Computational modelling of defects in fibre optic cables

Abstract not available

Mathematical modelling: a laser soldering process for an optoelectronics butterfly package

For sensitive optoelectronic components, traditional soldering techniques cannot be used because of their inherent sensitivity to thermal stresses. One such component is the Optoelectronic Butterfly Package which houses a laser diode chip aligned to a fibre-optic cable. Even sub-micron misalignment of the fibre optic and laser diode chip can significantly reduce the performance of the device. The high cost of each unit requires that the number of damaged components, via the laser soldering process, are kept to a minimum. Mathematical modelling is undertaken to better understand the laser soldering process and to optimize operational parameters such as solder paste volume, copper pad dimensions, laser solder times for each joint, laser intensity and absorption coefficient. Validation of the model against experimental data will be completed, and will lead to an optimization of the assembly process, through an iterative modelling cycle. This will ultimately reduce costs, improve the process development time and increase consistency in the laser soldering process.

Parallel bandwidth characteristics calculations for thin avalanche photodiodes on a SGI Origin 2000 supercomputer

An important factor for high-speed optical communication is the availability of ultrafast and low-noise photodetectors. Among the semiconductor photodetectors that are commonly used in today’s long-haul and metro-area fiber-optic systems, avalanche photodiodes (APDs) are often preferred over p-i-n photodiodes due to their internal gain, which significantly improves the receiver sensitivity and alleviates the need for optical pre-amplification. Unfortunately, the random nature of the very process of carrier impact ionization, which generates the gain, is inherently noisy and results in fluctuations not only in the gain but also in the time response. Recently, a theory characterizing the autocorrelation function of APDs has been developed by us which incorporates the dead-space effect, an effect that is very significant in thin, high-performance APDs. The research extends the time-domain analysis of the dead-space multiplication model to compute the autocorrelation function of the APD impulse response. However, the computation requires a large amount of memory space and is very time consuming. In this research, we describe our experiences in parallelizing the code in MPI and OpenMP using CAPTools. Several array partitioning schemes and scheduling policies are implemented and tested. Our results show that the code is scalable up to 64 processors on a SGI Origin 2000 machine and has small average errors.