Page 50 - The Art and Science of Analog Circuit Design
P. 50

Keitaro Sekine


        to cope with this situations. (The reason why they have been decreased
         shall be mentioned in later section of this paper.) But unfortunately it is
         believed that there should be no such a powerful analog CAD system like
         a digital for a while.
           Analog circuits design technologies have following features which
         prevent us from realizing unified approach schemes:

            1. While digital systems can be described with a couple of logic
         equations in principle, specifications of analog circuits are too much
         complicated to describe in a clear format. For instance, it sometimes is
         requested to design "excellent sound quality HiFi amplifier." We have no
         definition for "excellent sound quality" at all. It depends on individual
        judgment, some feels good the others feels no good, listening to the
         same amplifier. Besides a feeling judgment, amplifier has many charac-
         teristic items such as gain, frequency characteristics, dynamic range,
         distortion, temperature characteristics, input and output impedance,
         power consumption and so on. And normally we could not find evident
         correspondence between these characteristic items and the total perfor-
         mance.

            2. Several specifications on a single circuit usually conflict each
         other, so many trade off should be indispensable during the design proce-
         dure, taking restrictions such as performance of devices available, cost,
         deadline etc. into account. As these compromises could be done with the
         designer's personal experience and knowledge, there was no straightfor-
         ward scheme to do them. There were many papers with respect to the
         optimization of electronic circuits, but difficulties are not in how to do it
         but in where one should place the goal.

            3. To design a good analog circuits, a step by step method is quite
         insufficient and a breakthrough should be mandatory. Only man of talents
         can do that. But perhaps he cannot explain how he comes to the break-
         through.
            4. There are many circuit topologies and their combinations to real-
         ize the same specification. It should be so difficult for CAD to get a
         unique solution.

           Above mentioned features of analog circuits design are based on very
         essential characteristics of analog. We can not write any program without
         the knowledge about how it works. We think "computer-automated-
         design" of analog circuits are still one of challenging problems for us.
           We have, however, powerful tools for analog circuit design, a circuit
         simulator. Among them "SPICE" and its derivatives are widely used
         by the design engineers. It is very useful as far as he use as literally
         "computer-aided-design" tools. Circuit simulator requires good under-
         standing of circuits from the design engineer. We discussed about mer-
         its/demerits of using circuit simulator in the National Convention of
        IEICE in 1992 to find the following problems:

                                                                                        33
   45   46   47   48   49   50   51   52   53   54   55