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

RE: [EFM] Active Architectures




Hello Geoff, though I have read accounts of the scenario you describe being
an issue (I believe it was NTT deployments?), and these are the types of
things not generally thought of by the engineering community, it is
definitely a solvable problem. Put a connector on one; keyed connectors on
both; keyed duplex connector; color or size keyed, or even the dreaded
"human inspector" that post-inspects the connections as a double check. I
wish this was the most significant issue we had to deal with:). 

-----Original Message-----
From: Geoff Thompson [mailto:gthompso@nortelnetworks.com]
Sent: Tuesday, August 21, 2001 11:01 AM
To: Horne, David M
Cc: 'Carlos Ribeiro'; stds-802-3-efm@ieee.org
Subject: RE: [EFM] Active Architectures


David-

I believe that you are overlooking the chief merit of the single fibre (P2P 
or P2mP) solution.

That is: The installer will have a much more difficult time getting it 
wrong if there is only one fiber.

Having the cable-end appearance have a multi-mode and a single mode would 
destroy this advantage and increase the average termination time to:
         1) Guess which fiber
         2) Terminate that fiber
         3) Test the termination
         4) Aw #$%%*!
         5) Got to step 1
    Divided by 2

Geoff

At 07:20 AM 8/21/01 -0700, Horne, David M wrote:

>Hello Carlos, on your comment: "I see one immediate problem. I
>assume that the 850 nm lasers are coupled to multimode fiber, and that the
>downstream PON needs to use a singlemode fiber. If this is correct, then
>every home would have to be served by two fibers of different construction.
>I've heard news of fibers able to carry both single and multimode signals
>but haven't seen any hard data on this."
>
>Yes, that is why I made the below comment in my original message:
>"There are a couple options on the fiber between the node and the ONU but
>I'll leave that discussion for later."
>
>I don't think I'd use the word "problem" to describe it since it is quite
>solvable. It's more a question of which method is used. I would advocate
the
>use of microtubing or conduit between the node and each ONU. That way, the
>fiber or fibers would only be installed (blown-in or pulled) when a
>subscriber signs up for service. Likewise, the ONU would only be installed
>at that time. It doesn't have to be done this way, but to do otherwise
would
>leave the network operator with stranded assets in the field. In some cases
>these assets could sit idle for a very long time. Another option is the
>customer could own both the fiber and the ONU and have them installed by a
>third-party when they subscribe to a service. The second strand is still
>quite a minimal cost compared to the transmitter cost savings that the
>architecture provides. In a neighborhood of 1/4 acre lot sizes, the average
>run for a cluster of 16 homes would be about 170 meters or so.
>
>The extra fiber (if 2 are used) shouldn't be viewed as an additional cost
>per-se. It is just a part of the architecture decision and the OVERALL
costs
>are what matter. It is easy to single out an architectural difference and
>concentrate on what that costs, as an argument against something, but in
>reality the big picture is what matters. Big-picture-wise, this apparent
>extra cost disappears.
>
>If a single fiber were used, I believe there are methods such that it could
>be either singlemode or multimode. Anyone who can provide cost and
>performance details of such an arrangement please do. I find very little
>published on this, except to imply feasibility. In any case I tend to think
>this would be more expensive overall than 2 separate fibers since mux and
>demux components are needed. It may be straightforward off-the-shelf
>components but I think those cost more than 170 meters of fiber.
>
>
>
>
>-----Original Message-----
>From: Carlos Ribeiro [mailto:cribeiro@mail.inet.com.br]
>Sent: Tuesday, August 21, 2001 4:16 AM
>To: Frank Coluccio; david.m.horne@intel.com
>Cc: sganguly@yahoo.com; ramu_raskan@angelfire.com;
>stds-802-3-efm@ieee.org
>Subject: Re: [EFM] Active Architectures
>
>
>(in the context of David Horne's "half-PON" idea)
>At 20:22 20/08/01 +0000, Frank Coluccio wrote:
> >For extremely long runs, have you considered the tradeoffs of an
> >inexpensive cwdm
> >solution in lieu of multiple strands coming back from the ONUs?
>
>That was my first thought when reading David's proposal. In fact I already
>discussed similar views not only with David, but with lots of people in
>this list. That's the main reasoning to develop a wavelength alocation plan
>that allows for the future use of C/DWDM. Gerry Pesavento is going to
>discuss this issue.
>
>As far as Mr. Horne proposal is concerned, I see one immediate problem. I
>assume that the 850 nm lasers are coupled to multimode fiber, and that the
>downstream PON needs to use a singlemode fiber. If this is correct, then
>every home would have to be served by two fibers of different construction.
>I've heard news of fibers able to carry both single and multimode signals
>but haven't seen any hard data on this.
>
>
>Carlos Ribeiro
>CTBC Telecom