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

Re: [10GMMF] Task 2 and 4 slides for meeting Dec15th - OM3 fibr e measurements

Jonathan, thanks for the comprehensive measurements.

1. Jonathan's use of green-yellow-red is exactly the breakdown that I was trying to achieve in my motion at San Antonio where we would tentatively designate one value of PIE-D as a pass (i.e. GREEN) and another value of PIE-D as a fail (i.e. RED) with a gap in between (YELLOW).  I think he gets the point across much better than I did.
        Here Jonathan is proposing unanimous pass = 4.0dB (GREEN)
                                         unanimous fail = 5.5dB (RED)
   At the San Antonio meeting discussion on this 3-level sorting
        we received input that a PIE-D over 5.0 was unacceptable.
        Jonathan puts that in the yellow category.

2. In doing experiments on the various demo cables the 802.3aq task
        force needs to remember that for the most part these are
        not worst case fibers, but rather typical fibers.  In the case
         of the OM3 demo cables Jonathan tested, one cable was supposed
         to be "interesting" with DMD artifacts to test our
        specification-setting, and one cable was supposed to be
        "good" with all fibers expected to work at 10GbE 300m 850nm.
        It is not obvious to me from Jonathan's data but the B- cable has
        the better fibers.
3. This data seems to suggest to me that not only is a there a consistent problem if an offset patchcord is used but that the use of connectors with center launches will also occasionally have problems.  The 4.9dB PIE-D value on the champion White fiber really suggests that values over 5.0dB will not be rare.
4.  One idea I'd like to mention for discussion: For handling the data from demo cables to generate PIE-D, in some cases we might be able to make more rigorous estimates by scaling the impulse response (the DMD data) so that it just meets the spec. For example on the 12/96 FDDI cable, scale the impulse response consistent with a 160/500 FDDI spec for OM1 or 500/500 {or 400/400} for OM2. This also allows us to use fibers whose OFL BW fell slightly below by tightening the impulse response to increase it to 500.  It also provides a route to use some of the historical data HP measured on the installed base.  In the case of the OM3 fibers, we have round robin DMD and OFL data and can scale the impulse response so that they just meet the DMD mask.  I don't think this fully addresses the issue of "worst case fibers" but it might be a step in that direction.
5. Finally I wonder in examples like this if some of the measurements should be repeated so we know how much is "noise" and how much is repeatable.

John A.

-----Original Message-----
From: Jonathan King [mailto:jking@BIGBEARNETWORKS.COM]
Sent: Tuesday, December 14, 2004 5:22 PM
Subject: Re: [10GMMF] Task 2 and 4 slides for meeting Dec15th - OM3 fibre measurements

Here are some measurements of OM3 fibre PIE-D values for various launch
conditions, contribution for task 2 and 4 meeting on wednesday 15th Dec
best wishes


tel: 1 408 524 5110
fax: 1 408 739 0568

Jonathan King
Director, Optical Systems
BigBear Networks
345 Potrero Avenue
Sunnyvale, CA 94085