Back Reflection

[notabot]

We recently purchased and built out a DWDM mux solution that includes NCS 1001 amplifiers. The amplifiers are connected via an 11km dark fiber span. We have been running on it for about 3 months without any issues. Since we did not have any experience with DWDM's, the amps have been a bit of a learning curve (we used to have passive CWDM muxes in it's place, but were told we need amplifiers to upgrade from 16 CWDM channels to 40 DWDM channels). The build-out was rushed, so we unfortunately didn't have time to ensure all the metrics were in tolerance before having to go live. In following up after the install, we opened a TAC case to find out what the recommended tolerances were for TX/RX Power, Back Reflection, etc. After receiving the tolerances, the primary concern was that our Back Reflection value reported by our amplifier was at -1.60 dBm and should be at least <-25 dBm, and ideally <-30 dBm for minimal reflections. We were able to reconnect a fiber patch cable a week ago to bring it up from -1.6 to -18.7 dBm, a dramatic improvement. However, Cisco still recommends (as expected) to bring this fully into tolerance. Since we do not have an OTDR, we asked our vendor to give us their result but we still have not received it. Has anyone out there run at -18.7 dBm or worse for a long period of time and had any issues? Or is the <-25 dBm requirement just Cisco being overcautious?  My concern is the uncertainty of IF we will be able to find a way to get this into tolerance, then what?  I don't want to be in a position where we are a ticking time bomb.  I presume that as we add more channels, we might start seeing compounding effects of the Back Reflection. We currently have about 14 of 40 channels in use, and no transmission or receive errors on any of the SFP's.  Thank you

[tiagobsb1@gmail.com]

muito bom 

[Joseph W. Doherty]

Not my area of expertise . . .

"I presume that as we add more channels, we might start seeing compounding effects of the Back Reflection."

Unsure that matters.

". . . should be at least <-25 dBm, and ideally <-30 dBm for minimal reflections."

From a quick scan of the literature (beyond Cisco), -30 dBm is a good target, but possibly even -20 dBm is acceptable (although you've not reached that).

The two noted problems with an unacceptable level of back reflection is data corruption and/or damage to lasers.

The common cause of high reflection values is poor optical connections, often being dirty or just poorly connected (creating an air gap).  The back reflection of a connection, even when "clean", depends much on kind of connection.

It appears the first mitigation approach is to insure optical connection are clean and well connected.

I've read increasing attenuation is a possible mitigation technique.

Another is to is an "isolator".  (Which allows light to pass in one direction but not other.)

A couple of articles on this subject, that might be worth review:

https://www.truecable.com/blogs/cable-academy/basic-principles-of-fiber-optics-series-2-reflection

https://kingfisherfiber.com/application-notes/orl-back-reflection-guide/