Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/12373
Authors: Singh, Gurbax
Issue Date: 1976
Abstract: In the mass production of circuits where 100 per cent yield is not required, but an estimate of yield based on individual component behaviour assists in production decisions the tolerance analysis becores the final phase of circuit design. Practically it is very difficult to have the precise valued elements. Mora the precision, higher is the cost, so to reduce the overall circuit cost, the slackness in element values is allowed. But higher the slackness, more is the probability of circuit response to violate the specified limits i. e . , the circuit fails. So to avoid. circuit failure, tolerance analysis is required. Ascot and the circuit response are the functions of tolerance, we optimise the cost of the circuit subject to circuit respond constraints. In this study two itrative algorithms are presented for the tolerance assignment in a given frequency domain. The first algorithm is for continuous tolerance case and the second is for the discrete tolerance case. The presented algo-rithms can be used for both types of circuits whose element values are either correlated or uncorrelated and has the feature of allowing the designer to specify a circuit yield of less than.100 per cent and can handle any type of probability distribution. Two problems have been considered. The first problem is of a band pass filter circuit to illustrate the case of uncorrelated parameters for both continuous and discrete element tolerance assignment with normal distri-bution. And the second problem is of a three transistor low pass ampl3.fier used to illustrate the case of integrated circuits, i.e., with correlated parameters,
Other Identifiers: M.Tech
Research Supervisor/ Guide: Sharma, J. D.
Vasantha, M. K.
metadata.dc.type: M.Tech Dessertation
Appears in Collections:MASTERS' THESES (Electrical Engg)

Files in This Item:
File Description SizeFormat 
EED175342.pdf3.72 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.