Re: Supporting Installed Fibre
PAM5 signaling at 5 GBaud is capable of driving ANY installed multimode fiber,
with the possible exception of DMD=broken fibers, a distance of approximately 200
m. This is just 20 m short of the GbE 220 m distance number, but the effective
data rate is 10 times the GbE data rate. According to Alan Flatman's numbers in a
recent note, this covers anywhere from 58% to 74% of installed building backbone
links. Given these numbers, I don't understand your comment "unless it is limited
to shorter links".
Multilevel signaling is independent of electo-optics and gets rid of far more
than just optical muxes. It also gets rid of skew, multiple lasers, multiple
photodetectors, multiple preamps or TIA's multiple postamps, multi-channel
reliability concerns, cooling requirements for multiple lasers, 4X difficulty in
meeting laser safety requirements, optical crosstalk, etc.
It seems the multilevel signaling alone, without WWDM, presents a pretty darn
simple and cost effective solution for the installed base as well as new fiber.
On the other hand solutions which mix WWDM and multilevel signaling are the most
complex solutions proposed to date and still do not cover all of the installed
base of MMF.
Jaime Kardontchik wrote:
> In order to support the installed fiber the baud rate at the optical
> fiber must be of the order of the bandwidth of the fiber. The latest
> proposal of Transcendata in Montreal uses a baud rate of 5 Gbaud/sec.
> Hence, although it gets rid of the 4-WDM optical muxes it would
> appear to be more difficult to use with the installed fiber (unless
> it is limited to shorter links).
> All the proposals that want to maximize the reuse of the installed
> optical fiber use in one way or another WWDM optical muxes. These
> proposals presently are (I apologize if I forgot one):
> 1) PAM-5 + scrambling at 1.25 Gbaud/sec using 4-WDM
> 2) PAM-2 + scrambling at 2.5 Gbaud/sec using 4-WDM
> 3) PAM-2 + 8b/10b at 3.125 Gbaud/sec using 4-WDM
> I am not an expert in pricing curves, but it seems to me that the most
> expensive item in the link is installing new fiber.
> Jaime E. Kardontchik
> Micro Linear
> San Jose, CA 95131
> email :kardontchik.jaime@xxxxxxxxxxx
> jay.hoge@xxxxxxxxxxxxxx wrote:
> > The alternative to WWDM is probably a multi-level coding scheme (dibits or
> > quadbits per baud), which will reduce the baud rate to a managable
> > frequency. Transcendata is working on such a system. Others may be looking
> > at alternative coding schemes as well.
> > The disadvantage of this approach is that the IC's become very complex.
> > Reletive to WWDM, this is also the charm, as the IC's will be subject to
> > Moore's law, while the passive optics, plus 4X the number of devices, IC's
> > etc. of WWDM will probably not.
> > It seems unlikely to me that the H-P WWDM will conform to the pricing curve
> > your customers expect. I think waiting a little while before we make a
> > comittment might yield a scheme which better meets your customers
> > expectations.
Richard Taborek Sr. 1441 Walnut Dr. Campbell, CA 95008 USA
Tel: 408-370-9233 Cell: 408-832-3957 Fax: 408-374-3645