This master's thesis addresses the measurement of first-order polarisation mode dispersion (PMD) and its parallel second-order component, polarisation-dependent chromatic dispersion (PCD), in passive erbium-doped optical fibres (EDF). The erbium absorption resonance around 1530 nm severely degrades the signal-to-noise ratio, precluding reliable measurements in the telecommunication C-band, so the measurements are performed in two spectral windows outside the absorption region: the O-band (1240 to 1400 nm) and the extended L-band (1570 to 1715 nm). The wavelength-scanning fixed-analyser method is used, extended by cyclic scanning in which an automated polarisation controller sets a new random input polarisation state before each of the 100 measurement cycles. Three EDF samples of 10 m, 20 m and 30 m length are characterised.
For the O-band measurements, a dedicated broadband source based on a superluminescent diode (SLED) was designed and built in the laboratory, comprising a precision current source and a temperature stabilisation system with a current mirror supplying the internal thermistor, a bipolar current driver for the thermoelectric cooler, and a proportional-integral controller. Data acquisition and processing are fully automated in Python, from controlling the polarisation controller and the optical spectrum analyser to spectrum smoothing, automatic threshold selection for extrema detection using a stable-plateau algorithm, and the computation of final PMD values with an adaptive number of sub-bands.
The measured mean differential group delay (DGD) ranges from 0.175 to 0.541 ps in the O-band and from 0.264 to 0.820 ps in the extended L-band, with the corresponding PMD coefficient ranging from 1.75 to 3.12 and from 2.64 to 4.73 ps/√km, respectively. The PCD_rms value grows with fibre length and is consistently higher in the extended L-band, reflecting the increase of birefringence with wavelength in heavily doped fibres. The central finding of the thesis is that the DGD in the extended L-band was successfully predicted from O-band measurements by first-order Taylor extrapolation based on the measured PCD slope. Although this first-order approximation neglects higher-order dispersion effects across a wide spectral gap, it proves to be a valid engineering heuristic, with agreement with direct measurement remaining within 6 % for all three lengths.
|