Abstract
As wearable technology continues to evolve, the demand for seamless, intelligent, and unobtrusive integration of electronics into everyday garments becomes increasingly urgent. E-textile platforms offer a promising pathway toward this vision, enabling real-time sensing, computation, and communication directly within fabric-based systems. However, practical e-textile deployment is hindered by several challenges: limited power availability, rigid form factors of conventional electronics, integration difficulties, lack of reconfigurability, and limited scalability. Addressing these challenges requires the development of ultra-low power, compact, and mechanically compliant subsystems tailored for textile integration. This dissertation presents a collection of circuit and system-level innovations that collectively aim to enable scalable, energy-efficient, and comfortable e-textile platforms.
The first set of contributions focuses on sensing and data acquisition for energy constrained e-textiles. A 1.2 µW reconfigurable 10-Channel capacitance-to-digital converter is developed, featuring always on OTA-free architecture (replaced with duty-cycled buffer), programmable sensing range, and per-channel control to reduce power consumption. An on-chip offset correction and baseline calibration scheme further eliminates unnecessary digitization, improving conversion accuracy. A second contribution presents an end-to-end photoplethysmography sensing system that includes a custom analog front-end, LED driver, SoC, and BLE transmitter. The proposed custom end-to-end system reduces over all power consumption compared to existing COTS(commercial off-the-shelf)-based systems. The system also offers programmable duty-cycling of sub-components to further reduce system power. The next set of contributions focuses on addressing the area constraint challenges in e-textile systems. For data-intensive e-textile applications such as on fabric image capture, this work adopts a hybrid interconnect strategy—employing global I2C for reduced wiring complexity while enabling local integration of high-density SPI flash through a custom 0.36-mm² I2C-to-SPI converter. The converter performs on-the-fly protocol conversion without internal data buffers or clock generation, minimizing area and power overhead. Complementing these sensing and memory subsystems, a 0.36 mm² tiny audio compressor SoC is introduced for in-textile voice data processing. It supports 4× compression and integrates with braided yarn structures for acoustic sensing and recording. Finally, the dissertation demonstrates a monolithic haptics controller chiplet for yarn-level integration. This device features dual-channel touch inputs, impedance sensing, and an on-chip DC-DC converter to deliver distinct tactile feedback patterns based on input type—all while maintaining garment-level flexibility and comfort. Together, these contributions demonstrate how application-specific design at both the circuit and system levels can overcome energy and area constraints, bringing e-textiles closer to comfortable, long-lasting, and interactive intelligent garments.
