Memristor enabled reconfigurable analogue systems
File(s)
Author(s)
Szypicyn, Jakub Mateusz
Type
Thesis
Abstract
The following thesis re-investigates the idea of performing analogue reconfiguration using gyrator-based blocks. By employing memristors, basic impedance synthesis is achieved, allowing to realise PCB-based tunable inductors and capacitors with values as high as 50H and 2nF respectively. Clear limitations of the proposed approach - instability, high noise and limited bandwidth are thoroughly investigated through the use of test circuits such as filters and waveform generators. Accurate mathematical models are then proposed, which describe the behaviour seen in experiments and simulations. The thesis describes the advantages and drawbacks of using memristors in conventional analogue circuits. As the technology is still relatively young, current state-of-the-art devices lend themselves perfectly well in DC application, while high-frequency such as RF still need time. To contrast the use of synthetic inductors and capacitors, other methods are proposed which highlight the particular difficulties in realising true analogue computing machines. Notwithstanding, a generic analogue platform design is proposed which leverages the memristive tuning behaviour and enables systematic reconfigurability on chip. To complement the platform design, several computational methodologies are put forward, which leverage the use of component physics as means of computation, contrary to current digital architectures which are based on logic, commonly characterised by large areal and power overheads. While, the proposed methods do not give a clear answer as to whether field programmable analogue arrays can be realistically designed and used to perform complex tasks, the argument leaves the reader with clearly stated research questions which ought be answered in order to fully understand the scope into which FPAAs can be employed. The work concludes with a description of areas which will benefit from analogue reconfigurable systems, such as space exploration and IoT.
Version
Open Access
Date Issued
2021-05
Date Awarded
2021-11
Copyright Statement
Creative Commons Attribution NonCommercial Licence
License URL
Advisor
Papavassiliou, Christos
Sponsor
Engineering and Physical Sciences Research Council
Publisher Department
Electrical and Electronic Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)