|Abstract: ||This Thesis presents research on ultra low-power bioelectronics, focusing on purely ana- logue methodologies and techniques stemming from the systematic, in-house developed, transistor-level synthesis framework, termed the “Nonlinear Bernoulli Cell Formalism” (NBCF). The NBCF is capable of converting coupled, nonlinear biological differen- tial equations into coupled, nonlinear electrical differential equations and subsequently into ultra-low power log-domain electrical circuits, or “Cytomimetic Circuits”. The establishment of the aforementioned framework is achieved through the example re- alisation of several, novel, non-linear, multi-dimensional, cytomimetic topologies, that mimic a variety of biological function dynamics. The constructed topologies deal with increased complexity emanating from the high-dimensionality, the highly inter-coupled state variables and parameters, and the multitude of the potentially resulting dynami- cal behaviours. Their effective construction utilises design techniques in weak-inversion, log-domain bioelectronics, specialising in the analysis and synthesis of static and dy- namic translinear circuits. Moreover, statistical fabrication variability Monte Carlo analyses investigate the robustness of the proposed topologies. Subsequent low-level layout, post-layout simulations and fabrication of one of the proposed topologies (mam- malian cell cycle) lead to the production of novel proof-of-concept measured chip results. The measured results validate the feasibility of the NBCF.
Further to the establishment of the NBCF and its real-life feasibility, this Thesis serves to suggest area and performance optimisation techniques for widely-used static and dynamic translinear circuits which are pivotal blocks for our cytomimetic designs. Ini- tially, the area optimisation is outlined, by proposing an area minimisation path for the research group’s cochlear implant design. Novel circuit design techniques are imple- mented, in order to reduce the size of log-domain cochlear implant processors. Moreover, performance optimisation is performed by articulating a methodical, step-by-step sym- bolic analysis of a variety of common, static and dynamic, translinear topologies in weak-inversion electronics. The analysis relies upon a simplified EKV-based approxi- mation. The impact of transistor-level design parameters upon performance and ideal behaviour is captured, analysed and discussed in a quantitative and qualitative manner. Finally, optimisation approaches are detailed whereby novel design rules are provided for popular translinear topologies.|