1. Symptoms
In a newly completed Class A office building designed to smart building standards, the comprehensive computer cabling system, which includes both horizontal and vertical cabling systems composed of Category 5e cabling and multimode fiber optics, has undergone strict on-site acceptance testing and inspection based on the Category 5e cabling certification standards. The network and communication systems are currently being set up and tested while the building is being leased out. Most channels in the intelligent control system use IP protocols, and the originally separate control platforms for 17 subsystems have been redesigned and integrated into a single, high-speed 100Base-Tx Ethernet, significantly reducing the cost of the network system. The building’s cabling integrator contacted the Network Hospital today, reporting network connectivity issues on the 66th floor, which were previously working correctly. The problem began two days ago, with users on the 66th floor unable to connect with users on other floors and unable to access the Internet via the building’s frame relay line. The 66th floor is connected to the network monitoring center on the 2nd floor via a 200-meter multimode fiber link. Upon inspection, it was discovered that the connectors in the optical cable splice box on the 40th floor had been contaminated by water leakage from the upper floors. As a temporary solution, the engineering team rerouted the optical cable from a different telecommunications room, adding approximately 30 meters of cable and an additional connector. The originally contaminated connector had been replaced, but the network connection remained unavailable.
2. Diagnostic Process
Based on the pattern of fault statistics, it is generally more likely for problems to occur where maintenance personnel have made changes or modifications during network maintenance. This follows the first principle of fault diagnosis: “check where you’ve made changes.” Based on the reported fault conditions, it was initially suspected that the optical cable was more likely to be the problem. However, network equipment issues, such as optical cards and switches failing simultaneously, could not be ruled out. During today’s inspection, maintenance personnel also checked the optical cards by inserting and removing them. Twenty minutes later, we arrived at the site, connected the network tester to the 2nd-floor network center, and checked the network’s status. The network was functioning correctly, but the 66th-floor users couldn’t be located. After a phone conversation with the 66th-floor users, they mentioned that while connectivity was possible, it wasn’t very smooth. Sometimes, the speed was very slow, and occasional disconnections occurred. We replaced the cable tester with a multimode fiber test module, moved the main unit to the 66th floor, and left the far-end unit on the 2nd floor to test this specific fiber optic link. The attenuation value for fiber A was 3.7dB, while fiber B had an attenuation of 7.8dB. While the attenuation of fiber B was quite high, it still fell within the acceptable range for most optical cards, as it remained within the optical card’s specified range of sensitivity. Therefore, it should not have interfered with the optical card’s ability to receive signals unless the optical card had sensitivity-related issues. To simplify the diagnostic process, we conducted a swap test by replacing both the optical cards on the 2nd and 66th floors simultaneously. Using a network troubleshooter (One Touch) to access the network on the 66th floor, we found that the users on this floor were discoverable, but users on other floors still couldn’t be located. This indicated that the fault still lay within the optical cable link or the optical card’s interface. To determine the exact location of the fault, we substituted another pair of optical cables from a different telecom room and connected the original optical card with jumpers. After the optical card was inserted into the switch, the network immediately returned to normal operation. This confirmed that the switch, the optical card, and the optical card interface were all working correctly. The focus was on examining the optical cable link. The retest results were consistent with the previous test. We reversed the testing direction and found that the attenuation of fiber B was 27dB, whereas fiber A remained at 3.7dB. The attenuation measurement for the segment of cable between the 44th and 66th floors was 20dB, which was significantly out of tolerance. This indicated a serious problem with this link. We removed the optical card connector on the 44th floor, closely examined it with a magnifying glass, and found that the fiber core diameter was consistent with other connectors, with a diameter of approximately 62.5μm. However, when inspecting the optical cable connector on the 66th floor, it was observed that the fiber core diameter was much smaller than that of the other connectors. It was identified as a single-mode optical cable connector. The solution was to swap the receive and transmit positions of this optical cable. Upon inserting it into the optical card, the network immediately returned to normal operation.
3. Conclusion
Optical cable links should undergo bidirectional testing as required in standardized acceptance testing procedures. However, in this building, all optical cable links were only tested in one direction. When fiber diameter mismatches, fiber bubbles, or poor connector quality occur, the attenuation values of fiber optic links in both directions may differ. In general, this difference should not exceed 10%. In this case, the difference in attenuation between the two directions was as high as 20dB, which indicated a significant mismatch in fiber diameter, likely due to connector issues. Therefore, we examined the optical card connectors on the 44th and 66th floors to determine that a single-mode optical cable connector had been incorrectly used. Single-mode optical fiber typically has a core diameter of approximately 9μm and low attenuation for 1310μm and 1550μm single-mode lasers. Multimode optical fiber has a core diameter of approximately 62.5μm and is primarily used for transmitting 850μm multimode signals in computer networks. Single-mode and multimode optical fibers use entirely different light modes, dominant wavelengths, and attenuation mechanisms, and they should not be interchanged. In the case of this faulty connector, during the forward test of fiber B, light energy from the multimode fiber was still able to enter the single-mode fiber connector and passed through the small diameter (single-mode, 9μm) connector to reach the large diameter (multimode, 62.5μm) connector on the optical card. This caused an attenuation measurement higher than a normal link (measured as 7.8dB), but the signal was still usable. During the reverse test, the multimode fiber’s large-diameter core mostly blocked light energy from the connector, resulting in high reverse attenuation (measured at 27dB). Before the “water flooding” incident, the received signal energy from the optical card remained on the edge of the optical card’s sensitivity range. Consequently, the network’s performance was unstable, and sometimes the speed was very slow, while occasional interruptions occurred (influenced by changes in temperature and air pressure). After the “water flooding” incident, when additional 30 meters of cable and an extra connector were introduced during link restoration, the optical card’s received energy exceeded the edge value of sensitivity. As a result, network connectivity was lost. In cases where multimode optical cards are used, the fiber cables are paired unidirectionally, meaning one fiber is used for transmission, and the other for reception. Swapping the fiber cables, both transmission and reception for the optical cards, used the direction with lower unidirectional attenuation values, leading to stronger light signal reception. The network was able to return to normal operation.
This problem could have been directly observed using an Optical Time-Domain Reflectometer (OTDR), as the instrument’s screen displays discontinuities in the echo curve. Experienced testers would immediately recognize it as a problem caused by mixed fiber types.
4. Diagnostic Recommendations
First, replace the incorrectly used single-mode connectors as soon as possible. Second, ensure that all fiber optic links undergo bidirectional testing according to standardized construction and acceptance requirements. Third, the building’s design blueprints lack the specified attenuation values for fiber optic links. The tested fiber attenuation values, regardless of whether they belong to normal or abnormal links, exceed the design values. It is suspected that substandard fiber optics and connectors are being used, and it cannot be ruled out that multiple segments of fragmented fiber optic links have been spliced together. Therefore, it is recommended that the building owner instruct the integrator to check the number of actual connectors and fusion splices.
5. Afterword
Five days later, the users reported that they had tested most of the fiber optic cable links, and the actual tested attenuation values (minus the connector loss) were generally in line with standard values. No evidence of fragmented spliced links was found. However, most of the connectors and terminations were found to be substandard and of low quality. Fortunately, replacing the connectors is relatively straightforward and should not significantly impact existing network users. The plan is to replace all of these connectors.