## Efficient approximation of symbolic expressions for analog behavioral modeling and analysis (2004)

Venue: | IEEE Trans. Computer-Aided Design Integr. Circuits Syst |

Citations: | 1 - 1 self |

### BibTeX

@ARTICLE{Tan04efficientapproximation,

author = {Sheldon X. -d. Tan and C. -j. Richard Shi and Senior Member},

title = {Efficient approximation of symbolic expressions for analog behavioral modeling and analysis},

journal = {IEEE Trans. Computer-Aided Design Integr. Circuits Syst},

year = {2004},

volume = {23},

pages = {907--918}

}

### OpenURL

### Abstract

Abstract — Efficient algorithms are presented to generate approximate expressions for transfer functions and characteristics of large linear analog circuits. The algorithms are based on a compact determinant decision diagram (DDD) representation of exact transfer functions and characteristics. Several theoretical properties of DDDs are characterized, and three algorithms, namely, based on dynamic programming, based on consecutive k-shortest path based, and based on incremental k-shortest path, are presented in this paper. We show theoretically that all three algorithms have time complexity linearly proportional to |DDD|, the number of vertices of a DDD, and that the incremental kshortest path based algorithm is fastest and the most flexible one. Experimental results confirm that the proposed algorithms are the most efficient ones reported so far, and are capable of generating thousands of dominant terms for typical analog blocks in CPU seconds on a modern computer workstation. Index Terms — analog symbolic analysis, circuit simulation, determinant decision diagrams, matrix determinant, behavioral modeling I.

### Citations

8982 |
Introduction to Algorithms
- Cormen, Leiserson, et al.
- 1999
(Show Context)
Citation Context ...ortest (weighted) path in a coefficient DDD, which is a DAG (direct acyclic graph), can be found by depth-first search in time O(V + E), where V is the number of DDD vertices and E is number of edges =-=[28]-=-. So it is O(V ) in DDDs. Once we find the shortest path from a DDD, we can subtract it from the DDD using SUBTRACT() operation [21], and then we can find the next shortest path in the resulting DDD. ... |

73 |
Sansen, Distortion Analysis of Analog Integrated Circuits, 1st ed
- Wambacq, M
- 1998
(Show Context)
Citation Context ... GmbVbs testability analysis and yield enhancement [10]. Efficient symbolic techniques for linear(ized) analog circuit analysis are the basis of distortion analysis of weakly nonlinear circuits [11], =-=[12]-=-, symbolic modeling of time-varying systems such as mixers [13]. Previous attempts to generate interpretable expressions use various symbolic analysis methods to generate sum-of-product representation... |

55 |
Symbolic Analysis for Automated Design of Analog Integrated Circuits, Kluwer Acad
- Gielen, Sansen
- 1991
(Show Context)
Citation Context ...functions until errors are within the user-specified error bounds for the frequency range. M3 M4 M6 Vin1 Vin2 Cc M1 M2 Vbias M5 Fig. 1. A simplified two-stage CMOS opamp [8]. M7 VDD As illustrated in =-=[9]-=-, simple yet accurate symbolic expressions can be interpretable by analog designers to gain insight into the circuit behavior, performance and stability, and are important for many applications in cir... |

40 |
Symbolic analysis methods and applications for analog circuits: a tutorial overview
- Gielen, Wambacq, et al.
- 1990
(Show Context)
Citation Context ...ONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, VOL 23, NO. 4, APRIL 2004 2 g Cgd GmVgs Cgs Cbg Fig. 2. MOSFET small signal model. d GmbVbs testability analysis and yield enhancement =-=[10]-=-. Efficient symbolic techniques for linear(ized) analog circuit analysis are the basis of distortion analysis of weakly nonlinear circuits [11], [12], symbolic modeling of time-varying systems such as... |

27 |
A unified approach to the approximate symbolic analysis of large analog circuits
- Yu, Sechen
- 1996
(Show Context)
Citation Context ... generated in the final stage (as the model is recentered after every stage). Such nominal point approximation strategy used in this work is also adopted by most of symbolic approximation methods [1]–=-=[7]-=-. During the approximation, we monitor both magnitudes and phases of the transfer functions until errors are within the user-specified error bounds for the frequency range. M3 M4 M6 Vin1 Vin2 Cc M1 M2... |

26 |
Symbolic network analysis
- Lin, M
- 1991
(Show Context)
Citation Context ...pts to generate interpretable expressions use various symbolic analysis methods to generate sum-of-product representations for network functions. This area has been studied extensively in 1960s-1980s =-=[14]-=-. The resulting approaches, however, are only feasible for very small circuits, since the number of expanded product terms grows exponentially with the size of a circuit, and the resulting expressions... |

21 | Canonical symbolic analysis of large analog circuits with determinant decision diagrams
- Shi, Tan
- 2000
(Show Context)
Citation Context ...nt efficient algorithms of generating dominant terms for deriving interpretable symbolic expressions based on a compact determinant decision diagram (DDD) representation of circuit transfer functions =-=[21]-=-, [22]. We show that dominant term generation can be performed elegantly by DDD graph manipulations. Since we start with exact symbolic expressions in DDD representations, our new approximation method... |

11 |
Lazy-expansion of symbolic expression approximation
- Seda, DeGrauwe, et al.
- 1992
(Show Context)
Citation Context ... been developed. Approximation after generation is reliable but it requires the expansion of product terms first [6], [9], [15]. Some improvements based on nested expressions have been proposed [16], =-=[17]-=-. But they generally suffer symbolic term-cancellation and alignterm problems. Approximation during generation extracts only significant product terms [2], [5], [7]. It is very fast, but has two major... |

11 |
DC small signal symbolic analysis of large analog integrated circuits
- Hsu, Sechen
- 1994
(Show Context)
Citation Context ...functions. Other smallsignal characteristics such as sensitivities, symbolic poles and zeros, cannot be extracted in general. At the same time, several approximation before generation techniques [7], =-=[18]-=- were proposed in which the complexity of a circuit is simplified before symbolic analysis methods are applied. Recently, a signalflow graph based approximation before generation method was proposed [... |

10 |
A symbolic analysis tool for analog circuit design automation
- Seda, Degrauwe, et al.
- 1988
(Show Context)
Citation Context ...terpretable by analog designers. Recently, various approximation schemes have been developed. Approximation after generation is reliable but it requires the expansion of product terms first [6], [9], =-=[15]-=-. Some improvements based on nested expressions have been proposed [16], [17]. But they generally suffer symbolic term-cancellation and alignterm problems. Approximation during generation extracts onl... |

9 | Rodrıguez-Vazquez A.: Symbolic Analysis of Large Analog Integrated Circuits: The Numerical Reference Generation Problem
- Fernandez, O
- 1998
(Show Context)
Citation Context ... nested expressions have been proposed [16], [17]. But they generally suffer symbolic term-cancellation and alignterm problems. Approximation during generation extracts only significant product terms =-=[2]-=-, [5], [7]. It is very fast, but has two major deficiencies: First, if accurate expressions are needed, the complexity of the approach becomes exponential. Second, it works only for transfer functions... |

8 | Efficient DDD-based symbolic analysis of large linear analog circuits
- Verhaegen, Gielen
- 2001
(Show Context)
Citation Context ...ability in approximation-before/after-generation for analyzing large analog circuits. Experimental results show that our algorithms outperform the best dominant term generation method reported so far =-=[4]-=-, [23]. Some preliminary results of this paper were presented s Go Cdb Csb b in [24], [25]. This paper is organized as follows. Section II reviews the concepts of DDDs and s-expanded DDDs. Section III... |

8 | A cancellation-free algorithm for the symbolic simulation of large analog circuits - Wambacq, Gielen, et al. - 1992 |

8 |
Symbolic Analysis in Analog Integrated Circuit Design
- Floberg
- 1997
(Show Context)
Citation Context ...es and phases of the transfer functions until errors are within the user-specified error bounds for the frequency range. M3 M4 M6 Vin1 Vin2 Cc M1 M2 Vbias M5 Fig. 1. A simplified two-stage CMOS opamp =-=[8]-=-. M7 VDD As illustrated in [9], simple yet accurate symbolic expressions can be interpretable by analog designers to gain insight into the circuit behavior, performance and stability, and are importan... |

7 |
On simplification techniques for symbolic analysis of analog integrated circuits
- Fernández, Martı́n, et al.
- 1992
(Show Context)
Citation Context ...s have been developed. Approximation after generation is reliable but it requires the expansion of product terms first [6], [9], [15]. Some improvements based on nested expressions have been proposed =-=[16]-=-, [17]. But they generally suffer symbolic term-cancellation and alignterm problems. Approximation during generation extracts only significant product terms [2], [5], [7]. It is very fast, but has two... |

6 | Efficient derivation of exact sexpanded symbolic expressions for behavioral modeling of analog circuits
- Shi, Tan
- 1998
(Show Context)
Citation Context ...he original DDD in which s is contained in some vertices is called complex DDD. The s-expanded DDD can be constructed from the complex DDD in linear time in the size of the original complex DDD [22], =-=[26]-=-. Before we generate the dominant terms, one problem we need to consider is symbolic cancellation. Symbolic canceling terms arise from the use of the MNA formulation in analog circuits. For instance, ... |

4 |
Symbolic modeling of periodically time-varying systems using harmonic transfer matrices
- Vanassche, Gielen, et al.
- 2002
(Show Context)
Citation Context ...ent symbolic techniques for linear(ized) analog circuit analysis are the basis of distortion analysis of weakly nonlinear circuits [11], [12], symbolic modeling of time-varying systems such as mixers =-=[13]-=-. Previous attempts to generate interpretable expressions use various symbolic analysis methods to generate sum-of-product representations for network functions. This area has been studied extensively... |

4 | Interpretable symbolic small-signal characterization of large analog circuits using determinant decision diagrams
- Tan, Shi
- 1999
(Show Context)
Citation Context ...n approach on the s-expanded DDDs before de-cancellation. IV. CONSECUTIVE K-SHORTEST PATH ALGORITHM FOR GENERATION OF DOMINANT TERMS We proposed an efficient algorithm for finding k dominant terms in =-=[27]-=-. The algorithm does not require DDDs to be native or to satisfy aforementioned graph theoretical property (Theorem 1), and thus can be applicable to any DDD graph. To differentiate this algorithm fro... |

3 |
Evaluation of error-control strategies for the linear symbolic analysis of analog integrated circuits
- Daems, Verhaegen, et al.
- 1999
(Show Context)
Citation Context ...odel generated in the final stage (as the model is recentered after every stage). Such nominal point approximation strategy used in this work is also adopted by most of symbolic approximation methods =-=[1]-=-–[7]. During the approximation, we monitor both magnitudes and phases of the transfer functions until errors are within the user-specified error bounds for the frequency range. M3 M4 M6 Vin1 Vin2 Cc M... |

3 |
Symbolic Distortion Analysis of Analog Integrated Circuits
- Verhaegen, Gielen
- 2001
(Show Context)
Citation Context ...del. d GmbVbs testability analysis and yield enhancement [10]. Efficient symbolic techniques for linear(ized) analog circuit analysis are the basis of distortion analysis of weakly nonlinear circuits =-=[11]-=-, [12], symbolic modeling of time-varying systems such as mixers [13]. Previous attempts to generate interpretable expressions use various symbolic analysis methods to generate sum-of-product represen... |

3 |
Circuit simplification for the symbolic analysis of analog integrated circuits
- Daems, Gielen, et al.
- 2002
(Show Context)
Citation Context ...] were proposed in which the complexity of a circuit is simplified before symbolic analysis methods are applied. Recently, a signalflow graph based approximation before generation method was proposed =-=[19]-=- and demonstrated successfully to symbolic pole and zero generation. Symbolic analysis based on the concept of signal paths in control theory was employed for pole and zero location analysis [3], [20]... |

3 |
representation and efficient generation of s-expanded symbolic network functions for computer-aided analog circuit design
- “Compact
- 2001
(Show Context)
Citation Context ...practical analog circuits. For each circuit, DC analysis is first carried out using SPICE and our program reads in small-signal element values from the SPICE output. The algorithms described in [21], =-=[22]-=- are used to construct complex DDDs and s-expanded DDDs. First we apply the proposed dominant term generation algorithms to derive interpretable symbolic expressions for transfer functions and poles f... |

3 | Parametric analog behavioral modeling based on cancellation-free DDDs - Tan, Shi - 2002 |

1 |
Analog small-signal modeling - part I: behavioral signal path modeling for analog integrated circuits
- Leyn, Gielen, et al.
- 2001
(Show Context)
Citation Context ...osed [19] and demonstrated successfully to symbolic pole and zero generation. Symbolic analysis based on the concept of signal paths in control theory was employed for pole and zero location analysis =-=[3]-=-, [20]. In this paper, we present efficient algorithms of generating dominant terms for deriving interpretable symbolic expressions based on a compact determinant decision diagram (DDD) representation... |

1 |
new reliable approximation method for expanded symbolic network functions
- “A
- 1996
(Show Context)
Citation Context ... longer interpretable by analog designers. Recently, various approximation schemes have been developed. Approximation after generation is reliable but it requires the expansion of product terms first =-=[6]-=-, [9], [15]. Some improvements based on nested expressions have been proposed [16], [17]. But they generally suffer symbolic term-cancellation and alignterm problems. Approximation during generation e... |

1 |
Analog small-signal modeling - part II: elementary transistor stages analyzed with behavioral signal path modeling
- Leyn, Gielen, et al.
- 2001
(Show Context)
Citation Context ...[19] and demonstrated successfully to symbolic pole and zero generation. Symbolic analysis based on the concept of signal paths in control theory was employed for pole and zero location analysis [3], =-=[20]-=-. In this paper, we present efficient algorithms of generating dominant terms for deriving interpretable symbolic expressions based on a compact determinant decision diagram (DDD) representation of ci... |

1 |
Symbolic determinant decision diagrams and their use for symbolic modeling of linear analog integrated circuits
- Verhaegen, Gielen
- 2002
(Show Context)
Citation Context ...ty in approximation-before/after-generation for analyzing large analog circuits. Experimental results show that our algorithms outperform the best dominant term generation method reported so far [4], =-=[23]-=-. Some preliminary results of this paper were presented s Go Cdb Csb b in [24], [25]. This paper is organized as follows. Section II reviews the concepts of DDDs and s-expanded DDDs. Section III prese... |

1 |
ddd-based term generation algorithm for analog circuit behavioral modeling
- “Efficient
- 2003
(Show Context)
Citation Context ...erimental results show that our algorithms outperform the best dominant term generation method reported so far [4], [23]. Some preliminary results of this paper were presented s Go Cdb Csb b in [24], =-=[25]-=-. This paper is organized as follows. Section II reviews the concepts of DDDs and s-expanded DDDs. Section III presents a dynamic programming based term generation algorithm based on the work of [4], ... |