RE: Multi-mode PMDs
Ed/Mike,
I would add the 1310 Serial PMD as being even more attractive for VCSEL
implementation than WWDM. These will happen, but the timing is uncertain
until there are more definitive results. [Recent press announcements on
1300nm VCSELs show far less information than the 1170nm VCSEL data I
presented at the March 2000 Plenary.] The best way to proceed is to draft a
spec that works for DFB's and has flexibility to accommodate a VCSEL
solution. This is present, in the form of asterisk's and a Note, on slides
11 and 12 of Del Hanson's presentation at the May 2000 Interim. So the
approved 1310nm Serial PMD is in good shape for short and long term, though
I would still argue for about 1dB greater receive sensitivity in exchange
for 1dB lower minimum launch power.
Now you might expect a longwave VCSEL promoter like me to push longwave
VCSELs for all solutions. Not the case however. Use in MMF is more
complicated than it might look. First of all, there is no 802.3z spec for
MMF at 1310nm which has a bandwidth greater than 500MHz-km. Some fibers are
only 400MHz-km at 1310nm. So a straightforward reach 100m on MMF is no
easier at 1310nm than at 850nm. If it did, such solutions would have been
proposed and probably well-received. {I think one involving coding may have
been presented at the November meeting.) Additionally, my understanding is
that the mode-conditioning patch cord is required to achieve the 500MHz-km
on FDDI-grade fiber. The DMD effects in MMF are very strong at 1310nm, thus
necessitating restricted launch. Now at 850nm, we already have the VCSELs,
and restricted launch conditions appropriate for 850nm VCSELs are defined in
the TIA FO2.2 for achieving 385MHz-km over FDDI-grade fiber. Not much
different from the 400MHz-km minimum for patchcord-launched 1310nm.
Additionally, new MMF allows 300m serial links at 850nm. It is compatible
with 300m 1310nm WWDM links without the patchcord, but it will not support
100m serial links at 1310nm.
Hence my consistent support for the 850nm VCSEL-based PMD for short-reach
MMF links. Technical demonstrations are in relative abundance by multiple
vendors. [This week at NFOEC, a 10Gb/s 850nm VCSEL-based full module demo
is on the Exhibit floor - I won't mention any company names.] Serial
solutions at all wavelengths maximize the use of Si (the most cost-effective
technology known) and minimize the complexity of the optics. For MMF, the
850nm serial PMD and the 1310nm WWDM PMD form an optimized pair to satisfy
the MMF needs.
Regards,
Jack
-----Original Message-----
From: Edward Chang [mailto:edward.chang@xxxxxxxxxxxxxxxx]
Sent: Tuesday, August 29, 2000 7:13 AM
To: Mike Bennett
Cc: Roy Bynum; hssg
Subject: RE: Multi-mode PMDs
Mike:
The application of 1310 VCSEL is primarily in the multimode utilization of
1310 WWDM to achieve 300 meter. It is not enough for serial application to
achieve 100 meter --- using BW of 500 MHz-km.
The main question is the 1310 VCSEL volume production schedule and price,
which the 850 VCSEL has well established for a long time without any doubt.
For 1310 VCSEL, it is an expectation, which needs time to prove.
Regards,
Edward S. Chang
NetWorth Technologies, Inc.
EChang@xxxxxxxxxxxxxxxx
Tel: (610)292-2870
Fax: (610)292-2872
-----Original Message-----
From: Mike Bennett [mailto:MJBennett@xxxxxxx]
Sent: Tuesday, August 29, 2000 7:26 PM
To: Edward Chang
Cc: Roy Bynum; hssg
Subject: Re: Multi-mode PMDs
Ed, Roy, et. al,
I sent mail to the reflector (subject 1300 nm VCSELs, date:Wed, 16 Aug 2000)
in
order to get the discussion going on short reach PMDs. I hope folks are
really
busy and just haven't had the time to discuss the issue here on the
reflector.
I'd like to hear your opinions regarding the 1300 nm VCSEL and its potential
impact on the low cost short reach application space. I'm of the mind that
fewer PMDs is better, but I was not willing to vote in favor of fewer PMDs
without being assured of a low cost option for the short reach application
space. If 1300 nm VCSELs are going to be available in a reasonable time
(less
than 1 year), and the mode conditioner is not needed for <= 100m then it
seems
much more difficult to argue in favor of more PMDs. I'd much rather see the
discussion start now, rather than burn time at the interim with the same old
arguments ending in up with no concensus.
Regards,
Mike
Edward Chang wrote:
> Roy and All:
>
> It seems that we did not get the consensus of which specific PMDs we need
in
> addition to 1550 and 1310 serial. However, it seems we have the
consensus,
> or the majority of desire, to have short reach, cost-effective PMDs, which
> will include the VCSEL based approaches.
>
> Since many factors we have been debating to further narrow down the
> selection lists are based on the future market trends and many technical
> assumptions, we have had quite verities of different opinions; as a
result,
> we were unable to obtain a consensus. I do not think the situation will
get
> better, unless we take a lot of time in debating and testing by prolonging
> the schedule, which is unlikely the path we can afford to take.
>
> I think we should let all PMDs on the table to be adopted, and let us move
> on to the next target, instead of trying to pin down factors which are not
> under our control.
>
> Regards,
>
> Edward S. Chang
> NetWorth Technologies, Inc.
> EChang@xxxxxxxxxxxxxxxx
> Tel: (610)292-2870
> Fax: (610)292-2872
>
> Subject: Multi-mode PMDs
>
> All,
>
> Where are we in the discussion of the multi-mode PMDs? There has not been
> much on the reflector of late. The last that I have on this does not
> indicate to me that there has been any consensus established. The last
> that I have, indicates that the system vendors and other members of the
> 10GEA are still wanting the 1310nm solutions, and the rest of the TF are
> wanting the 850nm solutions.
>
> Has this subject been moved to an Ad Hoc? Has this discussion been moved
> to a different forum/TF?
>
> Thank you,
> Roy Bynum