This work proposed a Design Methodology for Manufacturing (DFM) using Geometric Programming (GP) and Newton-like methods to solve non-linear optimization problems, which define and aid the design of analog circuits. Afterwards, this methodology is applied and validated through the design of a voltage reference circuit.Over the last years, the tendency of the increasing on the transistor density predicted by the Moore Law has turned the circuit design problem dimensionally more complex. Additionally, a higher transistor density implies shrinkage on the feature dimensions of the process making it more sensitive to the process variations and environmental conditions. The differences between the designed circuit and the tested one give an evidence of yield losses, which are attributed in a great proportion to the design process. Due to the high responsibility of the designer on this problem, the analog design must be focused on new approaches that jointly manage performance and yield.In first place, the proposed methodology obtain a initial point with a suitable set of performance specifications, which will be used to analyze the impact of the mismatch and process variation over the design specifications. Once the statistical and deterministic behavior of the circuit was characterized, a new yield improvement strategy is implemented using Geometric Programming. Attempting to obtain a design with a set of high performance specifications directly involves the circuit yield, because an optimal performance set obtained by the traditional framework of GP does not assure the obtaining of a marketable and competitive design. So, this works establish a design method that combine the advantage of obtaining global optimum in Geometric Programming with a new mismatch and worst-case analysis that enabled a reduction in their computation time and maintain the initial nominal performance values. Using the design methodology for manufacturing proposed in this work, a voltage reference design with 37% better yield than one obtained with a typical design strategy without any significant penalty on their performance specs was achieved.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2025 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.