802.3ap Channel Group Ad-Hoc Conference Call on Friday, 21 May 04 from 10am-12am PDT Attendees: john.dambrosia@tycoelectronics.com rmoubarak@ti.com ahealey@agere.com william.r.peters@intel.com zwong@altera.com J.Mitchell@winchesterelectronics.com mike@mike-lerer.com m_oltmanns@comcast.net Bill.Hoppin@synopsys.com popescu@quaketech.com joec@lsil.com glen@vitesse.com aniruddha.kundu@intel.com b_panos@worldnet.att.net GOleynick@fciconnect.com schelto.vandoorn@intel.com Richard.Wolcott@taec.toshiba.com jcain@cisco.com tomaz@force10networks.com michael.w.altmann@intel.com cathyl@lsil.com joel@force10networks.com + Joel Goergen Ad-hoc Chair + Peter Tomaszewski to record minutes. Meeting Agenda + Review and discuss mask approach for channel definition (Joel) + SDD21 MagMask using Four Port data.John D has input here. + SDD11 Magand SDD22 MagMasks using Four Port data + Group Delay Mask (input from Jeff) or equiv.Petre... thoughts ... + NEXT Magand phase for diff aggressors (input from Adam and Jeff) - multi-disturber. (Jeff) + FEXT Magand phase for diff aggressors (input from Adam and Jeff) - multi-disturber. (Jeff) **measurements from NEXT and FEXT should be 2port differential with open ends terminated aggressor and victim. Discuss proposal for the interim meeting in Long Beach (Joel) The goal of the meeting is to reach consensus on the agreed upon six mask set defining an informative channel model starting point for discussion at the May 2004 interim meeting. Meeting Page numbers refer to backplane proposal conf call 21May04 - rev002.pdf + This call is preparation for next week’s IEEE Interim Meeting + Return loss @ Rx, Bring up informative means to measure it. Joel to present. + Review pages 1 – 9 + Pg. 6 – Where did 12Gbps come from? Worst case probable coding scheme. It is just to specify a max. or cap. Give us some kind of limit for analysis/testing and specification. + Pg. 7 – Graeme to include the inaccuracies for tightly coupled diff pairs in his presentation for IEEE meeting next week. + Pg 8-14 – Go through masks to develop proposal for next week and discuss group delay + John D’Ambrosia’s Presentation. + Devote an entire call to John’s slide #7. This slide addresses packaging effects to the channel. SDD21 Discussions: + -80dB levels out because you run into the noise floor of the machine. Location of the curve (y-direction) could be moved as John shows on Pg.5 & 6. Joel is comfortable on the current curves from the standpoint of TP1 & TP4. + Jeff Cain agrees with the curves, John D’Ambrosia qualified his data against the boards and materials that he got his data from. + Adam restated the objective of the group regarding material/connectors/distance. + Mike Altman would like to see a comparison of broad data measurements before we decide to use this line (equation). Joel stated that this was a starting point and there will be change to it before the final equations are set. + John wants to know how much of an improved FR4 the system guys are willing to use for their line cards. + What definition of improved FR4 is the group going with? We need to revisit this in a future call and close this issue. Joel is planning to present something on this…stay tuned. + Bill Hoppin speaks in support of the curves. + Petre would like to see them 4-5dB higher. This is the curve and mathematics for the first informative mask definition of SDD21 that will be presented to the group by Joel. TI would like a better curve 4-5dB higher. Joel feels that this would put us into hi performance material and not improved FR4. SDD11 and SDD22 Discussions: + Pg. 9 – should read SDD22 (not SDD21). We discussed the equations of CX-4 and noted that the current ones needed adjustments based on Zhi stating that the loss crosses zero at around 6Ghz. Zhi was going to address this and thought we could lower the equations by 5dB. Zhi will send me data to discuss this either at the meeting or at future calls. Discussions then continued on the current CX-4 equations. + Adam thinks that we should accept the equations but that we should show data to support them in the future. Jeff – good starting point. People have a good history with these equations. Brian Seeman – Steve Anderson – good starting point. + As an ad-hoc group; 54.7.2 and 54.7.3 will be added to our informative proposal. Zhi will send his data on the equations and we will adjust from there if required. NEXT and FEXT + Pg. 10 – Effect of multiple aggressors. Combining NEXT and FEXT into a total aggregate effect? Specify a limit on the net effect, not the individual contributors? Want to consider a combination of NEXT and FEXT as we go normative. Brian Seeman will create a proposal. + For 54.7.X on NEXT and FEXT, Joel would like to move forward with this for next week with Jeff Cain’s extrapolation (to 20G) for multiple disturbers and include that in the mask set. Discussion of whether or not this is a “good stake in the ground” to move forward with. + Individual straw poll – Is there any disagreement on using the NEXT & FEXT curves from CX4 extrapolated out to 20G if we lower them by 10dB? 18 members, 15 – yes, 1 – no, 2 - abstain. + Jeff will forward the shifted curves to Joel. Group Delay + Group delay discussion. To get phasing information in on the S21. + How important is phasing information to the silicon guys? Petre wants group delay ripple. + Does anyone have any objections to using this curve (Pg. 13) as a starting point? + Change the curve y-axis label and normalize it about zero. Zeros the average group delay value. + Are there any objections to this? No detractors. Joel will check with Jeff to get his technical opinion. We now have the start for the informative six masks that are our goal! Congrats!!!