Open app

Link errors because there is too much light

common

“Every loss test passes and the link still flaps and errors. Short run, brand new optics, and cleaning the connectors made it worse.”

Likely causes

  • Long-reach or extended-reach optic on a short link

    most common

    Measured receive power sits above the optic's stated maximum input. A short singlemode hop (inside a data hall, across a campus, a metro span that got shortened) running an optic built for tens of kilometres. Cleaning the connectors raises the power and makes the errors worse, which is the giveaway nobody expects.

  • Inline attenuator missing, wrong value, or pulled during a move

    common

    Worked for a year, moved or repatched, broke. Somebody found a little barrel in the path and took it out, or fitted the value that was in the bag instead of the value the link needs.

  • Link shortened or rerouted after the optics were chosen

    common

    As-built says kilometres, the actual path is hundreds of metres. New building entry, new duct route, a span that got cut over to a closer node. Measured loss is far below the budget the design assumed.

  • Wrong optic part number fitted at one end only

    less common

    One direction errors and the other is clean. Read both part numbers, not the two labels on the cage.

  • Not overload at all: the module's own reading is out

    occasional

    Switch reports a level the meter doesn't agree with. A module's digital diagnostics are a diagnostic, not a calibrated measurement, and accuracy varies by part. Confirm with a meter before you fit hardware.

What to bring

  • optical power meter
  • switch or transport optics readout (module digital diagnostics)
  • transceiver data sheets for both part numbers
  • fixed-value inline attenuators, correct wavelength, assorted values
  • OTDR for real length
  • as-built and design budget

Safety

This is the one link on the job with more light in it than usual, so treat it accordingly: prove it dark with a power meter before you unmate, inspect or scope anything, never look into a port, adapter or cord end, and cap every port you open. ANSI Z136.2, IEC 60825-2.

Steps

  1. Get the optic's receive window and compare it to the actual level

    Pull the transceiver data sheet for the exact part number at each end. You want two figures: minimum sensitivity and maximum input power. Both vary by part and by vendor, so read the part's own numbers rather than a class rule of thumb. Then read the reported receive power at both ends off the switch's optics display.

    If that doesn’t do it

    No data sheet available for that part: measure the level anyway and escalate to whoever specified the optic. You cannot judge a level against a window you don't have.

  2. Confirm the level with a meter at the receive connector

    Meter set to the right wavelength, taken at the connector the receiver actually sees, not at a panel two jumpers upstream. Module diagnostics are useful for spotting the problem and not accurate enough to size the fix.

    If that doesn’t do it

    Meter agrees with the budget and the module doesn't: suspect the module. Swap the optic before you touch the cabling.

  3. Check measured loss and real length against the design

    Overload is a design mismatch, so prove the mismatch. Measured loss well under budget, on a path much shorter than the drawing, with an optic rated for a long span, is the whole story. Get the real length off the OTDR, not off the print.

    If that doesn’t do it

    Length and loss match the design and the level is still too high: read the optic part numbers again at both ends.

  4. Size an attenuator instead of guessing one

    Aim for the middle of the optic's receive window, not the edge: attenuation needed equals measured power minus your target level. The value is specific to this link, so calculate it. Use a fixed-value attenuator rated for the wavelength in use, fit it on the receive side, and record the value on the port label and the as-built.

    If that doesn’t do it

    Correct value fitted and errors persist: measure again with the attenuator in, then treat it as a normal loss problem and work the loss-budget sequence.

  5. Or change the optic to the right reach class

    Cleaner than an attenuator and one less part to lose in three years. If the link is short, fit a short-reach optic at both ends. Confirm both ends are the same wavelength and class after the change.

    If that doesn’t do it

    No suitable optic available: run the attenuator and flag it in the closeout so the next tech knows the level is engineered, not accidental.

  6. Clear the counters and watch it under load

    Reset error counters, then load the link and watch. Overload errors follow traffic and temperature, so a quiet five minutes proves nothing. Record the receive level at both ends after the fix.

    If that doesn’t do it

    Level is now inside the window and it still errors: the fault is not optical. Hand it to the network side with the levels attached, or check fiber grade and mode conditioning in the fiber-type sequence.

References

  • IEEE 802.3 (Ethernet optical PMD receive power ranges, per PMD)
  • SFF-8472 (module digital diagnostic monitoring, where the reported level comes from)
  • ITU-T G.957 and G.959.1 (optical interface parameters for transport systems)
  • transceiver vendor data sheet (minimum sensitivity and maximum input; varies by part)

Check it with a calculator

More Fiber Optic guides

All 54 guides