Expected fluctuation of dwdm span loss.

[spemiller]

We recently added a redundant connection between two sites with ONS15454 chassis by adding two spans and an amplifier node in the middle. This was added to our existing five nodes. The original spans have stable span losses staying well within 1dB loss fluctuation. The new spans with the amplifier node are showing up to 8dB fluctuation in span loss. The new spans on either side of the amplifier node are 59 and 117km long. We have other spans of 112 and 123km which don't show any fluctuation. I noticed that the span loss looks to correlate to change in outside air temperature. Even if these spans are mostly aerial, should we expect this much change? We looked at a spec sheet from Corning fiber and according to their spec we should see less than 0.05dB/km across a 96degC temp range. We saw 8dB span loss change with a 33degC temp change on the 117km span.

[Tom Randstrom]

I've never heard of that level of fluctuation.

Are both the TX and RX fibers impacted on the span?

Is the amplifier node located in a CO or environmental enclosure?

My suspicion is a bend in the fiber somewhere in the outside plant. The change in temperature may be impacting some plastic or metal device that may be up-against the fiber (slice closures are always suspect due to tight fiber routing spaces.

[spemiller]

Attenuation is impacted in both directions on both spans and looks to be about the same in each direction. A third party characterization shows loss at what we expected. The TX levels from each end stay stable. The amplifier node is co-located in an ISPs equipment room and environmentals are controlled well.

Any other ideas? Do you think we should present this to the fiber vendor?

Thanks for your help.

[Tom Randstrom]

Interesting... two spans with both TX and RX fibers impacted.

Are the impacted spans on opposite sides (East and West) of the amplifier node?

Maybe an OTDR shot at high and low outside temps would isolate a high attenuation point in the network; splice point, pedestal, CO, other, that might point to an issue.

Gut tells me it is not a fiber mfg issue, but they may be able to lend some ideas due to experience.

[spemiller]

Thanks for your reply.

The impacted spans are on opposite sides of the amplifier node. Just to be more clear, my reference to the fiber vendor is the company we lease the fiber from, not the manufacturer. It might be difficult to schedule a OTD and try to predict high and low temps etc. Can anyone think of why an amplifier node would cause the these fluctuations if it isn't fiber. The output power on the nodes at the end points of these spans shows very stable output +-0.1 dB.

[Tom Randstrom]

Are you amplifying a single-channel on these spans?

What optical amplifier is being used?

What are the terminating cards on the end of the spans.

Are there any alarms?

Sorry, but I may be just not understanding the installed network and am overlooking something.

[spemiller]

Hi,

Thanks for your interest.

There are 4 channels on the affected spans. There haven't been any alarms since we added attenuation to the MXPP trunk RX ports to cover the power fluctuation and we increased the range of expected span loss. The terminating cards are MXPP configured with client ports as 1Gig Ethernet. One end of all 4 channels terminate at “E” site at one end of the amplified spans with the other end of the channels at 4 different sites, one of which is the other end of the affected spans. The amplifier node “NL-1” is actually an OADM without any client cards installed or circuits provisioned with a local drop. Slot1=OPT-BST, slot2=OPT-PRE, slot3=AD4C, slot15=AD4C, slot16=OPT-PRE, slot17=OPT-BSTE. The AD4C express ports connects sides A and B on NL-1. NL-1 side A connects to node “E-1” side B over 59km of fiber with MXPP cards in second chassis. NL-1 side B connects to node “S-2” side A over 117km. All other sites including E and S have a ROADM type setup with either OPT-BST or OPT-BSTE boost amps, OPT-PRE, 32WSS and 32DMX cards. At the E site, there is a second chassis, E-2 with 4 MXPP cards and side A of E-1 has BST, PRE, WSS, and DMX cards. The S site shelf 2 has a second chassis S-1 with BST, PRE, WSS, and DMX cards on both side A and B. S site shelf S-2 has 4 MXPP cards on side B and BSTE, PRE, WSS, and DMX cards on side A. Since the addition of these new spans was to finish our build-out and provide a redundant path, I think the E site is the only one without any pass-thru circuits.

I'll try attaching a basic network drawing as a png file. Distances are in miles.

[Tom Randstrom · Accepted solution]

I can't figure out why your system is responding the way it is. The need to add external attenuation on the MXPP definitely shows an issue.

As you mentioned earlier, you probably should contact the fiber's owner and see if they can troubleshoot or provide assistance. Maybe Cisco TAC can provide some insights.

I wish I had an answer.

[viyuan700]

Have you check the gain of your amplifier,when you are experiencing this fluctuation?. Can be a problem with your amplifier gain?

Like Tom i dont suspect the fiber as problem with fiber can be digital like effect either something or nothing.

[spemiller]

Thanks.

Sorry, I grouped my reply to you in with another. The gain and output at the end nodes of the problem circuit stays very stable.

[viyuan700 · Accepted solution]

if these spans are mostly aerial, should we expect this much change

Was reading ur message and saw mostly the cable is aerial. We shd not expect that much change.

Do you have any extra fiber where u can check power during the time it fluctuate in your 15454. Or can ask the vendore who is leasing you to check what is loss during that time.

Possible that aerial cable is experiencing something w.r.t temperature fluctuation .

As you said there are no alarms but if something is between the range of equipment why will it raise an alarm. Have you checked output power of all cards involved in between during that time.