the bios so I had to do the qflash again after all that and finally back to a working reference board. At least in the processes of causing this issue I learned a lot about the vddio circuit and tracing its enable signals since that was the first area I noticed a missing voltage after attempting to power the board. Still no idea what is causing the other board to cycle debug lights. It seems like its just restarting after waiting several seconds on the dram light. I will keep at it, if I can figure this out it will be the 4th x870i aorus I have fixed, already decent progress with the trace repair and ecd replacement