issue on CWDM high Rx power

[Tony Cheng]

i have 2 test switches setup in a LAB, each switch connecting to each over CWDM-Mux-8A, using various differenet wave CWDM SFPs on both switches, and i have  tried it using different ports on the Cisco switch, different  CWDM MUX-8A, different CWDM SFP 1490, 1590, 1550, I even tried different Cisco switch, different switch port, and using different single mode fiber patch cable. All are new SFP and new patch cable and new CWDM MUX-8A.  I even bypass the CWDM 8-port MUX directly, just connecting CWDM SFP 1490 ---to-----CWDM SFP 1490 directly, and also did the same for CWDM 1590….etc.   But i'm gettting below:

BAD Case:  if I push the LC connector on both ends of the single mode fiber cable, all the way into the CWDM SFPs on both ends of this fiber cable , and stop when you hear the “clicking” sound, the interface just won’t come up on both sides of the switch, below is the show interface transceiver:

Test-SW01#

Test-SW01#sh int trans module 1

If device is externally calibrated, only calibrated values are printed.

++ : high alarm, +  : high warning, -  : low warning, -- : low alarm.

NA or N/A: not applicable, Tx: transmit, Rx: receive.

mA: milliamperes, dBm: decibels (milliwatts).

                                 Optical   Optical

           Temperature  Voltage  Tx Power  Rx Power

Port       (Celsius)    (Volts)  (dBm)     (dBm)

---------  -----------  -------  --------  --------

Gi1/1/1      38.2       3.55 +     8.1 +   -40.0

Gi1/1/4      39.3       3.30       2.1      -9.1

Test-SW01#

Good Case:  but if I let one end of the fiber cable loosely hangs on the SFP port, not inserting fully by avoiding hearing the “click” sound when LC connector connecting to the CWDM SFP on one of the switch, then the interface will be UP on both side:

Test-SW01#

Test-SW01#

Test-SW01#sh int trans module 1

If device is externally calibrated, only calibrated values are printed.

++ : high alarm, +  : high warning, -  : low warning, -- : low alarm.

NA or N/A: not applicable, Tx: transmit, Rx: receive.

mA: milliamperes, dBm: decibels (milliwatts).

                                 Optical   Optical

           Temperature  Voltage  Tx Power 

Rx Power

• •12.Port       (Celsius)    (Volts)  (dBm)     (dBm)
• •13.---------  -----------  -------  --------  --------
• •14.Gi1/1/1      39.2       3.29       2.8      -8.0
• •15.Gi1/1/4      39.3       3.30       2.2      -9.0
• •16.
• •17.
• •18.Test-SW01#

why is that ?  how do i attenuate the Rx Power from -40 down to -8 ?  do i need an external optical attenuator to adjust the power ?  or is it because these are in LAB environment, the patch cable are too short, thus power is high ?

thanks,

[Marvin Rhoads]

-40 dBm is a LOW power reading - essentially what DOM tells you when the received signal is below the Signal to Noise S/N) ratio floor of the receiving module. You will see that when there is either extremely low or no signal at all.

Based on what you described, you most likely have a bad physical layer connector at one end or the other.

[Tony Cheng]

thanks, let me try differenet patch cables then, again these are just short single mode fiber patch cable, no more than 2 meter in length and, i have replaced them twice already, each time they are all brand new fiber patch cable and brand new CWDM SFP, or even doing direct connection from CWDM SFP ----to---- CWDM SFP.

let me try replacing them again.

[Tony Cheng]

yea, i tried using new cisco switch, new cwdm sfp, new single mode fiber patch cable 15 meter, and bypass the CWDM Mux-8A, just do a direct connection from new cwdm sfp ----to ----- cwdm sfp, same result, each time when i connect both end of the LC connector fully, by inserting into the cwdm sft and stopped when i hear the "click" sound, the  dbm becomes - 40, and interface is Down on both end.  And by having it loosely connected on one end, meaning not inserting fully...let it hangs there without hearing the LC connector "click" sound,.then interface will be UP on both side with say -7 or -10 dbm...

any idea why this is so ?  thanks for your help

[Tony Cheng]

i think i maybe experiencing LC connector insertion power loss, but how to fix this.

[Marvin Rhoads]

Perhaps your CWDM SFP has a mechanical issue with the housing, making full insertion misalign the fiber cores.

[Tony Cheng]

thanks, yea i have a TAC case open.

[Tom Randstrom]

I'm very curious what the issue is. Strange that it would work when the connector is fully seated in the SFP port.  Are the optical jumpers from a name brand manufacturer? Is the connector end green or blue in color?

When connecting a CWDM SFP back to back, it should have 10dB attenuators placed in the path to prevent overloading the optical receiver in the SFP.  It is possible, even with the MUX8A filters in-line with a short 15m fiber jumper, that you are over-loading the receivers; causing the system not be be able to decipher the signal. The info you provided above shows a transmit level of +8.1 dBm (higher than the typical +3dBm specification).    

When you determine what the problem is, please post it on this site! 

Thank you,

[kevin.hill]

sounds like you have a burnt SFP RCV port or more than likely a bad fiber path or bad jmpr. what you could check is make sure your fibers are tapped correctly. if you are using molex jumpers or really any manufacture, the fiber that has writing on it usually is the XMT fiber and the one that is blank is RCV. also check the SFP specs for optical PWR. it sure does sound like either you have a SFP WL mismatch, bleeding SFP or a bad jumper. before attaching the fiber to the SFP's. test the fiber with a test scope

Sent from Cisco Technical Support iPhone App

[chris-hart]

By not clicking the connectors fully home you introduce an air gap/attenuation, try inserting a 10dB PAD