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

Re: [10GMMF] TP3: Suggested stressor levels collected responses

Jim et al.,

We have reviewed the responses to your request for suggested TP3
stressor levels and would like to summarize Corning's position as
verbally outlined at the Atlanta meeting. Attached is a single Slide
denoting a triangle of design tradeoffs as we see it for the LRM

The 3 corners of the triangle are:

1. Worst case design philosophy - this has been the hallmark of most
IEEE standards, certainly all of the optical standards that we have been
a part of in recent years. It means that when all specified link
parameters are at their worst case values, the link should operate. In
some cases, this worst case has been relaxed to 99th percentile values
so we often talk about "coverage" in terms of how much of the installed
base would be supported by a given set of specified link values.

2. EDC chip design capability/complexity - we use PIE-D here because
PIE-D is the metric that 802.3aq is using to quantify how hard it is to
equalize the distorted pulse using electronic dispersion compensation.
It depends on the specific fiber (the mode delays), the specific launch
assumptions (i.e. the mode power distribution - offset launch, center
launch, etc.) the length, as well as assumptions about residual noise.
The current proposed value is 4.5 dB.

3. Quantitative requirements - we use length = L.  The current proposed
value is 300m.

The tradeoffs are interdependent and we agreed in Atlanta that we cannot
support all three as specified (worst case design, PIE-D = 4.5 dB and L
= 300m) - something must "give." 

Corning's position is that either (1) and (3) are fixed requirements and
(2) should be specified to support 300m, or (1) and (2) are fixed
requirements and (3) should be de-rated to support the specified

The other known position is that (2) and (3) are fixed requirements and
(1) should be de-rated to support the specified complexity and link

Underlying all of this is an unprecedented use of a dual launch which
has not been modeled accurately, specifications that leave little, if
any margin, and as respondent D notes "numerous uncertainties in
modeling and budgeting involving center launch, connectors, performance
of real laser transmitters, lack of OM2 models, measurement methods,

Corning has a vested interest in the success of all applications using
optical fiber.  Based on experience, we expect 802.3 will encourage
consistency and will agree that 802.3aq should use the same design
philosophies used in 802.3ae for all the 10G PMD types (10GBASE-SR,
10GBASE-LR, 10GBASE-ER and 10GBASE-LX-4); to do otherwise will subject
our customers (and their customers) to increased risk of failing links
and negatively impact the market potential of LRM to support the
installed base of multimode fiber. Any field failures will reflect
negatively not only on IEEE but also negatively on optical PMDs in
general and multimode PMDs specifically.
 <<EDC Tradeoffs.ppt>> 

EDC Tradeoffs.ppt