AnsweredAssumed Answered

error while simulating PCIe signals

Question asked by judson_antu on Apr 10, 2019
Latest reply on Apr 11, 2019 by weston_beal

Hi,

     I am trying to simulate eye diagram for  PCIe(gen2), between 2 FPGA's(XC7K410T-2FBG900C) on separate boards. There is a driver(DS80PCI402SQ/NOPB) on one board for the PCIe signals. Once the simulation reaches 100%, a report is generated with the following errors/warnings.(attached screenshot of the report as "report.PNG")

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------

   Messages from AMI models:

I:    AMI_Init of Rx [ U2.2p (at die)/U2.2n (at die) ] returned:

       *** Xilinx 7 Series GTX Rx model ***

       

W: Warning building plots for probe U2.2p (at die)/U2.2n (at die):

   Time=2.57923e-005: clock generated by the Rx AMI model looks unrealistic; this may result in eye closure

    

   The problem above could be due to a failure in the AMI model’s code (e.g., data-rate limitation, unsupported # of samples per bit, etc.); or could indicate the actual Rx behavior under the simulation conditions. Try modifying the AMI model’s settings and/or simulation stimulus first; if failures still occur, contact the silicon vendor for detailed support for the AMI model.

    

   Time=2.57923e-005: clock generated by the Rx AMI model looks unrealistic; this may result in eye closure

  Time=2.87043e-005: clock generated by the Rx AMI model looks unrealistic; this may result in eye closure

  Time=2.91443e-005: clock generated by the Rx AMI model looks unrealistic; this may result in eye closure

Time=2.99003e-005: clock generated by the Rx AMI model looks unrealistic; this may result in eye closure

Further warnings of this type suppressed

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

in the error they mentioned "Try modifying the AMI model’s settings and/or simulation stimulus first" but I'm not sure which are the settings that need to be modified.

I have attached the following photos for reference.

 

eye diagram

linesim

settings while running IBIS-AMI channel analysis.

IBIS model of the driver

 

Strange thing is that, while i simulated it without using the driver, i was able to get a pretty good eye. Please do point me in some direction

I will be happy to provide further informations if required.

Attachments

Outcomes