This history is going to be about my most lucky repair of the Gigabyte GA-X58-UD7, that went straight from recycling bin to my workbench.
For starters, let’s analyze physical condition. We have broken PCI-E slot, what I assumed was a choke, but it only was a holder for the broken slot, bent pins in CPU socket and burn marks on 24 pin ATX plug and flux under one of the capacitors. “Oh boy, this is going to be a trip” – that’s what I thought, but no, in retrospective this diagnosis repair was by no means easy, but also far from the worst.
Diagnosis
First of all, burn marks on ATX plug, specifically on 3.3V line. Probably as bad as signs of previous attempted repairs. I started by measuring resistance and… a few kiloohms. Not bad, not bad at all. At this point, since it wasn’t a short, I just let it slide.
After that I took al look at the socket. It didn’t look half bad, just two bent pins. Unfortunately burned. I straightened them and took a look at 1366 CPU pinout. Bent pins were VTTD and VSS, so even if I removed them motherboard would still work.
At this point I decided to hook up a PSU and fire it up. It obviously didn’t work, hell, it was even worse than not working, as it was bootlooping. Probably the worst case scenario, since half-working things are the hardest to repair.
What was weird, the motherboard would stay on without 12V CPU power plugged in and that got me thinking. It couldn’t be a short, since then PSUs OC protection would kick in. Short on VTTD? Nope, resistance looks fine. On to the internet then, in search of schematics or boardview.
I noticed that some leds didn’t light up, such as DDR_LED1. Quick look up revealed it being connected to the SuperIO chip, the IT8720F. VCC? Fine, 5V is present. RST? Goes high for a bit then low, it’s fine. For now everything seems ok, so I left it alone.
I saw different behaviour on NBT_LED1 and NB_LED1 – they would light up a second before each loop with 12V present. That got me thinking, maybe NB wasn’t getting any voltage, but no, 1.2V was present. Onto the schematic then. It shows VTTD, NB voltage, being controled by ISL6312CRZ. At first I looked for any references to 12V. And there is one, pin 29 PVCC1_2. After a quick googling for the controller chip datasheet it turned out to be a gate drive level supply.
But then I took a look at the typical application schematic. Mosfets were fed from 12V here. I quickly rushed to the schematic, only to be disappointed, as it seemed that on the motherboard transistors were fed from VCC, which was in fact 5V.
After that I saw pin 23, EN_PH4, being responsible for switching the amount of active phases. Maybe the not active phase was shorted?
No, as quick voltage check on pin 23 shows. Both phases were always active.
At this point I almost gave up. Nothing seemed to be wrong with this board. But then it hit me. I was looking at the schematics, but not looking at what do these leds indicate. Thermal/voltage indicators. Maybe the NB was just overheating?
I took a spare heatsink and pushed on the NB…
The board sprung back to life! And then as soon I let the heatsink go it stopped responding.
“Hmmm… Maybe it wasn’t necessarily thermals that kept this board from working, but pressure?”
With that in mind I pressed on the substrate next to the NB flip chip and the board worked, meaning broken BGA connection.
It was around 3 am that I made this discovery. I was so excited, that I couldn’t wait untill morning to throw this board onto the BGA machine and reflow the bridge chip. I pondered for a second whether to reflow or reball, but a quick look at my reball stencils revealed that I didn’t have the one I needed, and besides, why reball when reflow will be a sufficient and reliable repair?
30 minutes later board was cool enough to see if the reflow process succeeded… and it did! Motherboard fired up without me holding it and was responsive all the time.
Glorious success, if I do say so myself.
Unfortunately this wasn’t the end of my problems. The bracket that the NB heatsink screws into has a broken screw and the USB plug didn’t entirely go into three USB sockets. Quick enough fix – there was something stuck in 2 of 3 sockets. Third one was a bit of a problem – bent pins and broken plastic. I managed to bend pins back into their place, but the plastic divider broke entirely. I had to mask that socket, but others work.
In conclusion, we sometimes tend to overthink things, make them seem worse than they are in reality, as it was in this case. With that life lesson in mind I’m off to reapply thermopads (as each and every one is dry) and repair the NB screw.
Thank You for reading!