RE: 3PMD Proposal
My Friend Bernd:
It seems you picked a popular and interesting topic; as a result, everyone
likes to write to you including myself.
I agree with you that, the 1000BASE-LX product does not really need an
offset patch code for the majority of installations. Typically, the
required minimum fiber BW is 70% to 80% of the bit rate, or around 900 MHz
for the GbE links at 1.25 Gbps data rate. The 1000BASE-LX specification
using 62.5 um MM fiber with 500 MHz-km BW at operating distance of 550 meter
is a near worst case design, which should satisfy all installed systems.
The reason of using an offset patch-code is to avoid the BW loss (but not to
increase BW) caused by the defected, abnormal, DMD fibers, which are very
small portion of the installed MM fibers. In fact, I told my users not to
use the offset patch code initially, but to use an offset patch code only
when a DMD fiber is identified by excessive BER.
The 10 GbE CWDM/WWDM using 500 MHz-km (850 nm, 50 um /1300 nm, 62.5 um)
operating at one quarter of 10 Gbps (2.5 Gbps/ 3.125 Gbps) is similar design
as 1000BASE-LX by extrapolating data rate and operating distance to achieve
the 300 meter operating distance. Therefore, in case of 1310 nm 62.5 um MM
fiber, it does not require an off-set patch code, except for a defected, DMD
fiber, which is a very small portion of the installed base.
I guess it is a good question why bother with all those defected fibers?
Why we are not removing them from the installed base?
For a 10 GbE serial link, using the 62.5 um, 500 MHz-km, fibers, the
operating distance can be extrapolated from 1000BASE-LX specification in the
same way the WWDM's 300 meter being derived. The extrapolated 10 GbE serial
link operating distance is about 70 meter, which was also indicated by Brian
Lemoff. Again, no offset patch code is needed, unless the fiber used is a
defected DMD fiber. I like to remind that an offset patch code is used to
avoid the bad section of the fiber cross-section to restore its normal
bandwidth, but the offset patch code cannot increase (or amplify) a fiber
bandwidth.
From the FO 2.2 task test data, it showed that those good 62.5 um fibers (no
DMD at all) using 1310 nm can have the upper limit BW nearing 2.GHz-km to
3.0 GHz-km. However, there are only very small portion of installed 62.5 um
fibers can reach those upper limit BW. Even if we can get BW of 2.0 GHz-km,
the 10 Gbps serial link operating distance is 300 meter according to the
serial link presentations in Ottawa. Furthermore, if we can get those 3.0
GHz-km BW fibers, the operating distance will be 500 meter at BER of 10^-12.
The problem is that they are a very small portion of the installed base.
You have one of the best BW fibers in you hand.
Regards,
Edward S. Chang
NetWorth Technologies, Inc.
EChang@NetWorthtech.com
Tel: (610)292-2870
Fax: (610)292-2872
David,
"min. overfilled launch" as in the Ottawa presentation from Del may imply
the use of a patch cord. At least myself (and maybe others) missed it.
Anyway, a patch cord is a solution that guarentees greater distance over old
fiber and that's good.
For our 5 distance objectives this could mean that a serial 1300nm solution
fulfills them all probably long term at lower cost then a CWDM just by
adding a patch cord were necessary.
In this matter I agree with Ed Cornejo that most (90%+) of the LX GBE
applications today work without patch cord. We tested serial 10 Gig on
random fiber over 500m without any problem!
Best wishes
Bernd