Symbolic analysis for large scale networks

number: 
1850
إنجليزية
Degree: 
Author: 
Firas Abdullah Thweny Al-Saidi
Supervisor: 
Dr. Fawzi M. Al-Naima
year: 
2000
Abstract:

The known techniques for symbolic analysis (matrix, signal-flow-graph, tree enumeration, parameter extraction, and interpolation methods) suffer from several limitations, like the limitation in symbolic terms and circuit size. The presented technique overcomes these limitations. It is proved that the presented technique produces exact symbolic terms for any circuit size and any number of symbolic elements. In this thesis, a new hybrid technique for evaluating the transfer function of large networks symbolically is presented. It is based on mixing the two-graph modified nodal formulation besides the determinant properties of the closed system matrix. The presented technique leads to an exact sorted and compact symbolic transfer function with single expansion of the closed loop system matrix determinant. To verify this technique practically, a software package based on the rules and steps supported by the technique; is created under windows environment. This software package named "SAPLN" from "Symbolic Analysis Package for Large Networks" evaluates the symbolic transfer function for any electrical circuit composed from the fundamental elements (passive and active) exactly with reasonable execution time. SAPLN is created using the Borland C++ Builder Environment, supported under Windows operating system. This software package supports the user with the basic electrical elements from passive, to active like, operational and transconductance amplifiers, nullors, current conveyers, dependent sources, low and high frequency bipolar and FET transistors. All these elements 'guide the-user to rebuild other custom elements using the powerful library supported by the package. Beside the supported library, SAPLN supports user interface analysis tools: Text editor, schematic editor, frequency response evaluator and plotter. As a result, the user can evaluate all four types of transfer functions for any network symbolically. Networks with different sizes and different configurations were tested using the presented technique evaluated by SAPLN. The results were encouraging, especially in the number of generated transfer function terms (9226593 distinct terms for 17 stage RC ladder). Due to unavailability of any complete symbolic analyzer, the frequency response evaluator supported by SAPLN enables the verification of the results (by numerical substitution of all symbolic elements). The results are compared with the well-known PSPICE numerical simulation package. The results were found identical.