[ibis-macro] Back-Channel for pure statistical flow

  • From: Taranjit Kukal <kukal@xxxxxxxxxxx>
  • To: "'IBIS-ATM'" <ibis-macro@xxxxxxxxxxxxx>
  • Date: Thu, 14 Jun 2012 11:40:18 +0530

Hi All,
The current back-channel communication proposals have been limited to Tx-Rx 
getwave calls. What about Tx-RX init calls for pure-statistical flow.
High level steps would be

1.       Tx-init should send IR convoluted with initial FFE taps to Rx-init.

2.       Rx-init should modify IR based on Rx-filters like CTLE -curves 
(expressed as IR), RDL-parasitics (expressed as IRs) etc.

3.       Rx would then estimate what FFE taps would make this modified-IR even 
better and send recommendations back

4.       Tx would modify....and so on...

I understand that the back-channel for pure-statistical flow would have limited 
use but then if we are in a mode of allowing statistical flow to completely 
characterize (though less accurate and ltd.) the Serdes, we need to extend the 
back-channel to init functions as well.

I would like to hear people's opinions so that we can take next steps on 
bringing more details to this proposal.


Other related posts:

  • » [ibis-macro] Back-Channel for pure statistical flow - Taranjit Kukal