[BP] Signaling ad-hoc conference call minutes 14-01-05

Thread Links	Date Links
Thread Prev	Thread Next	Thread Index	Date Prev	Date Next	Date Index

Mike Altmann starts with am overview of the meeting agenda (re. http://www.ieee802.org/3/ap/public/signal_adhoc/altmann_s1_0105.pdf , slide #2). Mike continues with an overview of progress vs. original schedule. Mike asks for feedback:

[Joe A.] I am going to target to have some sims for the meeting. I am struggling to get it complete. I will not be able to have it complete to the level of detail indicated by the spreadsheet.

[Shannon] I am planning to publish some StatEye results.

[Amir] We will bring some results

[Grant] ... some simulation results, although it may not be complete.

[Petri] ... also some sims.

[Mike A.] I am planning to have some PR4 results.

[Adam H.] Not sure about PAM4.

[Mike A.] The 19^th is the date we had defined to “post” simulation results. By the 19^th it must be communicated to Adam that a presentation is requested.

[Mike A.] Moving on, I would like to review the changes to the signaling spreadsheet (which should say Rev. 4.2 and not 4.1 as it presently does).

Mike reviews the changes.

[Mike A.] Are there questions/comments?

[Charles] I though we had discussed scaling crosstalk till failure point.

[Adam] The results of the straw poll was 9-6-6

[Mike A.] My interpretation was that there was some significant opposition. Also it would not be clear on how to report that.

[Brian] How about reporting the magnitude of the crosstalk? Charles, what was your intent?

[Charles] My intent was to report a numerical measure of the robustness of a given solution.

[Mike A.] We are reporting voltage margins already. Are we not covering this? When we talked about scaling crosstalk we concluded that it would be different from channel to channel.

[Charles ] We are already reporting timing/voltage margins on a channel by channel basis. This would be just another output to report.

[Mike A.] Aside from NEXT/FEXT scaling, is there anything else?

… silence …

[Mike A.] Since we have discussed this before, and we have done so again, maybe we should re-issue the straw poll from November regarding the scaling of the crosstalk.

STRAW POLL #1: Should we linearly scale NEXT/FEXT until a given solution breaks (fails to meet the BER requirements) and report the result?

[Joe A.] We had a lot of discussions already and I do not feel it’s necessary to overturn it now. Besides, this is an aspect that I will not be able to cover for this coming meeting.

[Charles] I would accept as a friendly amendment the fact that this will not be done by January but, rather, represent a long term goal.

[Mike A.] Let me try this by acclamation first. Is there anybody that would vote “NO”?

… discussion follows .. straw poll called : YES [4], NO[5], Abstain[14]

[Mike A.] I am inclined to interpret these results as mixed (mostly as “don’t care”) consequently I am going to leave this item off the spreadsheet.

At this point it was observe that the MOLEX channels were not included in the Signaling Spreadsheet (SS) and Mike agreed to include them in Rev. 4.3

Mike resumes going to his presentation, focusing on slide #6, LINK elements.

[Mike A] What does the group think we should use for the cap or TP4-TP5?

[Charles] The 4.7 nF is in the range. The SP4 mostly relates to the measurements of the traces to the cap.

[Shannon] Yes, this is a very specific data point which, structurally, may not be representative of what we are going to have between TP4-TP5.

[Mike A.] Well, the choices are : either we use this, nothing or everyone does his thing. Last time, a straw poll indicated split in the group, Hence this attempt to clarification.

[Shannon] I am planning to use Rich’s package and this cap.

[Mike A.] Is there a feeling that his is not adding any benefits to the simulation results? If so, we should not bother.

[Shannon] The reason is to include a reasonable estimate of the parasitics and the effects that they may have on the system.

[Charles] I personally feel that the inclusion of the model is going to cause more trouble than benefit. Therefore I would like to propose to make it optional.

[Mike A.] Again, if it does not add much why bother at all?

[Charles] Because Shannon wants to ..

Straw Poll #2: Should we use a model for TP4-TP5?

Yes [10], No[5], Abstain [5]

Straw Poll #3: Should we use Shannon’s 4.7 nF model for simulation work to be presented to January meeting?

Yes[5], No[8], Abstain[7]

[Mike A.] Since the responses to SP3 were mostly “NO” or “Abstain”, it may be valuable to re-ask should we use at all a model for TP4-Tp5 for sim results to be reported in January?

Straw Poll #3a: should we use at all a model for TP4-Tp5 for sim results to be reported in January?

“NO” by acclamation

[Mike A.] Similarly for packaging: should sims for Interim meeting in January use a package model?

Straw Poll #3b: should sims for Interim meeting in January use a package model?

By acclamation, 21 decided to use Mellitz package model

[Adam H.] Let’s adopt a policy of reporting what we use as simulation input parameters.

[Mike A.] Excellent suggestion. I will updated and resend out the SS.

Conference Call adjourned.