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

Re: New 802.3ae link model




Piers,

As a prospective customer to deploy 1300nm serial PHY, I took a look at 
that specific link model.  Is the total link budget the 10db (Power Budget) 
plus the 2db (Connector Budget) or is it only the 10db (Power Budget)?  If 
it is only the 10db (Power Budget), then the total fiber attenuation budget 
is 8db.  What was the physical connector definition that was used to make 
the determination of the number of connectors used to determine the 
Connector Budget?  Was it one each connector at the 802.3ae port interfaces 
on each end of the link and a total of  only two patch panels?  If so, this 
may not be enough connections.  There will normally be a total of four 
patch panels in addition to the connections at the interface port, as well 
as an internal patch from the optical transceiver and the interface surface 
bulkhead.

Thank you,
Roy Bynum

At 08:00 PM 1/2/01 +0100, DAWE,PIERS (A-England,ex1) wrote:

>All,
>
>There is a new version of the link model at
>http://www.ieee802.org/3/ae/public/email_attach/10GEPBud2_4_1.xls
>
>This updated spreadsheet contains:
>
>Proposal for 850 nm,
>OMA definition of transmit power,
>RIN and RIN(OMA) are distinguished,
>Worst extinction ratio is calculated,
>1550 nm powers raised by 2 dB,
>A simple Polarisation Mode Dispersion formula.
>
>The new features can be overridden as explained in the notes page.
>
>For those who like equations, some detail follows.
>
>Piers
>
>
>The PMD (Polarisation Mode Dispersion) formula uses an input parameter
>DGDmax, which the equalization ad hoc tell me is 19 ps for 1550nm.  I have
>taken a guesstimate of 10 ps for 1300 nm.  The calculation I used is as
>follows:
>
>If DGDmax is used               (box E7)
>         Effective BW1(MHz.km) = 10^6*Ltarget(km)/(3*DGDmax(ps))
>else
>         Effective BW1(MHz.km) = BWm(MHz.km)
>endif
>         Effective BW2(MHz) = Effective BW1(MHz.km)/L(km)  (column E)
>
>Notice two quantities with identical names but different units.
>
>You can get the factor of 3*DGDmax by considering
>         sin(2pi.f.t+a) + sin(2pi.f.t-a) = sin(2.pi.f.t)
>where f is the bandwidth and 2a represents the phase splitting of a pulse
>caused by DGDmax.  This result was known to the 802.3z community.  You can
>get a very similar result by working through the standard deviation of a
>pair of Gaussian pulses.
>
>The effect of Polarisation Mode Dispersion seems to be less than we had
>feared.
>--
>         Phone   +44 (0)1473 465654 or Telnet (31)25654
>         Fax     +44 (0)1473 464965   piers_dawe@xxxxxxxxxxx
>         Agilent Technologies   White House Road   Ipswich
>         Suffolk   IP1 5PB   England