Using Neural Networks to Design Transistor Amplifier Circuits
This paper is an extension of previous work that addressed the application of bipolar transistor amplifier design using neural networks. That work addressed the design of common emitter amplifiers by first mathematically determining specific output parameters from a large selection of biasing resistors. Once the outputs had been determined, a neural network was trained, using the aforementioned results as inputs and the biasing resistors as outputs. This was initially performed with ideal emitter bypass capacitors, but was then followed-up by employing several non-ideal capacitors, making it much more interesting and useful. This paper focuses on the common collector and the common base configurations. Bipolar junction transistor amplifier parameters often include voltage gain, input impedance, output impedance, and the voltage difference between the collector and emitter. These will be addressed in this paper as before. There are several methods that can provide a suitable solution for each design, however the objective of this work is to indicate which external resistors are necessary to yield useful results by employing neural networks.
