電力取舍和低功耗的模擬CMOS集成電路

The enormous rise of digital applications in the last two decades arouses the
suggestion that analog techniques will lose their importance. However in applications that work with digital signals analog techniques are still very important for a number of reasons. First the signal that must be processed or stored may be analog at the input and output of the system. Second when digital circuits must operate at high speed the analog behavior becomes
important again. And third when only limited bandwidth and signal to noise ratio is available the theoretical maximum data rate is determined by Shannon’s law. This theoretical limit can only be approximated in practice when complex modulation schemes are used, and after this modulation process the signal is analog again. Of course this does effect the tremendous advantage of digital signals compared to analog signals. Where analog signals deteriorate every time they are processed or stored, digital signals can be recovered perfectly when they
are tailored to the properties of the system they are used for. The accuracy of digital signal processing is only limited by practical constraints and many digital signals can be compressed very effective so that after compression they use less bandwidth then their analog counterparts.
In any aplication there will thus be analog and digital parts and often the choice has to be made if an analog or a digital solution is preferred for a certain function. It is often very difficult to give a founded opinion to this because there are many factors that play a role in this decision. This book tries to bring up methods and information that can help you in making the right choices. Things like the fundamental and practical limits of power dissipation for a specified accuracy and signal to noise ratio are discussed. Other factors like
the process choice and related with that the chip area, the maximum supply voltage and the number of pins and external component also play an important role in the choice between analog and digital. Of course if a chip is mainly digital it becomes more difficult to integrate high quality analog circuit first because the digital cross-talk will interfere with the analog signal, second because the process choice will be based on the digital properties.
Where many different processes are needed for different state of the art analog applications, CMOS processes that are optimized for digital applications have a tendency to go through comparable technical and lithographical improvements in different companies. Because the economy of these processes is much more determined by feature size than in analog processes, the huge drive for improvement. As smaller feature sizes in digital circuits also means lower energy dissipation per function more and more circuits that are still analog now
will become digital in the future. Every new process generation the digital function gets smaller chip area and lower energy dissipation while the analog counterpart may even become worse because of a lower supply voltage, more noise and less linearity. The only analog parameter that still improves with every new CMOS generation is speed.
This book starts with the theoretical foundation of the power considerations in sub-micron digital and analog circuits. Next, as there are many circuit solutions that can fulfil a certain specification, classes of circuits are discussed in order to help the reader in making the right initial choices. Then a number of circuits that are realized by the author are discussed in detail and the measurement results are presented.
I think that this book can help the mixed signal designer to make the right choices on architectural and circuit level.