Wrong module, wrong distance, or a module that's simply gone bad — picking the right optic and diagnosing one that's already installed both come down to the same handful of display commands and the same handful of numbers on a spec sheet. This is the selection logic, the compatibility rules, and the fault ladder taken straight from Huawei's own S-series switch maintenance handbook.
By the AtlasCommTech engineering team — 13 years of carrier & enterprise network deployments · Updated July 2026
A mismatched optic and a genuinely failed one look identical from the port's LED — the spec sheet is what tells them apart.
Picking the right optical module up front and diagnosing one that's already installed are really the same job worked from two directions. Selection means matching rate, wavelength, distance and single- or dual-fiber format to the actual run — get any one of those wrong and the symptom looks exactly like a hardware fault: a port that won't come up, or comes up and drops packets. Diagnosis means reading the same handful of numbers — RX/TX power, FEC error counts, port-flap logs — off a module that's already in the field.
What follows is both halves in order: how to read a module's standard name well enough to predict compatibility before you order anything, a practical selection table by rate/wavelength/distance/fiber count, the display transceiver commands that confirm what's actually installed, and the field fault ladder for a dark link, marginal RX power or an error floor that won't go away — all taken from Huawei's own S-series switch maintenance handbook.
A standard name like 1000BASE-LX10 already tells you the rate, the wavelength and the coding — read it correctly and you can predict compatibility without a datasheet in hand.
Take 1000BASE-LX10 apart piece by piece: "1000" is the rate (10M, 100M, 1000M, 10G, 40G and 100G all follow the same pattern), "BASE" means baseband transmission, "L" is the laser's center wavelength, and the trailing "10" is the channel count (no digit means a single channel).
| Wavelength letter | Meaning |
|---|---|
| S | Short wavelength — 850nm |
| L | Long wavelength — 1310nm |
| E | Extra-long wavelength — 1550nm |
| B | Bidirectional long wavelength — a single-fiber (BiDi) module |
The basic compatibility rule follows directly from the name: two modules that support the same standard can talk to each other — the standard already fixes rate, wavelength and distance, and the package (SFP, SFP+, XFP, QSFP+...) doesn't have to match for that. Modules on different standards may or may not interoperate; treat that as a question for support, not a guess, and the same rule holds when a Huawei switch is facing a different vendor's router or switch on the other end.
The numbers that actually decide which module goes in a given run — rate, wavelength, maximum distance and whether it needs one fiber or two.
| Standard | Rate | Common package | Wavelength | Typical max distance | Fiber |
|---|---|---|---|---|---|
| 1000BASE-SX | GE | SFP | 850nm | Up to ~550m, multi-mode | Dual-fiber |
| 1000BASE-LX10 | GE | SFP | 1310nm | Up to 10km, single-mode | Dual-fiber |
| 1000BASE-BX10 (BiDi) | GE | SFP | 1310nm / 1490nm, paired | Up to 10km, single-mode | Single-fiber (BiDi pair) |
| 10GBASE-SR | 10GE | SFP+ / XFP | 850nm | Up to ~300m, multi-mode | Dual-fiber |
| 10GBASE-LR | 10GE | SFP+ / XFP | 1310nm | Up to 10km, single-mode | Dual-fiber |
| 10GBASE-ER | 10GE | SFP+ / XFP | 1550nm | Up to ~40km, single-mode | Dual-fiber |
| 40GBASE-SR4 | 40GE | QSFP+ | 850nm | Up to ~100-150m, multi-mode (MPO) | Dual-fiber |
| 40GBASE-LR4 | 40GE | QSFP+ / CFP | 1310nm (CWDM) | Up to 10km, single-mode | Dual-fiber |
| 100GBASE-SR4 | 100GE | QSFP28 | 850nm | Up to ~70-100m, multi-mode (MPO) | Dual-fiber |
| 100GBASE-LR4 | 100GE | QSFP28 / CFP | 1310nm (CWDM) | Up to 10km, single-mode | Dual-fiber |
| 100GBASE-ER4-lite / ER4 | 100GE | QSFP28 / CFP | 1310nm | Up to 30-40km, single-mode | Dual-fiber |
100GBASE-ER4-lite interoperates with 100GBASE-ER4 up to a maximum of 30km — one of the few cross-standard exceptions the handbook calls out explicitly. Beyond ~10km on any long-distance module, check the actual fiber run length against the module's stated distance before ordering — a long-distance module driven over a short, low-loss run is its own separate problem, covered below.
Two commands tell you everything the module itself is reporting — full detail, and a fast pass/fail read.
display transceiver interface ... verbose gives the complete picture: type, connector, wavelength, rated distance, and the live diagnostic block — temperature, voltage, bias current, and both RX and TX power in dBm, each against its own high/low threshold.
<Quidway> display transceiver interface gigabitethernet 0/0/1 verbose
Gigabitethernet0/0/1 transceiver information:
-------------------------------------------------------------
Common information:
Transceiver Type :1000_BASE_SX_SFP
Connector Type :LC
Wavelength(nm) :850
Transfer Distance(m) :300(50um),150(62.5um)
Digital Diagnostic Monitoring :YES
Vendor Name :SumitomoElectric
Vendor Part Number :HFBR-5710L
-------------------------------------------------------------
Diagnostic information:
Temperature(°C) :26.00
Voltage(V) :3.29
Bias Current(mA) :4.57
RX Power(dBM) :-40.00
RX Power Low Threshold(dBM) :-16.99
TX Power(dBM) :-5.03
-------------------------------------------------------------display transceiver diagnosis is the faster read when you just need a pass/fail: it puts current value, low/high alarm threshold and a normal/abnormal status side by side for every parameter.
<HUAWEI> display transceiver diagnosis interface GigabitEthernet 0/0/8
Port GigabitEthernet0/0/8 transceiver diagnostic information:
Parameter Current Low Alarm High Alarm
Type Value Threshold Threshold Status
------------- --------- --------- ---------- --------
TxPower(dBm) -5.44 -11.00 -1.00 normal
RxPower(dBm) -30.22 -22.02 -1.00 abnormal
Current(mA) 12.05 0.00 60.00 normal
Temp.(°C) 53.69 -10.00 80.00 normal
Voltage(V) 3.26 2.97 3.63 normalTo confirm a module is genuinely Huawei-certified rather than just labeled that way: on V200 devices, check display elabel for a Manufactured date after the certification cutover (2013-07-01 for GE/10GE-and-below modules, 2016-01-01 for 25GE/40GE/100GE), confirm display version is on a supported release, and confirm display transceiver's Vendor Name field reads HUAWEI. On V600 devices, the equivalent commands are display device elabel, display version and display interface transceiver. All three conditions have to hold together — any one missing means the module isn't certified.
Once RX/TX power looks marginal rather than clearly good or clearly dead, the deeper read of what those numbers mean — and the difference between a center-module and remote-module reading on a passive optical link — is its own note: see PEN Optical Module Diagnostics: Reading Rx/Tx Power Like a Pro.
A dark link, a suspiciously low reading or an error floor that won't clear — this is the order that actually isolates the fault instead of swapping parts at random.
Diagram labels are kept in English for engineering clarity.
A 100GE-specific version of step 3 and 4: when a 100GE port won't come up, display phyregister information's FEC counters tell you directly whether the module itself is bad — a Fec Uncorrect Count that keeps climbing alongside RX Loss Of Signal points at hardware, not configuration.
<HUAWEI> system-view
[HUAWEI] diagnose
[HUAWEI-diagnose] display phyregister information interface 100GE 0/0/2 status
Unit 0 Port ce0 Information:
Physical State(PMD): DOWN
RX Loss Of Signal: Yes
Current Lane1 RX Power(dBM): -40.00
Local Fault: Yes
FEC Status: Yes
Fec Correct Block Count: 0
Fec Uncorrect Block Count: 0
Fec Error Count: 6
Fec Correct Count: 0
Fec Uncorrect Count: 6
// Fec Uncorrect Count rising alongside RX Loss Of Signal -- treat as a hardware fault on the module itselfThe ones that get mistaken for a hardware fault, or the other way around — pulled from real field cases.
SYMPTOMA port cycles up and down repeatedly over a few seconds, with no configuration change and no obvious cable problem.
CAUSEA non-Huawei-certified module raises a hwOpticalUnAuthorized alarm and, in real cases, has been observed to cause exactly this kind of repeated up/down cycling — a real log shows the same GigabitEthernet interface flapping every few seconds right alongside the unauthorized-module alarm.
FIXCheck the alarm log for hwOpticalUnAuthorized before chasing a cable or configuration explanation, and replace with a certified module if it's present.
%%01DEVM/1/hwOpticalUnAuthorized(t)... The optical module was not certified by Huawei Ethernet Switch.
(EntPhysicalName=GE1/0/26, Reason=... Huawei cannot ensure that it is completely adaptive...)
...IFPDT/4/IF_STATE(l)[0]:Interface GigabitEthernet1/0/26 has turned into UP state.
...IFPDT/4/IF_STATE(l)[1]:Interface GigabitEthernet1/0/26 has turned into DOWN state.
...IFPDT/4/IF_STATE(l)[2]:Interface GigabitEthernet1/0/26 has turned into UP state.SYMPTOMA GE (1000M) module dropped into a 10GE port either refuses to link at all, or links but only in a fixed non-negotiated mode.
CAUSEA 10GE port packaged as XFP does not support a GE module at all; only SFP+-packaged 10GE ports can take one, and even then support depends on the specific switch model and software version. On S6700EI specifically, an inserted GE module forces the port to 1000M in a fixed non-auto-negotiation mode unless a specific patch is applied — meaning the peer port has to be forced to the same non-negotiated mode to come up at all.
FIXConfirm the port's package type (XFP vs SFP+) before assuming a GE module will work, check the specific chassis/model's own support table, and on affected models either apply the relevant patch or force both ends to matching negotiation modes.
SYMPTOMTwo BiDi (single-fiber) modules are installed, the fiber is connected, and the link still won't come up.
CAUSEBiDi modules must be used in matched pairs by wavelength direction, not simply as "the same part number on both ends" — one end needs a TX1310/RX1490 module and the other end needs the complementary TX1490/RX1310 module. Two modules with the same TX/RX wavelength assignment on both ends will never link.
FIXConfirm each end's module reports the opposite TX/RX wavelength pairing before troubleshooting anything else on a BiDi link.
SYMPTOMRX Power reads above the high-alarm threshold rather than below the low one — the opposite of the more familiar "weak signal" complaint.
CAUSEA long-distance module's transmit power is calibrated for the loss a long fiber run is expected to introduce. Run it over a short, low-loss link instead and the far end receives far more signal than intended — over-driving the receiver and, over time, risking damage to it.
FIXWhenever a module rated for more than roughly 10km is installed on a run shorter than that, add an inline optical attenuator sized to bring RX power back within the module's normal range — this is routine, not a sign anything is wrong with the module.
SYMPTOMA 100GE port won't come up, RX power reads essentially nothing (around -40dBm), and it isn't obvious yet whether the module, the fiber or the peer is at fault.
CAUSEdisplay phyregister information's FEC fields are built specifically to separate a hardware fault from a configuration or fiber problem: RX Loss Of Signal true, combined with Fec Uncorrect Count that keeps incrementing rather than sitting at zero, is a documented signature of a hardware-faulty 100G module.
FIXRead the FEC counters before swapping fiber or re-checking configuration on a dark 100GE port — a climbing Fec Uncorrect Count with RX Loss Of Signal set means order a replacement module, not a cable.
SYMPTOMA 10G port with a GE module installed links successfully against some peer devices and fails intermittently against others, with no pattern in the optics themselves.
CAUSEA documented chip-level defect on certain S-series releases causes exactly this: once a 10G port adapts to 1000M for a GE module, a small clock-frequency deviation between the two ends — normally within tolerance for Ethernet's asynchronous timing — sits inside a range the affected chipset can't handle at 1000M, making the link fail probabilistically rather than consistently.
FIXOn confirmed-affected releases, apply the documented patch or upgrade (the handbook lists specific SPH/version fixes per release train) rather than continuing to swap modules or fiber — this is a firmware defect, not a hardware or cabling problem.
The questions that come up first, whether you're ordering a module or already troubleshooting one.
Yes, as far as the optical standard is concerned — package format (SFP+, XFP, QSFP+...) isn't part of what has to match for two modules on the same standard to interoperate. What still has to match is the physical port itself: the port has to be built to accept that package in the first place, which is a separate question from optical compatibility.
No. It depends on the port's package: XFP-packaged 10GE ports don't support GE modules at all; SFP+-packaged 10GE ports can, but whether they actually do — and whether the resulting 1000M link auto-negotiates cleanly — depends on the specific switch model and software version. Check the model-specific support table before assuming it will work.
Too low usually means the signal has been attenuated more than expected — check the actual fiber run against the module's rated distance, and check the fiber path itself (connector loss, excessive bend, single-mode/multi-mode mismatch) and the far end's TX power. Too high usually means a long-distance module is running over a shorter, lower-loss path than it was designed for, and needs an inline optical attenuator to protect the receiver.
BiDi modules have to be paired by wavelength direction — one end TX1310/RX1490, the other end the complementary TX1490/RX1310. Two modules with the same wavelength assignment on both ends, even if they're otherwise identical, will never link.
It's one of the three usual sources, alongside CRC errors and outright discards — see Network Packet Loss: Optics, CRC Errors and Discards Diagnosis for how to tell which of the three is actually responsible before replacing anything.
Three conditions have to hold together: the device's elabel Manufactured date has to be after the certification cutover for that module's speed class (2013-07-01 for GE/10GE-and-below, 2016-01-01 for 25GE/40GE/100GE), the software version has to be a supported release, and the transceiver's own Vendor Name field has to read HUAWEI. Missing any one of the three means it isn't certified, regardless of what the physical label says.
This handbook is built from the optical-module TechNotes chapter and the physical-interface-DOWN FAQ chapter of Huawei's own S-series switch maintenance handbook. Model-specific support tables — which exact chassis, line card or software version accepts a GE module in a 10GE slot, for instance — change across hardware generations; always cross-check the specific device's own data sheet before ordering, and treat any result that deviates from what's expected as a question for support rather than a guess.
Send us the port type, the run distance, and — if it's already installed — your display transceiver diagnosis output, and we'll help you land on the right part or the right fix.