High-speed Optical Interconnects in harsh environments

Licentiatavhandling, 2024

Advancements in AI, machine learning, AR/VR, IoT, and cloud computing, along with the anticipated arrival of 6G, are driving innovations like telepresence and autonomous machine communication. These technologies demand robust data center infrastructure to manage exponentially growing data traffic. The speed of data centers depends on their interconnects, and for short-reach connectivity, optical interconnects (OIs) with VCSELs are preferred due to their high speed, compact size, and energy efficiency. However, when tested with real-world data, these systems often experience additional receiver sensitivity penalties, which remain insufficiently addressed in prior studies. Additionally, while co-packaging the transceivers which enhances system capacity, it subjects components to harsh environments, underscoring the need for advanced optical interconnects that maintain performance across wide temperature ranges. This thesis examines the operation of optical interconnects in real-world scenarios. A comprehensive study on Pseudo-Random Binary Sequences (PRBS) examines longer patterns that closely resemble real data, addressing additional receiver sensitivity penalties associated with longer sequences through a line coding technique. The study, conducted through simulations and experiments, uses a directly modulated VCSEL as the transmitter, a photodiode for direct detection, and multimode fiber for transmission. Results demonstrate 50 Gbps error-free transmission with a PRBS-31 sequence and 50-Gbps error-free transmission using PRBS-15 at 50°C when line coding is applied. The effectiveness of line coding increases under higher system stress, such as elevated data rates and temperatures. Additionally, improved VCSEL Quantum Well (QW) designs show robust performance across a broad temperature range, achieving 25-Gbps error-free transmission at 140°C. These findings are essential for advancing interconnect systems capable of handling real data in harsh environments.