RE: 3PMD Proposal is not enough
Tuvia,
Thanks for adding the information on micro-bend impact at 1300 nm, and sorry
for the delayed response (I've been out of the office all week). You are
correct that the attenuation specification at 1300 nm was harmonized for
both 62.5 and 50 um MMF types at 1.5 dB/km in the ISO/IEC 11801 draft
revision. The value has always been 1.5 dB/km in the TIA 568 standard. Note
also that the attenuation specification at 850 nm was lowered to 3.5 dB/km
from 3.75 dB/km for both fiber types in the newly published TIA 568-B
standard. Now both the 568 and the 11801 standards will agree on the
attenuation specifications for both 62.5 and 50 um. During the discussion in
ISO/IEC it was shown that the 1.5 dB/km specification would not impact the
supportable distance for 1300-nm applications.
Regards,
Paul Kolesar
----------
From: Tuvia Liberman [SMTP:teldorfo@xxxxxxxxxx]
Sent: Sunday, June 25, 2000 2:26 AM
To: Kolesar, Paul F (Paul)
Cc: 'Gair Brown'; 'stds-802-3-hssg@xxxxxxxx'
Subject: Re: 3PMD Proposal is not enough
paul,
just to complete the picture regarding the microbend sensitivity of
the 50 micron fiber as compared to the 62.5, this issue was raised and
presented in the ISO/IEC JTC1 SC25 WG3, the body drafting the next version
of the 11801, the international equivalent of the TIA/EIA 568A (or soon to
be published 568B) in berlin last year.
based upon liaison letters from the US delegation, the Israeli
delegation and the IEEE (!) it was agreed across the board that there still
remains a genuine difficulty for most fiberoptic cablers around the world
using tight buffered constructions (preferred for LANs) to consistently meet
a max. attenuation of 1.0 dB/km as was previously specified in the
international premises wiring standard.
since it had been agreed by consensus that both the american and
international standards present unified multimode requirements (50 & 62.5 um
without differentiation) the microbend sensitivity of the 50um fiber in
tight buffered constructions (generally considered to be about three times
that of the 62.5um fiber) was the reason that the international body
"loosened" its max. attenuation @ 1300nm from 1.0dB/km to 1.5 dB/km (again
with IEEE backing).
so this is still an issue, and one of the reasons that fiber
manufacturers are still seriously looking into second window (1300nm)
optimized 62.5/125um fibers.
regards,
tuvia
"Kolesar, Paul F (Paul)" wrote:
Gair,
Your perceptions are fair and I will attempt to respond to
each. I know you
are intimately familiar with these issues, so much of this
explanation is
for the benefit of others who may not be.
The US market has been a 62.5 um market. True. There has
been reluctance to
switching to 50 um in the US up to now, even though 50 um
offered extended
distance capability up to 550 m for GbE. My view is that the
added advantage
of the extra distance was insufficient to persuade the
customers to switch.
We know that 300 m is a critical "magic number" for the
great majority of
building backbones. Neither the installed 62.5 or 50 um
fibers can fulfil
this need for serial transmission at 10 G. The next
generation of 50 um
fiber does fill this need. Since this is a critical coverage
requirement for
customers planning on 10G networks, customers are willing
to switch core
sizes to gain this advantage. Our sales force confirms this
with reports
from their contacts with customers.
50 um fiber does not couple as much power from LEDs as 62.5
um. We account
for this in the applications and structured cabling
standards. This loss of
power translates into smaller budgets which reduce the
number of connection
supportable, or in some cases reduce the link distance. For
example, short
wavelength (850 nm) LED-based applications such as
10BASE-FL, when operating
at minimally-compliant levels (worst case power budget),
will typically not
have sufficient power to support 2km on 50 um fiber.
Depending on the
connections, the worst case supportable distance is usually
between 1 and
1.5 km. But long wavelength (1300) LED-based applications
like 100BASE-FX
easily still achieve 2km with margin. This is due to the
lower attenuation
of the fiber at 1300. There is no penalty for VCSEL-based
applications like
1000BASE-SX, since coupling efficiency is not an issue for
lasers.
50 um fiber is more bend sensitive, both to macro-bends
(loops in the cable)
and micro-bends (microscopic deflections due to fiber buffer
and coatings).
The macro-bend sensitivity causes some small amount of power
loss from
highest order modes excited by LEDs. 50 um macro-bending
losses are
generally on the order of 0.5 dB higher than 62.5 um fiber
for overfilled
(LED) launches. Macro-bending is not an issue for laser
launches. Since the
next generation 50 um fiber is aimed at 300 m link lengths,
this small
additional loss is inconsequential to the support of
LED-based systems,
which are virtually all specified to support much longer
lengths on 50 um
fiber. There is more than adequate power budget to cover
this loss.
Micro-bending loss, which is induced by the cabling process,
was more of an
issue for earlier cabling processes (1980s). Today major
cablers around the
globe produce 50 um cables without problems. The loss
specification in the
structure cabling standards specifies cabled (as opposed to
un-cabled) fiber
loss. Therefore, compliance to the structured cabling
standard specs assures
product that meets the specifications of the applications
standards. The
IEEE link model uses the worst case values in calculating
link distance
capability. So whatever micro-bending loss exists, it is
accounted for. As a
testament to the minimal impact for cabling 50 um relative
to 62.5 um, the
specs for cable attenuation at 850 nm for both 50 and 62.5
um fiber are the
same in the structure cabling standards. Both are specified
at 3.5 dB/km.
We at Lucent looked hard at all of these issues before
deciding that 50 um
offered the best trade-off of all parameters and market
acceptance for next
generation MMF. Our customers both in the US and abroad are
telling us that
we made the right choice.
Regards,
Paul Kolesar