UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

AR# 34418

Virtex-5 GTP sis kit known issues

Description

This article describes the known issues of the Virtex-5 GTP SIS KIT.

Work-arounds are provided.

Solution

HSPICE
 
  1. The Hspice WinXP issue in v1.0 with environment variables has been fixed by Synopsys.
    v2008.03 does not have the issue
     
  2. The intdattyp parameter for the LXT/SXT package model wrapper needs to be updated to ri.
    ma gives suboptimal results. 
    Intdattyp determines the type of interpolation used based on the data format.

 
ELDO
 
v2.0 Known Issues
 
  1. V2.0 will ONLY work with HyperLynx 8.0 Beta2.
     
    1. The models (GTP and GTX) have been updated with a new encryption which will NOT work in HyperLynx 7.7.
    2. The .ffs project files in this kit have been updated to a new version which will also NOT work in HyperLynx 7.7.
  2. v1.0 WILL work in HyperLynx 8.0
  3. A slight DC shift, about 4-5 mV was observed between kits v1.0 and v2.0.
    That appears to be a DC operating point startup issue.
    More investigation will be done as time permits.
  4. The *_REFCLK.ffs projects do not seem to work when stored and run on a network drive.
    Local drive works fine.

 
v1.0 Known Issues
 
  1. The Run_**_ batch files dont start up HyperLynx

The batch file does not obtain the path for the Hyperlynx executable automatically.
 
One reason is that the windows utility called command has to be directly accessed from the windows\system32 directory.
 
Solution:
 
1.     Get the exact path for bsw.exe (the HyperLynx executable)
 
IBIS-AMI (Updated January 5th 2010)

  1. With high input amplitudes at the RX, the output of the RX models (both GTP and GTX) will be different in Statistical vs. Time Domain Simulation.
    This has been observed in SiSoft QCD.
    Statistical results should be more optimistic than Time Domain simulations.
    1. Both GTP and GTX Linear Equalizers saturate around 350-500 mVppd depending on Termination mode.
    2. The Time Domain model has a memory-less non-linearity built-in to capture that behavior, however a statistical simulation cannot take the non-linearity into account.


 
NEXXIM (Last Updated August 18th 2008)

Ansoft Nexxim Designer cannot handle commas (,) in .model statement
 
  1. Ansoft Nexxim Designer cannot handle .tran statement with tdelay
  2. Ansoft Nexxim Designer cannot handle .if, .then, .elseif, .else statements
  3. Ansoft Nexxim Designer does not interpret a as atto by default.
    Needs to have .option unit_atto=1 to work.
  4. For channels with lots of delay (long channels) Ansoft Nexxim state-space does not work and convolution is needed.
    Needs to have .option s_element.convolution=1 for long channels
  5. Ansoft Nexxim cannot handle separate quote types in .lib statement.
    Hspice has no problems with this.
    For example:

     .LIB  '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/rx65_1.01_e.lib T_res
 

6. Ansoft Nexxim cannot handle multiple .includes with library calls in the subckts, for example:


.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_typ.sp'
.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_fast.sp'
.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_slow.sp'


Change to
 

.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_typ.sp'
*.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_fast.sp'
*.include '$XILINX_V5_RIO_SIS_KIT/v5_gtp_refclk_models/nexxim/v5_gtp_refclk_slow.sp'

 
It seems to pick parameters from the last library call.
 

7. Ansoft Nexxim and Synopsys Hspice differ when s-parameter models are inserted.


GTP TX -> GTP RX without any s-parameter channel/package models seems to be close.
GTP TX -> GTP RX with s-parameter channel/package models is somewhat off.
 
This has to do with causality issues when the s-parameter models were generated.
Ansoft did a different s-parameter model extraction for a channel and in that Hspice and Ansoft are quite close.
 

8. The intdattyp parameter for the LXT/SXT package model wrapper needs to be updated to ri.
ma gives suboptimal results. 
Intdattyp determines the type of interpolation used based on the data format.


Linked Answer Records

Master Answer Records

Answer Number Answer Title Version Found Version Resolved
34417 HSSIO sis kit Known Issues - Master Article N/A N/A

Associated Answer Records

Answer Number Answer Title Version Found Version Resolved
34417 HSSIO sis kit Known Issues - Master Article N/A N/A
AR# 34418
Date Created 02/17/2010
Last Updated 06/10/2015
Status Active
Type General Article
Devices
  • Virtex-4 SX
  • Virtex-5 LX
  • Virtex-5 LXT
  • More
  • Virtex-5 SXT
  • Virtex-5 TXT
  • Virtex-5 FXT
  • Less