模拟VLSI电路和原则

This book presents an integrated circuit design methodology that derives its
computational primitives directly from the physics of the used materials and
the topography of the circuitry. The complexity of the performed computations
does not reveal itself in a simple schematic diagram of the circuitry on the
transistor level, as in standard digital integrated circuits, but rather in the
implicit characteristics of each transistor and other device that is represented
by a single symbol in a circuit diagram. The main advantage of this circuitdesign
approach is the possibility of very efficiently implementing certain
‘natural’ computations that may be cumbersome to implement on a symbolic
level with standard logic circuits. These computations can be implemented
with compact circuits with low power consumption permitting highly-parallel
architectures for collective data processing in real time. The same type of
approach to computation can be observed in biological neural structures, where
the way that processing, communication, and memory have evolved has largely
been determined by the material substrate and structural constraints. The data
processing strategies found in biology are similar to the ones that turn out to
be efficient within our circuit-design paradigm and biology is thus a source of
inspiration for the design of such circuits.
The material substrates that will be considered for the circuits in this
book are provided by standard integrated semiconductor circuit technology and
more specifically, by Complementary Metal Oxide Silicon (CMOS) technology.
The reason for this choice lies in the fact that integrated silicon technology
is by far the most widely used data processing technology and is consequently
commonly available, inexpensive, and well-understood. CMOS technology has
the additional advantages of only moderate complexity, cost-effectiveness, and
low power consumption. Furthermore it provides basic structures suitable for
implementation of short-term and long-term memory, which is particularly important
for adaptive and learning structures as found ubiquitously in biological
systems. Although we will specifically consider CMOS technology as a physical
framework it turns out that various fundamental relationships are quite
similar in other frameworks, such as in bipolar silicon technology, in other
semiconductor technologies and to a certain extent also in biological neural
structures. The latter similarities form the basis of neuromorphic emulation of
biological circuits on an electrical level that led to such structures as silicon
neurons and silicon retinas.