Analog Circuit Topology Development: Practice methods for technology and teaching based on comprehensible transistor models
Doctoral thesis, 2007
To obtain adequate knowledge for design of robust analog circuits in industry, circuit engineers gain main knowledge from own practice experience, because theory taught by universities rarely states validity limits and result spreads. Therefore, when going from university to industry, the engineers find this theory nontrustworthy and relinquish it. Starting the practice learning from scratch, their learning time can be tens of years. There are, therefore, needs for effective work and learning methods in practice, in the area of analog circuit design.
In this dissertation, I address these method needs. I begin with classifying and describing the required knowledge for circuit engineers by using Aristotle's knowledge taxonomy. I point out and discuss the rather different methods in the industry in contrast to universities. The conclusion is that both the industry and universities need to adapt their work and teaching methods, respectively, for the engineer to get a more unified and effective learning situation, enabling continuous learning throughout the career. I propose a concrete work method --- the "expectation method" --- by which the engineer learn practice experience effectively. It is made to be used in everyday design work (in the industry) and in course laboratory exercises (in universities).
From technical point of view, a circuit engineer needs knowledge in two main areas: component behaviors and circuit performance. For understanding of component behaviors, I present a method how to develop comprehensible behavior models for electronic components. Since transistors have proven to be difficult to fully understand for circuit engineers, I have developed two new transistor behavior models by using this method: the SLT-FET~model for field-effect transistors and the SLT-BJT~model for bipolar transistors. When finalized, these models should include parameter tolerances and equation validity limits, which I believe are vital for effective learning of practice experience.
To improve efficiency of circuit-synthesis work and learning, I present a work method for analog-circuit topology development in practice. This work method helps circuit engineers to find their own ways to more structured design methods, including development of circuit topologies.
During the evaluation of this work method, I designed several circuits in different analog design areas. In addition to the evaluation itself, I have obtained design methods for: linear voltage regulators in general, a fully-differential op-amp topology, and an LNA topology with exceptional dynamic range.
I developed a five credit-point graduate course in analog circuit design in practice, and delivered it to industry engineers and PhD students. Four labs and a design project constituted the core of this course. Each lab and the project were performed according to the expectation method. In the course evaluation interviews, the industry engineers gave comments like: "The course gave knowledge in another way than other courses; it gave experience". To work and teach according to this method can bridge the gap of knowledge and methods between the industry and universities today.
HC1, Hörsalsvägen 14, Chalmers
Opponent: Professor Erik Bruun, Ørsted DTU, ElectroScience,Technical University of Denmark