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

[802.3ae_Serial] Possible simplifications of triple trade off tables



Serial PMD-ers,

A by-product of the improvements in the serial optics test methods made at
the last meeting is that for 10GBASE-S and 10GBASE-L we have both OMA and
TDP limits derived from triple trade off curves.  Particularly for
10GBASE-L, this seems over complicated.  There are several ways to simplify
the situation and I have tried to illustrate them in the attached.  Bear in
mind that for 10GBASE-L, the difference in fiber attenuation from one end of
the waveband to the other is about 0.4 dB; for 10GBASE-S it is negligible.
The table entries are divided up into areas with the same spec value, as
Dave Kabal suggested.

Option 1 is what we have in D4.1.  It's over-complicated, and allows full
freedom in laser spectral properties but no freedom in dynamic performance.

Option 2 is much simpler, allows limited trades between spectral and dynamic
performance, aims for consistent signal quality.

Option 3 is also much simpler, presents a very simple face to the user with
an OMA meter, trades  some signal quality for predicted attenuation.

Option 4 (Specify (OMA-TDP) )is what we have for 10GBASE-E: allows the
implementer to trade signal quality and transmitted power over a wide range.

Option 5 is a triple trade off like option 1, but with the range of
parameter values truncated.

Of course there is an
option 6: don't have a triple trade off at all, and
option 7, the two-categories proposal I put in the comments database last
time,
and there could be more options which we haven't thought of.  Are there any
more to be added to the discussion?

Options 2 and 4 seem the most plausible.  Comments welcome, and we can
discuss on the teleconference tomorrow.

Would we make the same choice for 10GBASE-S, or different?  Why?

Piers

OMA_TTOv01a.pdf