Subject: Meeting Notes from 26apr04 Conf Call to get a perspective on direction The following are my thoughts on this call to share with everyone. This call was not a general meeting, everyone in the ad-hoc was invited to listen in or talk if they wanted, and the goal was to help me establish a more firm direction. I think it was a good discussion. -------------------------------------- Attendence: Peter T Brian S John D Adam H Mike L Steve A Graeme B Schelto Mike A Brian Seeman ... Gave a presentation to help identify problem and solution space in multiple dimensions. It was a good discussion. Three of his many points are 1) Channel physics are different from transciever physics, 2) Try to make the channel set a small subset of the transciever space, and 3) Real channels will very all over the map from design to design. We discussed concepts such as pulse response and statistical eye. Discuss advantages of stateye -best fit -includes the large channel set and small transciever set -provides channel BER Discuss disadvantages of stateye - not many have validated it. - black box - SI designers feel constrained to receiver. For developing the initial model for informative use and maybe later, normative, there are three basic directions: - Simple Compliance Approach - S-Param Approach - Pulse Response Approach The focal point of the discussion on this was if pulse response would cover our needs better then s-param approach. To try to get a feel for the way people on the call where thinking, I asked each what they felt about the approaches. Pete- Stateye is complex and many people have not had a chance to follow. How do you scale the value crosstalk and jitter. What other tests do we include. *** How is jitter included in stateye. no intrinsic jitter included, just transmitter Dj/Rj. unaddressed tx and rx jitter. still needs to be accounted for along with pulse response. - pulse response and what aspects does it lack, how do you read the output and what else do we need to include like deriving BER and compliance. Adam- Brian's slide 7 really covers a lot. - false positives are always a problem. False negatives are okay, makes IEEE standards robust. - simple methodology with limit lines and margins or we could have a complex method similar to OIF. - The advantages of stateye have yet to be proven. - what do we need to deliver for adhoc and what do we need to deliver in the standard. - feels we need more then a stateye approach that the adhoc needs to deliver. We need the channel. Not a definition that requires the tx and rx as well as the channel. - we need to come up with a channel definition that addresses a certain need. - once we have picked the modulation scheme, etc, we can refine. But we need to put a stake in the ground defining the channel. Steve - two types of channels exist ... existing and next gen designs. - wanted to develop the model around new channels. Mike L - was fine with finding an informative point to start with and refining as we go to something more complex. John D - - stateye evolved ... agreed with that - believes it to be integrated complex problem - comfortable with SDD21 as informative budget. Graeme - - stateye script is entirely based on pulse response. - a stateye approach is to far becasue we don't know the code, tx, and rx - forced into giving equation based form to get things moving. ---------------------------------------------------------------------- Joel's direction for Wed's call: - I think we have conformance on SDD21 Mag as an informative place to start. - I heard from people that we need to establish SDD11 Mag and SDD22 Mag (SDD12 should mirro SDD21 ????) Four Port data - Isertion Loss Mask and Return Loss Mask. - If we add group delay on this (input from Jeff Cain), it makes the SDD pretty solid. - We need NEXT Mag and phase for diff aggressors (input from Adam and Jeff) - multi-disturber. - We need FEXT Mag and phase for diff aggressors (input from Adam and Jeff) - multi-disturber. - measurements from NEXT and FEXT should be 2port differential with open ends terminated aggressor and victim.