Comparative experimental study of different fetch distances, interpolation algorithms, source substrate noise, sampling point counts, and resolution for NF of BDFA

Abstract

Noise figure (NF) is a key metric for Fiber amplifier (FA), which affects the performance of signal transmission. The main factors affecting the NF are related to the amplified spontaneous emission (ASE) characteristics of the active fiber itself. However, during our study of NF, we found some easily overlooked factors that can have a significant impact on the results, such as the fetch distances and interpolation algorithm for calculating the ASE where the signal wavelength is located. Further, to date, we have also found little relevant research on whether the sampling characteristics of the instrument and the substrate noise contribute to the NF. On the other hand, bismuth-doped fiber amplifiers (BDFA) have developed rapidly in recent years. In this work, we use BDFA as an example, using different fetch distances, interpolation algorithms, substrate noise, and sampling variables (resolution and point counts), to comprehensively compare and analyze the impact caused by NF. The experimental results suggest that the distance between the sampling locations on either side of the signal and the center wavelength must exceed the signal’s bandwidth to obtain the actual NF; Linear interpolation and cubic interpolation are more suitable for calculating the NF; Even using a high power input signal of 0 dBm, the NF is only degraded by about 1.8% compared to the case where the substrate noise is ignored; In addition, the effect of different sampling point counts and resolutions have less influence on the NF. We believe this work provides a reliable reference value for calculating the accuracy of NF and will be useful for practical applications of active-fiber amplifiers.