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

[10GMMF] Proposed resolutions to comments on noise specification

Nick asked me to review and reconcile the noise specification comments 47, 48, 92-95 and 97-99.

Here are my proposed responses.  There may be a little more to be added, but it provides a basis to improve.

47      PROPOSED ACCEPT IN PRINCIPLE   Represent both quantities in both units per responses to comment 48.  Include equations to convert per responses to comments 92 and 97.

48      PROPOSED ACCEPT IN PRINCIPLE   Add the alternative metrics in brackets (29 in table 68-3, -120 dB/Hz in table 68-4).  Following common practice, keep with dB/Hz as primary units for RIN and linear units for noise calibration.

92      PROPOSED ACCEPT IN PRINCIPLE   Don't say 'Conformance shall be determined'.  The conformance is what is required, not the determination of it.  The proposed equation can be simplified from
OMA/(2*rms noise) = 1/SQRT(10^((RIN+10*LOG(NBW))/10)) to
OMA/(2*rms noise) = SQRT((10^(-RIN/10))/BW) .  Use BW not NBW following precedent in other clauses.  Other proposals are addressed in comments 93-95 and 97-99, 101, 102.

        1.      Eliminate the first note.
        2.      We have to choose a primary method: scope or noise meter?  Simplest and safest to use the established noise meter way.  Change wording of 'Conformance shall be determined':  the conformance is what is required, not the determination of it.
        3.      The editors note and the comment part 3 are mistaken: clearly says 'using the pattern specified for the PMD type'.  Remove the editor's note around line 20.
        4.      Use the example screen shot from the submitted figure, modified to show a histogram on 0 too.  Show an optoelectronic scope in place of several boxes.  Show the back reflection as in figure 58-4 or P802.3aqD1.0com99weiner68.6.4.pdf.  Use 'System under test' in place of 'Device under test'.
        5.      Leave this subclause in the section for parameters common to transmitter and receiver.  If wished, add a subclause to group up 68.6.1-4 (or move third level subclauses in 68.6.5 and 68.6.6 up one level).


95    PROPOSED ACCEPT IN PRINCIPLE   Do not accept 'as wide as possible' as it is both vague and not achievable at finite cost.  Change to '...1 UI wide histogram where the mean waveform has negligible slope or other vertical variations relative to the noise being measured.'  For editorial comment in last line: see response to comment 96.

97    PROPOSED ACCEPT IN PRINCIPLE   Accept comment but change last sentence to 'In this case, BW is close to 7.5 x10^9 Hz.'  Make clear that rms noise is mean of noise in 1s and noise in 0s.


        1.      Have the noise meter method as the primary method
        2.      Instead of Ratio of difference between (mean 1 and 0 levels) and
(mean of the two standard deviations), use Ratio of difference between (mean 1 and 0 levels) and (SUM of the two standard deviations).  This is more like the definition of Q used in ITU-T and the Ethernet link model.  Include equations to translate each way between the two units, per other comments.
        3,4     See response to 93, 101.  Include equation similar to P802.3aqD1.0com99weiner68.6.4.pdf p1 lines 39-47, without the '0.5 x'.