Alignment and Performance Considerations for Capacitive, Inductive, and Optical Proximity Communication

Abstract

We present a comparative analysis of different physical approaches to chip-to-chip proximity communication, PxC, based on capacitive, inductive and optical signalling. Each method is modeled theoretically and the tolerances for packaging are identified. Analytical formulas for performance in terms of the pad size and pad spacing are derived and compared to reported experimental data. The tolerance of each communication method to misalignment is reported. The design space in terms of channel density and chip separation for capacitive and inductive proximity communication is explored for a specified bit-error-rate (BER) or signal-to-noise ratio (SNR) and transmitter power or voltage. The relative merits of each technology are discussed. A general conclusion is that capacitive proximity communication is advantageous for dense communication with small pads at low voltages and when low raw bit-error rates are required; however a hard requirement for vertical separation between chips is identified, independent of the area of the pads, and fixed by the supply voltage and the technology parameters. On the other hand, inductive communication provides a larger working range of chip separations, and is advantageous when larger pad sizes are used; however the minimum voltage is similarly constrained in order to maintain low bit-error rates. Optical proximity communication potentially provides the largest chip separations, but has low tolerance to in-plane misalignment.