## 2001 Kluwer Academic Publishers. Manufactured in The Netherlands. Mixed-Technology System-Level Simulation

### BibTeX

@MISC{Martinez_2001kluwer,

author = {J. A. Martinez},

title = {2001 Kluwer Academic Publishers. Manufactured in The Netherlands. Mixed-Technology System-Level Simulation},

year = {}

}

### OpenURL

### Abstract

Abstract. This paper describes a computationally efficient method to simulate mixed-domain systems under the requirements of a system-level framework. The approach is the combined use of Modified Nodal Analysis (MNA) for the representation of a mixed-technology device and piecewise linear (PWL) techniques to overcome the costly numerical evaluation found in conventional circuit or device simulators. This approach makes the simulation computationally fast, as well as more stable when compared with traditional circuit simulation. The PWL solver, based in the frequency domain, is more robust to inconsistencies in initial conditions and impulse changes when compared to integration based solvers in the time domain. The advantage of this method is that the same solver enables the integration of multi-domain devices (e.g., electrical, optical, and mechanical) in the same simulation framework. The use of this technique for the simulation of multi-domain systems has proven to give better performance in simulation time when compared to traditional circuit simulators with a relatively small decrease in the level of accuracy. Comparisons with traditional solvers, such as SPICE, show two to three orders of magnitude speedup with less than 5 % relative error. The ability to adjust the level of accuracy, either by varying the sampling rate or the number of regions of operation in the models, allows for both computationally fast and in-depth analysis in the same CAD framework. Key Words: MEM simulation, modified nodal analysis (MNA), optical MEM CAD tool, piecewise linear simulation (PWL), optoelectronic simulation, microsystem modeling and simulation 1.

### Citations

519 | Ptolemy: A framework for simulating and prototyping heterogeneous systems
- Buck, Ha, et al.
- 1994
(Show Context)
Citation Context ...r. The actual approach for our modeling of mixed signal, multi-domain systems is a discrete event driven simulation model, which operates over the global system. An object-oriented framework, Ptolemy =-=[26]-=- is used to provide this degree of abstraction for the simulation of such systems. We choose Ptolemy’s “Dynamic Data Flow” (DDF) Ptolemy simulation method as our discrete event engine. Timing informat... |

452 | Asymptotic waveform evaluation for timing analysis - Pillage, Rohrer - 1990 |

451 |
Digital Integrated Circuits: A Design Perspective, second edition
- Rabaey, Chandrakasan, et al.
- 2003
(Show Context)
Citation Context ...The speed-up achieved for the same number of timesteps is at least two orders of magnitude. The increase in execution time vs. size of the circuit under test for SPICE is well known to be superlinear =-=[42]-=-. This behavior is evidenced in the time of execution used by the SPICE simulator to solve the three sizes of test circuits used in the experiment. On the other hand, each CMOS inverter is a separate ... |

353 | PRIMA: passive reducedorder interconnect macromodelling algorithm - Odabasioglu, Celik, et al. - 1998 |

259 | Efcient Linear Circuit Analysis by Padé Approximation via the Lanczos Process
- Feldmann, Freund
- 1995
(Show Context)
Citation Context ...ht of as characterizing the interconnection network and parasitic elements while the non-linear subblocks characterize active non-linear behaviors. In the second step, a Modified Nodal Analysis (MNA) =-=[27]-=- is used to create a mathematical representation for the devices, as shown in Fig. 3. Throughout this section we describe this representation in the electrical domain, without loss of generality, as w... |

134 |
Computer Methods for Circuit Analysis and Design
- Vlach, Singhal
- 1994
(Show Context)
Citation Context ... linear electrical network representation into a system of first-order differential equations and algebraic equations. For a development of this technique the reader can explore the work presented in =-=[31,32]-=-. This formulation, applied to the electrical domain as shown in Fig. 3, is based in the application of the KCL (Kirchoff’s current law) over the nodal representation of the circuit and the use of the... |

66 |
Theory of Matrix Structural Analysis
- Przemieniecki
- 1968
(Show Context)
Citation Context ...n to the PWL solver for evaluation. We explain this technique next. With damping forces proportional to the velocity, the equation of motion for a mechanical structure with viscous damping effects is =-=[40]-=-: F = KU + BV + MA (16) where, K is the stiffness matrix, U is the displacement vector, B is the damping matrix, V is the velocity vector, M is the mass matrix, A is the acceleration vector, and F is ... |

59 |
IRSIM: An incremental MOS switch-level simulator
- SALZ, HOROWITZ
- 1989
(Show Context)
Citation Context ...ient. As an alternative to traditional circuit simulation, a non-linear network modeling technique using piecewise linear (PWL) models was added to the timing simulator IRSIM by Kao [16] and Horowitz =-=[17]-=-. This technique is well suited for the delay estimation in dense non-linear digital-VLSI networks. However, the limited complexity of the MOSFET models and tree analysis technique used in this method... |

44 | Reduced-Order Modeling of Large Passive Linear Circuits by Means of the SyPVL
- Feldmann, Freund
- 1996
(Show Context)
Citation Context ... linear electrical network representation into a system of first-order differential equations and algebraic equations. For a development of this technique the reader can explore the work presented in =-=[31,32]-=-. This formulation, applied to the electrical domain as shown in Fig. 3, is based in the application of the KCL (Kirchoff’s current law) over the nodal representation of the circuit and the use of the... |

36 |
Bokhoven, Piecewise Linear Modeling and Analysis
- Leenaerts, van
- 1998
(Show Context)
Citation Context ...chosen to closely match the original function but are computationally more tractable. This technique has been applied with relative success in simulators such as NECTAR 2 [18], PLANET [19], and PLATO =-=[20]-=-. These simulators are much more stable when compared to traditional circuit simulators and provide flexibility for their use in hierarchical design. In these simulators, the complete PWL representati... |

32 | Pillage, “Time-domain macromodels for VLSI interconnect analysis - Kim, Gopal, et al. - 1994 |

22 | Micromaching for optical and optoelectronic systems
- Wu
- 1997
(Show Context)
Citation Context ...oduction Applications for optical MEMS (micro-electricalmechanical systems), or MOEMS, are growing to include switching, scanning, projection, display, printing, sensing, modulating, and data storage =-=[1]-=-. As these applications transition from abstract ideas to marketable products, the cost of designing, prototyping, and testing these systems has slowed down this development time. Instead of physicall... |

18 | Structured design of microelectromechanical systems”, DAC ’97. CONTACT Shuvra Das, Professor, Mechanical Engineering Department
- Mukherjee, Fedder
(Show Context)
Citation Context ...circuits and devices. Code V [2] and ASAP [3] are available for accurate optical modeling, and CAD tools for conventional MEMS are being designed in both academia and industry, including those by CMU =-=[4]-=-, Conventor [5], and MEMScaP [6]. However, these tools do not provide a single multi-domain design environment, which is needed for modeling mixed-technology applications. This paper concentrates on t... |

17 |
An Approach to Macromodeling of MEMS for Nonlinear Dynamic Simulations”, DSC-Vol
- Ananthasuresh, Gupta, et al.
- 1996
(Show Context)
Citation Context ... following substitutions: θ = bVth + η; ψ = γ Sn−1; ξ(in, out, j, k, bn−1, Sn−1) = (bn−1 + Sn−1λk)(T (in, j, k) − a · T (out, j, k)); N∑ Ɣ(k) = ξ(j, k); j=1 This gives: N∑ Ɣ(k) e tb/λk + ψtb + θ = 0; =-=(14)-=- k This expression is, in general, not analytically solvable. To obtain the desired values of tb, or roots for the expression, an algorithm such as Newton-Raphson can be used. There is, however, an im... |

15 |
Methodology for the Modeling and Simulation of Microsystems”, PhD
- Romanowicz
- 1997
(Show Context)
Citation Context ...ea of research. It involves the characterization of ports, defined as transducers (energy conversion devices), and elements, defined as actuators (unidirectional energy flow devices), as explained in =-=[9]-=-. Additionally, this technique relies on the abstraction levels offered by the mixed-simulation framework and, in doing so, it shares its drawbacks as well. For example, mixed-signal, multi-domain mic... |

12 |
Digital free-space optical interconnections: a comparison of transmitter technologies
- Fan, Mansoorian, et al.
- 1995
(Show Context)
Citation Context ...ultiple laser channels on the same chip that makes this technology so attractive. Consequently, VCSELs have become the optical source of choice for almost every optoelectronic link currently designed =-=[33,34]-=-. Typical L-I and V-I diagrams for a VCSEL are shown in Fig. 7(a) [34]. The segmented curve represents the PWL Modeling of the L-I Characteristics The optical output of VCSEL devices is a non-linear f... |

7 |
iSMILE: A Novel Circuit Simulation Program with emphasis on New Device Model Development
- Yang, Kang
- 1989
(Show Context)
Citation Context ...sentation for the non-electrical domain required to be simulated. The electrical equivalent can be simulated using any of the well known and established circuit simulators (e.g., SPICE, SABER, iSMILE =-=[10,11]-=-). This method has been used in [12– 14] for the simulation of micro mechanical devices, where a mapping of these devices to a SPICE netlist is proposed. In [10] Yang simulates optoelectronic intercon... |

7 |
Nodal Design of Actuators and Sensors (NODAS
- Vandemeer
- 1998
(Show Context)
Citation Context ...n depends on more nonlinear terms (i.e., the exponential in the expression). With these considerations, x(t) is given by: xi(t) = N∑ j=1 k=1 N∑ (T (i, j, k) × [(bn−1 + Sn−1λk)(1 − e t/λk ) + Sn−1t]); =-=(12)-=- where the input to the system is given by the piecewise linear expression: u(t) = bn−1 + Sn−1t; Using expression (12) in expression (11) with some algebraic simplification gives: N∑ j=1 k=1 N∑ (bn−1 ... |

7 |
Hierachical representation and simulation of micromachined inertial sensors,” Modeling and
- Vandemeer, K
(Show Context)
Citation Context ...0 Martinez et al. 5. Micro Mechanical Modeling In the field of MEM modeling, there has been an increasing amount of work that uses a set of Ordinary Differential Equations to characterize MEM devices =-=[11,38,39]-=-. ODE modeling is used instead of techniques such as finite element analysis, to reduce the time and amount of computational resources necessary for simulation. The model uses non-linear differential ... |

7 | Diffractive Optical Propagation Techniques for a Mixed-Signal CAD
- Kurzweg, Levitan, et al.
(Show Context)
Citation Context ...electrostatic attraction is performed using the PWL technique described throughout this paper. Chatoyant’s optical modeling technique is based on the RayleighSommerfeld scalar diffraction formulation =-=[44]-=-. We use RSoft’s BeamPROP [45] to simulate the light through the fiber, and have developed an interface between Fig. 13. (a) Switching system. (b) Mirror response and intensity distributions.Mixed-Te... |

6 |
Timing analysis for piecewise linear Rsim
- Kao, Horowitz
- 1994
(Show Context)
Citation Context ...pplication inefficient. As an alternative to traditional circuit simulation, a non-linear network modeling technique using piecewise linear (PWL) models was added to the timing simulator IRSIM by Kao =-=[16]-=- and Horowitz [17]. This technique is well suited for the delay estimation in dense non-linear digital-VLSI networks. However, the limited complexity of the MOSFET models and tree analysis technique u... |

6 |
Piecewise Linear Simulation of Optoelectronic Devices with Applications to MEMS
- Martinez
(Show Context)
Citation Context ...l variable dependencies can be added to the L-I VCSEL model following this approach (e.g., temperature, spot size, and threshold) that allow one to study their effect on complete systems. As shown in =-=[37]-=-, this VCSEL model is used to examine the dependency of BER on the temperature at which the device is operated. For the complete modeling of the VCSEL, the same technique is also used for additional n... |

5 |
The Mixed-Technology Modeling and Simulation of Opto-Electronic Microsystems
- Morikuni, Mena, et al.
- 1999
(Show Context)
Citation Context .... A mixed signal CAD tool is used as the simulation framework. Consequently, the support for co-simulation is already in place. This methodology is applied in the simulation of optical MEM systems in =-=[7,8]-=-. Even though the behavioral modeling approach appears to be a promising option for the modeling of mixed-signal, multi-domain systems, it is necessary to clarify the difficulties and limitations pres... |

5 | Mariappan M., “An Environment for Design and Modeling of Electro-Mechanical Micro-Systems
- Zaman, Bart, et al.
- 1999
(Show Context)
Citation Context ...b-functions are chosen to closely match the original function but are computationally more tractable. This technique has been applied with relative success in simulators such as NECTAR 2 [18], PLANET =-=[19]-=-, and PLATO [20]. These simulators are much more stable when compared to traditional circuit simulators and provide flexibility for their use in hierarchical design. In these simulators, the complete ... |

5 |
Reciprocation of Triply Partitioned Matrices
- Duncan
- 1956
(Show Context)
Citation Context ...s), they represent a set of non-linear ODEs. To reduce the above equation to a standard form, we use a modification of Duncan’s reduction technique for vibration analysis in damped structural systems =-=[41]-=-. This modification allows the above general mechanical motion equation to be reduced to a standard first order form, similar to electrical equation found in Fig. 3. This gives a complete characteriza... |

4 |
MOEMS Modeling for Opto-Electro-Mechanical Co-Simulation
- Jacquemod, Vuorinen, et al.
- 1999
(Show Context)
Citation Context .... A mixed signal CAD tool is used as the simulation framework. Consequently, the support for co-simulation is already in place. This methodology is applied in the simulation of optical MEM systems in =-=[7,8]-=-. Even though the behavioral modeling approach appears to be a promising option for the modeling of mixed-signal, multi-domain systems, it is necessary to clarify the difficulties and limitations pres... |

4 |
Computer-Aided Design of Optoelectronic Integrated Circuits and Systems
- Morikuni, Kang
- 1997
(Show Context)
Citation Context ...sentation for the non-electrical domain required to be simulated. The electrical equivalent can be simulated using any of the well known and established circuit simulators (e.g., SPICE, SABER, iSMILE =-=[10,11]-=-). This method has been used in [12– 14] for the simulation of micro mechanical devices, where a mapping of these devices to a SPICE netlist is proposed. In [10] Yang simulates optoelectronic intercon... |

4 |
Pull-in Dynamics of Electrostatically-Actuated Beams
- Gupta, Hung, et al.
- 1996
(Show Context)
Citation Context ... = bn−1 + Sn−1t; Using expression (12) in expression (11) with some algebraic simplification gives: N∑ j=1 k=1 N∑ (bn−1 + Sn−1λk) e tb/λk (T (in, j, k) −a · T (out, j, k)) + γ · Sn−1tb + bVth + η = 0 =-=(13)-=- where, the following substitutions were made: γ = η = N∑ j=1 k=1 N∑ N∑ (a · T (out, j, k) − T (in, j, k)); N∑ [(a · T (out, j, k) − T (in, j, k)) j=1 k=1 × (bn−1 + Sn−1λk)] In expression (13), in rep... |

4 |
NECTAR 2 a circuit analysis program based on piecewise linear approach,” by
- Kawakita, Ohtsuki
- 1975
(Show Context)
Citation Context ...ons. These sub-functions are chosen to closely match the original function but are computationally more tractable. This technique has been applied with relative success in simulators such as NECTAR 2 =-=[18]-=-, PLANET [19], and PLATO [20]. These simulators are much more stable when compared to traditional circuit simulators and provide flexibility for their use in hierarchical design. In these simulators, ... |

4 | An algorithm for finding all solutions of piecewise-linear resistive circuits - Yamamura, Mishina - 1996 |

4 | Chatoyant: A computer-aided-design tool for free-space optoelectronic system
- Levitan, Kurzweg, et al.
- 1998
(Show Context)
Citation Context ...hnology130 Martinez et al. simulator Chatoyant. Chatoyant is a multi-level, multidomain CAD tool that has been successfully used to design and simulate free space optoelectronic interconnect systems =-=[24]-=-. Static simulations analyze mechanical tolerancing, power loss, insertion loss, and crosstalk, while dynamic simulations are used to analyze data streams with techniques such as noise analysis and bi... |

4 |
A high-speed low-voltage stressinduced micromachined 2 × 2 optical switch
- Chen, Nguyen, et al.
- 1999
(Show Context)
Citation Context ...exhibit limited diffraction, while more computationally intensive scalar models are used when diffraction effects dominate the system. A Chatoyant schematic of a 1×2 optical MEM optical cross-connect =-=[25]-=- is shown in Fig. 1. In Chatoyant, each icon represents a component model with sets of parameters defining the characteristics of the component. Each line represents a signal path (either optical, mec... |

4 |
Nodal Analysis for MEMS Simulation and Design
- Zhou, Clark, et al.
- 1998
(Show Context)
Citation Context ...0 Martinez et al. 5. Micro Mechanical Modeling In the field of MEM modeling, there has been an increasing amount of work that uses a set of Ordinary Differential Equations to characterize MEM devices =-=[11,38,39]-=-. ODE modeling is used instead of techniques such as finite element analysis, to reduce the time and amount of computational resources necessary for simulation. The model uses non-linear differential ... |

2 |
is a registered trademark of Optical Research Associates
- CODE
(Show Context)
Citation Context ...ach of these specialized domains and it is unnecessary to reinvent these tools. For example, SPICE is the standard128 Martinez et al. simulation technique for electrical circuits and devices. Code V =-=[2]-=- and ASAP [3] are available for accurate optical modeling, and CAD tools for conventional MEMS are being designed in both academia and industry, including those by CMU [4], Conventor [5], and MEMScaP ... |

2 |
Research Organization
- Breault
(Show Context)
Citation Context ...specialized domains and it is unnecessary to reinvent these tools. For example, SPICE is the standard128 Martinez et al. simulation technique for electrical circuits and devices. Code V [2] and ASAP =-=[3]-=- are available for accurate optical modeling, and CAD tools for conventional MEMS are being designed in both academia and industry, including those by CMU [4], Conventor [5], and MEMScaP [6]. However,... |

2 |
Inside SPICE: Overcoming the obstacles of Circuit Simulation
- Kielowski
- 1994
(Show Context)
Citation Context ...theirMixed-Technology System-Level Simulation 129 reliability and long computation times. It is also well known that there are non-convergence problems faced during simulation of electronic circuits =-=[15]-=-. These techniques become computationally expensive with increasing size of the network being analyzed and when the network has a high degree of non-linearity. The use of such a computationally intens... |

2 | The generalized linear complementary problem applied to the complete analysis of resistive piecewise linear circuits - Vandeberghe, Moor, et al. - 1989 |

2 | An algorithm for solving piecewise linear networks that include elements with discontinuities characteristics - Pejovic, Maksimovic - 1996 |

2 |
VCSEL-based Digital Free-Space Optoelectronic Interconnections,” (unpublished
- Kibar
- 1999
(Show Context)
Citation Context ...ultiple laser channels on the same chip that makes this technology so attractive. Consequently, VCSELs have become the optical source of choice for almost every optoelectronic link currently designed =-=[33,34]-=-. Typical L-I and V-I diagrams for a VCSEL are shown in Fig. 7(a) [34]. The segmented curve represents the PWL Modeling of the L-I Characteristics The optical output of VCSEL devices is a non-linear f... |

2 | Vertical cavity surface emitting lasers and arrays for optical interconnect and optical computing applications,” SPIE Vol. 1563 Optical Enhancements to Computing Technology - Wang, Dziura, et al. - 1991 |

2 |
Small-signalequivalent circuits for a semiconductor laser
- Kibar, Blerkom, et al.
- 1998
(Show Context)
Citation Context ...decreases with further increases in electrical current. The reason for this behavior is the loss of efficiency in the laser, caused by the increase in temperature as a consequence of the ohmic losses =-=[36]-=-. To model this behavior of the device, however, is typically challenging because the optical power is not an electrical parameter. Nevertheless, an interesting feature of the PWL methodology is that ... |

1 |
is an electrical engineering Ph.D. student at the University of Pittsburgh. He received his M.S. from the University of Pittsburgh in 2000 in electrical engineering. He received the B.S. in electrical engineering from the Universidad de Oriente
- Rsoft, Martinez
- 1993
(Show Context)
Citation Context ...rformed using the PWL technique described throughout this paper. Chatoyant’s optical modeling technique is based on the RayleighSommerfeld scalar diffraction formulation [44]. We use RSoft’s BeamPROP =-=[45]-=- to simulate the light through the fiber, and have developed an interface between Fig. 13. (a) Switching system. (b) Mirror response and intensity distributions.Mixed-Technology System-Level Simulati... |