Channel Ad Hoc telecon minutes September 28, 2005 Adam - We plan on meeting early in Vancouver to go over comments. The channel ad hoc will meet Sunday afternoon and Monday morning then again with regular task force meetings on Tue, Wed, Thur. Have requests for presentation time in to Adam by midnight on 11/9. John - Can we reschedule the channel ad hoc from 10/26 to 11/2? Charles - approved, move channel ad hoc telecon to 11/2 Adam - Nashua did draft responses on Return Loss and voted to not revisit the comments. Haven't seen compelling arguments to include channel return loss. Gourgen - Do we have feedback from our Return Loss resolutions? Adam - Not yet. I'd like to see that issue closed. Charles - OK, channel return loss effort is closed until compelling arguments, and alternate remedies show up. Charles - Adam, should we cover ICR in this Ad Hoc? John - What's our focus? Adam - This group should focus on 69 and 69a issues. Charles - Today we'll cover comments 105, 128, 300. Is there any reason not to accept Joe's remedy for 300? John - Since he is asking for the same KR values for f2 and Nyquist, we need to talk to him to find out what he means. Charles - I think we should review equations in dambrosia_02_0905 John - For comment 128, propose to accept in principle that we review equations and make necessary corrections by Vancouver meeting. Adam - The goal here is to write responses in this meeting to be recommended in task force. I'd like to have the equations resolved prior to the meeting. Charles - I'll have someone review the equations for the next meeting....since I'm questioning them, I'll do it. Steve A. - With all this smoothing, what about spikes that cause failures? John - channels that failed ICR failed for other reasons not necessarily crosstalk, but other things like SDD21. Charles - If you have a narrow spike in ICR, the probability to have a channel fail is exponentially small as you get narrower. John D. - Ripple could cause a fail, while deviation would pass. The biggest issue I see is comparing this to BER 1E-12. Charles - Variation from sim to sim is comparable to the 10% in margin. John - if 2/3 passed-> PASS, If 2/3 fail-> FAIL Charles - Objection to review equations? If correct, then propose accept. Charles - Steve, do you object to us reviewing the comment 128 equations? Steve - a little bit, but I don't have anything to base it on. Charles - Ok, let's move on. moore_01_0905 slides 1-14 support comment 105 Rich M. - Is the crosstalk spec EIT? Charles - Yes, this is the EIT curve. Rich - If the crosstalk is less than the EIT, then the channel works. Charles - when you change the channel loss, the slope of the line changes. If you take an ICR item, the slope...I'm adding 4dB...we can get away with the same crosstalk at from low freq to nyquist. Rich - Can we just use ICR and add a line with a slope as a limit for crosstalk. It appears as a large change, can we just tweak ICR to get the same result? Charles - Using sine interference is valid with a correction factor. I'm assuming crosstalk is flat across frequency up to nyquist. John D. - Based on the channel models, the low freq crosstalk has more power, and none of the Tyco or Molex channels passed with the moore_01_0905 limits. Is there a way to look at the combined effects of the TX and the channel to get proposed model and annex 69a in sync? Rich - Charles is giving more margin, John says we have less? John - on page 10, is the ICR for the Goergen channel, to set the limit at -42db at Nyquist, it would fail. Fulvio - on page 5, the red curve is the total xt power. At high freq the red curve passes the limit. The model in Charles presentation... John - case 4 is 32in channel, case 5 is 29in channel, the limit changes all over the place. charles- if you look at the nyquist frequency, the difference between the two, the existing proposal use smoothed data, and this is unsmoothed data. The moore proposal needs smoothing on the PS curves instead of the ICR curves, to be a better comparison. The 3.4dB number is the assumption between sine and random. It may make sense to get rid of the margin, and only use smoothing. John - you want to smooth PSXT. Rich - you need more than a couple of dB on case 4 Charles - if you do a LSF on the data, you get less than 3.4dB worse than that line. John - I proposed LSF on channel and crosstalk. Rich - Is there a way to relate this to the ICR line. Charles - page 10 dambrosia_c1_0905, my curve on moore_01_0905 would be a flat line at 42dB. John - the current ICR is at 12.5dB. Charles - 12.5dB at nyquist, you get 38.5dB at nyquist and johns plot on page 10 has 40+, something doesn't look right with John's plot Rich - you are proposing changing ICR to 42dB as long as it's within the channel S21? Charles - Yes. Rich - We have ICR. what's the limit? Charles is suggesting move the limit with nyquist. Charles - the difference between my proposal and the current spec, is that i start the line lower, then draw it flat. The proposal is a rubber line, the spec is a rubber curve. Rich - let's get the previous limit with the new limit on the same curves to see the difference. Charles - how about a third curve on john's page 6 John - the PSXT.... Charles - It seems clear that John's presentation is confusing most of us. How about John checking his presentation, and let's continue this discussion in two weeks time.