[802.3ae_Serial] From Serial PMD 3 July: table numbers, model, overload, jitter, 850 rise time
Derivation and consistency of receiver stressed and unstressed sensitivity
--------------------------------------------------------------------------
Still not completely clear, will address this at face-to-face meeting on
Monday. In the meantime, Mike and Tom are asked to write down how nominal
and stressed sensitivity, link penalties and margin are calculated. Petar
was asked to email his derivation of the stressed sensitivity. Question of
whether the measure of stressed ISI now includes a standard Bessel filter.
Link budget
-----------
Piers was asked to prepare two spreadsheets: v2.4.1 with D3.1 numbers, and
new model with D3.1 numbers.
Receiver overload
-----------------
Mike and Piers hadn't had time to draw up and issue the linear programming
charts. Mike is basing his calculations on a 6 dB extinction ratio.
Juergen to submit a comment to flag the issue for comment resolution.
Jitter
------
Measuring CDR high frequency jitter tolerance is not easy. At least one CDR
intended for 10GE, and one SONET CDR, can't quite achieve the draft spec.
Further mention of whether the added SJ in the draft is double counting, and
whether it is worth relating SONET and Ethernet jitter methodologies:
perhaps the time for that has passed and we need to validate the Ethernet
methodology on its own terms.
I expect we will hear more about this in the feasibility meeting on Monday.
Rise time for 850 nm
--------------------
Mike's triple trade off curves are calculated with 35 ps "and a filter to
avoid exceeding 3.6 dB ISI". Mike and Petar to discuss off line, check with
Jack Jewell as necessary.
Next meetings
-------------
PMD feasibility, see Vipul's emails. Followed by:
ad hoc Face to face on Monday, probably Columbia room, downtown Portland
Marriott ~ 5 pm. See separate email (already sent) for proposed agenda.
Piers