Citation
Abstract
Two commercially available large-area silicon avalanche photodiodes (APDs) were characterized in the laboratory. The response of the APDs to a sequence of 8-bit pulse-position modulated (256-PPM) laser pulses, with and without additive background noise, was recorded and stored for post-analysis. Empirical probability density functions (pdf’s) were constructed from the signal and noise slot data and compared to pdf’s predicted by an analytical model based on the sum of Webb and Gaussian (Webb+Gaussian) statistics. The pulse sequence was used to generate bit-error rate (BER) versus signal photons-per-pulse plots. These BER measurements were also compared with analytical results obtained by using the Gaussian and Webb+Gaussian models for APD channel statistics. While the 532-nm APD data showed reasonable coherence with theory, more discrepancies were evident in the 1064-nm APD case. It is surmised that these discrepancies were caused by inaccurate knowledge of detector parameters as well as radio frequency interference. A simple slot synchronization loop was also tested on the laboratory-generated detector output data, resulting in little to no loss upon use of the tracking-loop-generated timing in calculation of bit-error probabilities.
Keywords
Details
- Volume
- 42-146
- Published
- August 15, 2001
- Pages
- 1–14
- File Size
- 236.6 KB