The rapid expansion in data communication due to the increased multimedia applications and cloud computing services necessitates improvements in optical transceiver circuitry power efficiency as these systems scale well past 10 Gb/s. In order to meet these requirements, a 26 GHz transimpedance amplifier (TIA) is presented in a 0.25-um SiGe BiCMOS technology. It employs a transformer-based regulated cascode (RGC) input stage which provides passive negative-feedback gain that enhances the effective transconductance of the TIA's input common-base transistor; reducing the input resistance and pro- viding considerable bandwidth extension without significant noise degradation or power consumption. The TIA achieves a 53 dBOhm single-ended transimpedance gain with a 26? GHz bandwidth and 21.3 pA/H z average input-referred noise current spectral density. Total chip power including output buffering is 28.2 mW from a 2.5 V supply, with the core TIA consuming 8.2 mW, and the chip area including pads is 960 um x 780 um.
With the advance of photonic devices, optical interconnects becomes a promising technology to replace the conventional electrical channels for the high-bandwidth and power efficient inter/intra-chip interconnect. Second, a silicon photonic transceiver is presented for a silicon ring resonator-based optical interconnect architecture in a 1V standard 65nm CMOS technology. The transmitter circuits incorporate high-swing drivers with non-linear pre-emphasis and automatic bias-based tuning for resonance wavelength stabilization. An optical forwarded-clock adaptive inverter-based transimpedance amplifier (TIA) receiver trades-off power for varying link budgets by employing an on-die eye monitor and scaling the TIA supply for the required sensitivity. At 5 GB/s operation, the ring modulator un- der 4Vpp driver achieves 12.7dB extinction ratio with 4.04mW power consumption, while a 0.28nm tuning range is obtained at 6.8uW/GHz efficiency with the bias-based tuning scheme implemented with the 2Vpp transmitter. When tested with a wire-bonded 150f- F p-i-n photodetector, the receiver achieves -12.7dBm sensitivity at a BER=10-15 and consumes 2.2mW at 8 GB/s.
Third, a novel Nano-Photonic Network-on-Chip (NoC) architecture, called LumiNoC, is proposed for high performance and power-efficient interconnects for the chip-multi- processors (CMPs). A 64-node LumiNoC under synthetic traffic enjoys 50% less latency at low loads versus other reported photonic NoCs, and ~25% less latency versus the electrical 2D mesh NoCs on realistic workloads. Under the same ideal throughput, LumiNoC achieves laser power reduction of 78%, and overall power reduction of 44% versus competing designs.