Polarization Extinction Ratio (PER) is one of the key parameters for Polarization Maintaining Fiber (PMF) connector. Based on our previous studies, the bending radius of fiber greater than 1.5 cm will not affect the insertion loss of PMF . Moreover, the measured PER of Panda PMF with LC/UPC connectors is more stable when that PMF is coiled around a hot rod with a minimum of 3-cm in diameter at 75°C temperature . Hence, the hot rod with less constrained 6-cm in diameter at constant 75°C was selected for this PER measurement. Two PER setups were verified and compared for measuring LC/UPC PMF connectors. The Polarized Laser Source (PLS) at 1550 nm wavelength and PER meter from OZ Optics were used in both setups, in which the measured connector was connected to PLS at 0° angle while the other end was connected to PER meter. In order to qualify our setups, the percentage of Repeatability and Reproducibility (%R&R) were tested and calculated. In each setup, the PER measurement was repeated 3 trials by 3 appraisers using 10 LC/UPC PMF connectors (5 LC/UPC PMF patchcords with 3.5±0.5 meters in length) in random order. The 1<sup>st</sup> setup, PMF was coiled at a larger 20-cm diameter for 3 to 5 loops and left in room temperature during the test. The 2<sup>nd</sup> setup, PMF was coiled around a hot rod at constant 75°C with 6-cm diameter for 8 to 10 loops for at least 5 minutes before testing. There are 3 ranges of %R&R acceptation guide line: <10% is acceptable, between 10% − 30% is marginal, and <30% is unacceptable. According to our results, the %R&R of the 1<sup>st </sup>PER test setup was 16.2% as marginality, and the 2<sup>nd</sup> PER test setup was 8.9% as acceptance. Thus, providing the better repeatability and reproducibility, this 2<sup>nd</sup> PER test setup having PMF coiled around a hot rod at constant 75°C with 6-cm diameter was selected for our next study of the impact of hot temperature on PER in LC/UPC PMF connector.
Bit Error Ratio (BER) dependence on received power was studied for 40Gb/s NRZ short optical fiber transmission, including a series of four low return loss (RL~21dB) and low insertion loss (IL~0.1dB) connections. The calculated power penalty (PP) was 0.15dB for BER~10<sup>-11</sup>. Although the fiber length was within DFB laser’s coherent length of ~100m and the multi path interference (MPI) value was 34.3dB, no PP of BER was observed. There was no PP due to low MPI probably because the polarization of the signal pulses were not aligned for optical interference, indicating that NRZ systems have a high resistance to MPI.
The fiber withdrawal of Group 4 (mated-thermal cycle) was observed up to 100 nm as in previous work<sup>1</sup>. We predict that this withdrawal is mainly caused by the impact of hot temperature (at 75ºC) based on GR-326<sup>2</sup> thermal cycle test profile repeated 21 cycles over 7 days; and thus, it was studies here for the purpose of reducing test time. All connectors were separated into four groups: 1) unmated-stored at room temperature, 2) mated-stored at room temperature, 3) unmated-stored at hot temperature, and 4) mated-stored at hot temperature. The hot temperature test was performed on Groups 3 and 4 for 1 hour, while Groups 1 and 2 was left at room temperature. The sample size of each group is 28 LC/UPC connectors. Radius of curvature, fiber height and apex offset were measured before and after that 1 hour. The fiber withdrawal up to 100 nm is found in Group 4 (mated-hot temperature), but no changes are observed in Groups 1-3. These results confirm the impact of hot temperature on fiber height, same as the thermal cycle test in previous work<sup>1</sup>. Afterward, Group 1-4 were unmated at room temperature for 1 day, 1 week, and 1 month. No significant change in fiber height is found. On the contrary, when Group 1-4 were re-tested as being mated at hot temperature for 1 hour, the fiber withdrawal up to 100 nm is now found in Group 1-3. However, the additional withdrawal up to 50 nm is still observed in Group 4.
The objective of this research was to investigate the impact of scratches on 40G NRZ optical link transmission and to understand if industry standards on connector scratches are sufficient for 40G applications.<p> </p>A sample set of twelve optical cable assemblies were prepared and characterized by TE Connectivity. Each assembly was a mated connector pair. The samples were divided into six groups with different levels of polishing scratches applied to vary the Return Loss (RL) at the mating interface. The measured RL across all groups ranged from a minimum of 26 dB to a maximum greater than 80 dB. Experimental links were built based on a 40G NRZ Bit Error rate Tester, EDFA with attenuator, cable assemblies and a fiber (maximum up to 3 km ITU-T G.652.D fiber). Bit Error Rate (BER) curve, Eye diagram, Jitter and Q-factor were measured for all experimental links. We found there was no significant change in any of the measured parameters for the links with different connector assemblies but with the same fiber length of optical link. The 40G NRZ link with multiple cable assemblies (up to 4 samples) and a fiber (1 to 3km) demonstrated significant robustness to the connector scratches.
Optimizing splicing of Erbium Doped Fiber (EDF) to Single Mode Fiber (SMF) is a critical requirement to maximize the
efficiency of Erbium Doped Fiber Amplifiers (EDFAs). This paper describes the key parameters which affect the splice
loss of EDF-SMF splices as well as the optimization process used to achieve 50% splice loss improvement. Before
performing the optimization process, the measurement system was validated with an evaluation including: laser stability,
detector linearity and Gage R&R (Repeatability & Reproducibility). The optimization of EDF and SMF splicing was
performed using a design of experiment with 2<sup>k</sup> factorial design and using MiniTab software for data analysis. A
commercially available fusion splicer was used. There were 53 parameters available for setting, They were selected and
divided into two groups. The first group included the parameters which might affect the splice loss and the second group
included the parameters which might affect the estimated splice loss. The optimization process for the first group of
parameters was performed until the target loss was met. The Arc1 Power and Arc1 Time were identified as the most
critical parameters for loss. Then the optimization process for the second group of parameters was performed until the
slope of the graph of estimated splice loss to actual splice loss was nearly one. This method reduced the average actual
splice loss from factory setting of EDF-SMF splicing (0.18 dB) to 0.10 dB and SMF-EDF splicing to 0.09dB. The
difference between estimated loss and actual loss was less than 0.05dB for either direction (measurements in the EDFSMF
and SMF-EDF direction). The proposed design of experiment can be used as a reference process to perform the
optimization of EDF to SMF splicing when Erbium Doped Fiber is changed to the other fiber types.