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

[10GMMF] Cambridge Rel 3.0 scaled profiles

I brought up an observation on the channel modeling teleconference earlier today regarding the treatment of fiber 41 in the Release 3.0 package some of you may have received recently that contains scaled index profiles.   Richard Penty asked that I capture my thoughts in an email to the reflector, so here they are.

First I would like to say that the modifications to the profile scaling process that modify the radial width of the center defect and the onset of the edge perturbation are probably good adjustments.  But one issue has crept in.  

The fiber in question started out as one devoid of intentional perturbations. The DMD scaling process magnifies the small alpha non-idealities to the point where they reach a limit of 500 MHz-km OFL BW. This apparently occurs at about 1 ns/km DMD.  The scale factor for this fiber turns out to be about 10, which is the largest of any fiber in the set by roughly a factor of 3.  The release note expressed concern that the profile scaling process could not be applied to a fiber with a perfect profile, and therefore the profile supplied was not scaled (although a DMD scaling factor is provided).  In examining the DMD structure of the scaled fiber, it looks realistic.  So the result of the DMD scaling did not produce a nonphysical result.  In addition, the structure appears to be of a type that is not commonly found in the other fibers of the set and therefore its presence could be of additional value to the simulations.  So I would recommend that this scaled fiber be retained in the analysis.

I think the issue is how to scale a perfect fiber using the profile scaling algorithm established for the other fibers of the set.  There is no center or edge perturbation to adjust and the alpha error is zero.  So for at least this one case (and more rigorously for all cases) I believe the correct approach may be to scale relative to the profile of a fiber that generates zero DMD.  Such a fiber will nominally have a constant alpha, but with some small variations that produce zero DMD.   These variations provide the basis for scaling the "perfect" fiber (and all others, but the added value is likely small in most cases).

Paul Kolesar
1300 East Lookout Drive
Richardson, TX 75082
Phone:  972.792.3155
Fax:      972.792.3111