Design rules for highly parallel free-space optical interconnects

Recently, a number of successful free-space chip-to-chip and board-to-board optical interconnects have been demonstrated. Here, we present some of the design rules that can be derived as a result of this work and also as a result of numerical and theoretical analyzes. We draw a number of conclusions. In the area of optoelectronic very large scale integration (VLSI) design, we suggest that differential electrical and optical transceiver designs provide the best performance. In the area of optical design, we present scaling and system partitioning laws for clustered optical relays and determine the interconnect distances at which microlens or macrolens systems are more suitable. We also show that the ease with which two modules can be aligned can be related to the optical invariant of the system and is, thus, a function of the size of the detector and the numerical aperture of the detector optics. Finally, we show that when multiple optical components must be aligned, very high individual component tolerances are required if the system as a whole is to have a high chance of success.