A method based on cutset analysis is presented to compact the network matrices written in the modified nodal formalism. Operational amplifier models with finite as well as infinite gain have been used. Simple algorithms for frequency response and sensitivity analyses based on these compact MNA equations are presented. The algorithms have been tested on a small 8086 based system.

An algorithm to reduce the size of MNA matrices for switched and nonswitched networks is proposed through the use of cutsets. The procedure is simple and eliminates all diagonal zeros. Closed switches and EMF sources are handled identically. The final size of the matrix may be reduced to the number of nodes when no current-controlled elements are present.

The fixed-point error analysis of long length RT's is presented assuming sign-magnitude or 1's complement arithmetic. The errors introduced by the input-addltion, multiplication and the output-addition stages are separately considered in each dimension and are suitably combined to give the average output noise variance. The various error components signifying the characteristics error parameters of a module are obtained for each short length RT. Filters of different transform lengths are chosen for illustration. The theoretical noise variance obtained in each case is found to have excellent agreement with the corresponding simulated results. The noise performance of RT and FFT filters is also compared. It is seen that the input noise has worsened by a factor of 3.8 bits in passing through an RT filter compared to 6.15 bits as in case of FFT filter.

Recently, Patterson and McClellan [1] have carried out the fixed-point error analysis of winograd Fourier transform algorithm assuming noise sources to be uncorrelated and have derived expressions for output NSR (noise to signal ratio). Such an assumption is not quite valid. A theoretically revised expression without making such an assumption has been obtained in case of Winograd short-length DFT algorithms. This involves the computation of correlation coefficients among noise sources. The revised results are verified by computer simulations and are found to be in close agreement with those predicted by Patterson and McClellan. It is, in general, observed that the basic modules introduce errors higher than those of FFT and require maximum of half a bit to maintain the same degree of precision.