Tool/software:
My customer currently has 2 new prototpyes., each equipped with 2 DP83826I channels.
Because of ETH-compliance test issues he wrote new values to extended registers 0x30b and 0x30c in order to increase the output voltage swing for a specific test case.
Now one of the two prototypes does NOT return back to the default values after power-cycle, (it even has been switched off over night!) The next morning it still reads back the altered values from these registers!
The customer even re-flashed the complete prototype with a Test-Firmware that does not access any PHY on the boards but still can read back only the altered values...
Up to my knowledge this device is purely register-controlled, which is a NVM-technology and should return back to initial status after a power cycle...
Is there any other explanation possible besides a defective device?