RE: Connector Return Loss (1550 nm versus 850/1300 nm)
Hi,
Well, folks, just for a historical note here (as best I can recall)...
Bellcore's (now Telecordia) reason for 40 dB are a few fold and the reasons
to retain it may be different from why it was put there to begin with, but
when the requirment was made 1550 nm wasn't such a big thing and definently
was not any sort of deciding factor.
The reasons at the time included...
1) A desire to drive the connector world to produce connectors which had a
physical contact. If anyone remembers the Biconic connectors they will
understand why. Also it was not fully understood what the proper connector
specification should be. This is also one of the reasons that the ST tended
to be eschewed by many of the Baby Bells.
2) At one point it was thought the analog optical transmission (as is (was?)
used in the cable industry) might be an important market and that this
market would require a much better return loss specification than the
digital world. (Of course it wasn't clear that 40 dB would be sufficient
for this application...)
3) Finally it was not fully understood what the digital world would need as
they went to higher bit rates. Especially, as someone else noted, if one
fiber would be used in the future for Tx and Rx at the same wavelength.
One could ask so why was not issue separate requirements for different
applications (say by analog vs. digital, different levels of the digital
hierarchy, etc.). The reason for this is that Bellcore in those days was
writing for its owners and NOT the industry suppliers. It was thought by
Bellcore that the owners wanted to be able to use one connector ubiquitously
throughout their network. Historically not all the baby bells agreed, but
that is another matter.
Either way, due to the desire to use one ubiquitous connector one return
loss requirement spec was given in GR-326, while the return loss spec given
in the Sonet documents were different.
At any rate I wouldn't get to hung up over the 40 dB spec in GR-326, but
would rather say the SONET spec may be of more interest since for 802.3 we
are interested in what will work for 802.3ae, i.e., we are interested in
what will work for 10 Gbps. This I imagine would be similar to the needs of
OC-192 (10 Gig SONET), maybe a bit more stringent since it appears that the
extinction ratio for 802.3ae is going to be less than OC-192.
Regards,
Joshua Brickel
> -----Original Message-----
> From: gwinn@xxxxxxxxxxx [mailto:gwinn@xxxxxxxxxxx]
> Sent: Thursday, April 12, 2001 5:27 PM
> To: Kolesar, Paul F (Paul); 'Geoff Thompson'
> Cc: 802.3ae
> Subject: RE: Connector Return Loss (1550 nm versus 850/1300 nm)
>
>
>
> Paul and Goeff,
>
> The Telecordia 40-dB return loss requirement may be intended for
> 1550-nanometer systems. There was a discussion of these issues at the
> Tampa 10GbE meeting in 2000, mostly revolving around what the receiver
> return loss should be. Basically, 850 and 1300 nanometer
> links are short
> (1 Km or less) and are build by connecting randomly-chosen
> parts from bins,
> while 1550 nanometer links are long (tens of Km) and are individually
> engineered. Use of optical isolators was discussed as well,
> but this is a
> transmitter design issue, and thus should not be required. The final
> agreement was to leave the connector return loss specs as
> they were, but
> for 1550 only tighten the receiver loss, and the RIN specs
> and measurement
> would be how the transmitter was specified. A tranmitter
> with an overly
> sensitive laser might require an isolater, but this is not
> for the standard
> to say. If memory serves, the return loss spec was changed
> from 12 dB to
> 20 or 26 dB; the current draft should tell the story.
>
> Anyway, we may wish to use Telecordia optical specs across
> the board for
> 1550 nanometer links, but 1300 and 850 nm links are a different issue.
>
> Joe
>
>
> At 10:24 AM 1/4/12, Kolesar, Paul F (Paul) wrote:
> >Geoff,
> >Good point on passive splitter systems. Similar concerns exist for
> >bi-directional transmission on the same fiber at the same
> wavelength. For
> >EFM, both of these systems will probably be only on
> singlemode fiber and use
> >singlemode connectors. While outside the scope of Structured Cabling
> >Standards, the improved return loss performance that may be
> required for SM
> >connectors is found elsewhere. For example, Telcordia now
> requires at least
> >40 dB return loss for singlemode connectors. Fortunately,
> products exist
> >that easily meet this performance level.
> >Paul
> >
> > ----------
> > From: Geoff Thompson [SMTP:gthompso@xxxxxxxxxxxxxxxxxx]
> > Sent: Wednesday, April 11, 2001 6:22 PM
> > To: Kolesar, Paul F (Paul)
> > Cc: 802.3ae; 'Doug.Coleman@xxxxxxxxxxx'
> > Subject: RE: Connector Return Loss
> >
> >
> > Paul-
> >
> > The only thing that I would add is that there is a
> 25 dB return loss
> >
> > requirement for MM in 10BASE-F (15.3.2.2) for
> 10BASE-FP (Passive
> >Star). As
> > we start to consider both passive splitter systems
> and single fiber
> >systems
> > in EFM return loss may again rear its ugly head.
> >
> > Geoff
> >
> > At 11:02 AM 4/11/01 -0400, Kolesar, Paul F (Paul) wrote:
> >
> > >Doug,
> > >Here are my views:
> > >The 20 dB level for MM connectors is historical in
> that that has
> >been the
> > >spec used by previous generations of application
> standards and is
> >in
> > >agreement with cabling standards.
> > >
> > >The 26 dB level for SM connectors initiated in
> Fibre Channel in the
> >early to
> > >mid nineties. I believe this was at the request of 1300 nm
> >transceiver
> > >vendors. It is also in agreement with cabling
> standards. It is
> >interesting
> > >to not that Telcordia (formerly Bellcore) used 27
> dB in the past,
> >which I
> > >believe was in concert with SONET equipment standards.
> > >
> > >The use of isolators in the transceiver is a design
> choice and not
> >dictated
> > >by IEEE 802.3. The receiver is specified with a
> minimum return loss
> >level,
> > >the value of which has been the subject of debate.
> > >
> > >What type of testing are you referring to?
> Transceiver or cabling?
> > >
> > >Paul Kolesar
> > >Lucent Technologies
> > >
> > > ----------
> > > From: Doug.Coleman@xxxxxxxxxxx
> >[SMTP:Doug.Coleman@xxxxxxxxxxx]
> > > Sent: Wednesday, April 11, 2001 9:23 AM
> > > To: 802.3ae
> > > Subject: Connector Return Loss
> > >
> > >
> > > Does anyone know the history behind the
> connector return
> >loss
> > >numbers
> > > proposed for 10 GbE?
> > >
> > > The return loss for multimode connections
> shall be greater
> >than 20
> > >dB.
> > > The return loss for single mode
> connections shall be
> >greater than 26
> > >dB.
> > >
> > > Do the transceivers integrate isolators
> into them to
> >address return
> > >loss?
> > >
> > > Has any testing be done to demonstrate
> compatibility?
> > >
> > > Thanks
> > >
> > > Doug Coleman
> > > Corning Cable Systems
> > >
>
>