EVGA GTX 780 TI

February 8, 2022
 |  No Comments

Seller Note “Untested”

Summary

  • No shorts, resistances looked ok except memory which is 76 Ohms
  • Blank screen, main voltage rails appeared to be there
  • vcore 0.99v 2.9Ω
  • vmem 1.62v 76Ω
  • PEX 1.08v 216Ω
  • 5v rail voltage is but, but the resistance is very high and climbing
  • 3.3V rail reads ~850Ω, but it’s voltage is reading 3.9-4V
  • The lights are on, but no one’s home! Need to establish GPU health
EVGA GTX 780 TI

Memory is Hynix 7.0Gbps H5GQ2H24AFR-R2C FBGA(170ball)16Bank, 1.6V – So 1.6v is correct.

EVGA 780TI BIOS

Since all voltages appear normal and resistances may be too, I want to check whether the GPU is healthy:

  • BIOS
    • VCC is about 3.3v
    • Cannot see /CS chip select, data in or data out on oscilloscope
    • Data in SI (pin 5) – None detected
    • Data out SO (pin 2) – None detected
    • Crystal frequency present SLK (pin 6) – I cannot see this on the chip pin, but the crystal is working.
    • Need to double-check check BIOS circuitry, but could be a sign the GPU is dead.
    • The BIOS contents can be read successfully using a CH341A reader
  • PCI EX Checks
    • REFCLK+- 3.3 MOhms – ok
    • SMCLK/SMBUS for shorts – none
    • Shorts on data lane capacitors – none
    • Differences on receiver differential pairs (diode mode)
      • 0.753v (lane 15)
      • 0.734v
      • 0.745v
      • 0.752v
      • 0.757v
      • 0.758v
      • 0.752v
      • 0.752v
      • 0.739v
      • 0.856v
      • 0.710v
      • 0.717v
      • 0.727v
      • 0.755v
      • 0.752v
      • 0.752v (lane 0)
    • Differences on transmitter differential pairs (diode mode)
      • 1.137v (lane 15)
      • 1.136v
      • 1.149v
      • 1.142v
      • 1.143v
      • 1.066v
      • 1.132v
      • 1.140v
      • 1.111v
      • 1.068v
      • 1.142v
      • 1.139v
      • 1.145v
      • 1.142v
      • 1.135v
      • 1.142v (lane 0)
  • PEX capacitors good (check capacitance & ESR) – All capacitors seem in range. Not checked ESR.
  • RESET Signal should go high – PEX reset drops and then returns to the 3.3V voltage (seems to read as 3.9V)
  • Crystal appears to output a signal

Update 29/02/2023 – Checking again

This card always leaves me feeling a bit uncertain, it most likely does have a core issue, it’s just that I cannot prove it.

Observations

  • The core gets hot, so is likely taking VCore.
  • I don’t think the card detects (BIOS not read).
  • I don’t understand why the 5V rail (voltage present), has a very high and climbing resistance?
  • The 3.3V rail resistance appears normal, but it’s voltage reads ~3.9/4V?
  • It feels like there could be something I am missing here.