Nvidia Quadro K2000 2Gb

June 16, 2022
 |  2 Comments

Seller Note “This card just displays a black screen on turning pc on sold for parts only”

Summary

  • Appears to be missing the VMem rail.
  • Resistances
    • VCore – 12.4
    • VMem – 104.8
    • PEX – 235
    • 5V –
    • 3.3V – 10.8K
    • 12V – 11K+
  • Missing VMem rail – replacing the buck converter EM5303A fixed this
  • Card is working

Investigation into Key Components

VCore PWM

Quadro K2000 – VCore PWM & Drivers

I think this is a Richtek RT8809BZQW https://www.richtek.com/assets/product_file/RT8809A=RT8809B/DS8809AB-02.pdf

It appears to drive a pair of MOSFETS (TBC)

Power Management

Quadro K2000 – Power Controller

Its APL3516C https://datasheetspdf.com/pdf-file/761940/ANPEC/APL3516/1

Missing VMem Rail

Quadro K2000 – VMem synchronous rectified PWM controller

Its EM5303A https://datasheetspdf.com/pdf-file/807405/CYStechElectronics/EM5303A/1

This also appears to drive a pair of MOSFETs (TBC). I need to check this in more detail since VMem is missing.

PIN 5 – VCC is 12v

PIN 7 – EN is 5v

So, on EM5303A 

pin 2 (UGATE) -> Gate of Q13 (SM3119NA N-Channel Enhancement type MOSFET)

pin 4 (LGATE) -> Gate of Q10 (SM3116NA N-Channel Enhancement type MOSFET)

Together these components, with a single inductor and capacitor, form the VMem VRM. My guess is that the PWM controller EM5303A could be faulty since it has EN and VCC are present. On a working K2000D, a PWM signal can be seen on the gate of Q10 and Q13, and a VMem of 1.37V is present. This was correct! Replacing EM5303A from a donor card has fixed the VMem rail and the card seems to be working now 🙂

Testing

  • MATS 10Mb, 25Mb, 50Mb – passed
  • Kombuster (some), the card worked by is not suited or powerful enough for this kind of benchmark
  • 2 x Display ports and DVI – working