Browse Prior Art Database

A novel methodology for fitting data to multiple compact models with improved accuracy

IP.com Disclosure Number: IPCOM000199832D
Publication Date: 2010-Sep-17
Document File: 3 page(s) / 130K

Publishing Venue

The IP.com Prior Art Database

Abstract

The stringent demands placed on circuit design in today’s advanced technologies, coupled with the increasing difficulties of capturing all of these phenomena comprehensively in a single compact model, has forced designers to rely on more than one compact model for creating their designs. In this scenario, one or more compact models are developed from a given set of data and delivered to the designer, which he (she) uses in his designs. In this case, it is important that the models be consistent with each other as well as with the data. The currently used modeling approaches are not equipped to handle the case of single data and multiple compact models as, in the current approach, the multiple models are developed and optimized independently of each other. As a result, it happens that even though each model may agree well with the data, the different models may not be in good agreement with each other. Disclosed in this embodiment is a novel compact modeling methodology which will ensure agreement not only between each model and the data, but among the different models as well.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 3

A novel methodology for fitting data to multiple compact models with improved accuracy

    Disclosed in this embodiment is a novel methodology for fitting data to multiple compact models with improved accuracy. In this embodiment the term 'data' refers to any collection of measurements obtained from a physical system, while a 'compact model' is a representation of the behavior of that system with the help of mathematical and/or physical equations. Compact models are used in engineering in order to mimic the physical behavior of a system in order to understand and design more complex systems. An example of the use of compact models is in the domain of semiconductors, where the characteristics of a transistor, which is the fundamental building block, is measured, yielding data, and the data is mapped on to a compact model. This compact model is then used to design more complex building blocks called circuits, which are in turn used to design semiconductor chips.

    Accurate semiconductor compact models are key to producing successfully working circuits and chips. As semiconductor technology becomes more advanced, accurate modeling of semiconductor phenomena is becoming increasingly challenging. In order to meet this challenge, a large number of solutions, i.e. compact models, have emerged from industry as well as academia, each claiming the advantage of accuracy of use.

N

 erfect, and each one has its own strengths and weaknesses. This fact, coupled with the increasingly stringent demands placed on circuit design in order to ensure a successful product, has forced designers to rely on more than one compact model for creating their designs. In this scenario, two or more compact models are used by designers in order to create their circuits. The compact models are created from measurements (data) as explained above, and as stated above, it is important that the models be as close to the data as possible, i.e the error between the data and the models be as small as possible.

    The currently used modeling approaches are not equipped to handle the case of single data and multiple compact models as, in the current approach, the multiple models are developed and optimized independently of each other. As a result, it happens that even though each model may agree well with the data, the different models may not be in good agreement with each other. Figure 1 is an illustration of the currently employed methodology matching the models independently to the data.

1

o compact model, however is

p

Page 2 of 3

Data

Model1 Model2

Link

Missing

Figure 1: Currently used approach

    Disclosed in this embodiment is a novel compact modeling methodology which will ensure agreement not only between each model and the data, but among the different models as well. In this approach...