IEEE802.3aq Channel model ad hoc Task 2

1
IEEE802.3aq Channel model ad hocTask 2 and 4
Launch study

• Teleconference notes, 9th Feb 2005
• Jonathan King

2
Agenda

• attendees
• Launch study discussion
• Correlation of OSL PIE-D vs OFL - John Ewen
• Correlation of OSL PIE-D vs OFL - John Abbott
• other business
• next meeting date/time if needed

3
1) Attendees

• Jonathan King, Big Bear Networks
• Petar Pepeljugoski, IBM
• Yu Sun, Optium
• Al Brunsting, Panduit
• Lew Aronson, Finisar
• John Ewen, JDSU
• Peter Hallemeier, Optium
• David Cunningham, Agilent
• Mike Dudek, Picolight
• John Abbott, Corning
• Paul Kolesar, Systimax

4
2) Launch study

• a) Correlation of OSL PIE-D vs OFL - John Ewen
• A graph of correlation of PIE-D for various
  offset OSL plotted against OFL bandwidth for the
  67YY Monte Carlo model was presented.
  Correlation was 0 for centre launches or OSL with
  offsets from 0 to 15um, while at 20um correlation
  was approx -0.65 there is a general trend that
  if fibre OFL Bw is high, the PIE-D is low for a
  20um OSL, whereas centre launch is unpredictable
  from OFL Bw.
• Discussion following
• John Abbott commented that the correlation plot
  was consistent with the conceptual model of
  fibres having large variability in the central
  region. Petar Pepeljugoski asked if fibre PIE-Ds
  correlated with measured path penalty when an EDC
  was used. David Cunningham said that his
  experience showed PIE-D correlated well with path
  penalty. Jonathan King confirmed similar
  findings.
• b) Correlation of OSL PIE-D vs OFL - John Abbott
• A scatter plot of PIE-D vs OFL bandwidth was
  shown for the 54YY and 67YY . Little correlation
  is seen between PIE-D and OFL for an 11um offset
  OSL, whereas a broad trend of higher PIE-D at
  lower OFL bandwidths is visible for a 20um OSL .
• Discussion following
• John Abbott said that his scatter plot did not
  include the effect of connection offsets he also
  said that data for 0um offset was not readily
  available, but that the correlation graph was
  flat for 0 to 15 um offset. John Ewen summarized
  that CL (centre launch) did not need to be
  tightly specified to get a good statistical
  chance of CL launch producing a lower PIE-D than
  the OSL launch David Cunningham referred to
  Ewen_1_0105, which shows that for a fibre with a
  90 percentile OSL PIE-D, there was a very high
  probability of the centre launch PIE-D being
  lower, largely because of the poor correlation
  between CL and OSL . This is what makes the dual
  launch approach so effective.
• Paul Kolesar commented that so far the task
  force has simulated centre launch as a single
  mode spot, but that the launch from a TOSA
  directly into a MMF patchcord may have a larger
  spot size and since the MPD differs, a different
  PIE-D statistic may be expected. There was
  general agreement that the current EF spec for CL
  is sufficiently tight to define a single mode
  spot-like launch, but that a looser EF spec could
  help reduce implementation costs, and that it
  would be useful to study the performance trade
  off likely with a progressivly looser EF spec for
  CL. The group concluded that it would be
  useful to postulate looser EF specs for CL and to
  simulate the impact on coverage. Petar
  Pepeljugoski agreed to initiate some preliminary
  work on this.
• NB Centre Launch CL Offset single-mode
  Launch OSL Mode conditioned launch MCP

5
3) Other business

• no issues raised

4) Next meeting

• Time slot available for next meeting Wednesday
  16th Feb 2005 9.30am.
• Call will proceed if there are submissions for
  discussion by the preceding Monday please send
  contributions via the reflector as usual, cc to
  Yu Sun