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

How much singlemode fiber is high-PMD?




Dear colleagues,

Let's conceptually put all the singlemode fiber in two bins - old
and new. We define new fiber as the one that conforms to the new IEC
guideline of 0.5 ps/sqrt(km) max. We define old fiber as the one
that doesn't.

(The new fiber can be understood and dealt with. In fact, in
parallel to this thread, I will soon resume discussion of estimating
PMD and the outage probability for new fiber links. Our goal will be
to go to 802.3ae with a description of the problem and some
recommendations affecting the design of 1550 nm links.)

The unknown, which makes me more uncomfortable, is the old fiber. We
need to answer three questions.

1. What is the magic year when production of old fiber stopped and
production of new fiber began?

This is best answered by fiber manufacturers. A good source of data
may be when the patents for processes that control PMD were filed by
fiber manufacturers. Another source may be to look at IEC and ITU
activities and identify the year when PMD specifications became
topics of discussion.

2. How much of the old fiber is in the ground today?

We are only interested in terrestrial metro routes, not long
distance routes. We are more likely to find new fiber in metro
routes - true?

3. What can we do to identify these old fibers? (I am assuming a
worst case scenario where 802.3ae members insist on answering this
question, regardless of the answer to question 2.)

My initial thought was that we could just read the date code on a
cable, but a colleague reminded me that often the markings fade or
get scraped off...also, in a manhole, one may have just a few feet
of cable to work with, so date code may not be accessible. Well,
then, what instruments can we use? Surely, the Jones Matrix method
is an overkill and impractical!! How about Interferometric test
sets? They are portable, though not hand-held (or inexpensive).
Perhaps the potential market size will encourage test instruments
vendors to come up with a good solution. Any thoughts?

It's important to answer these three questions, even partially, by
November 5.

Thanks,
Vipul

vipul.bhatt@xxxxxxxxxxx
(408)542-4113