Behavioral models for microwave devices from time domain large-signal measurements are developed. For the presented examples, the model is defined by representing the terminal currents as a function of the terminal voltages and their derivatives. When using these models as building blocks of higher level designs, the simulation speed is significantly improved.
An application is presented for online model-based fault detection and isolation (FDI) in a multitank fluid system. The tank system is equipped with a distributed measurement and control system that implements components of the IEEE standard for smart transducers, IEEE 1451 [1]. This standard includes an information model that provides programming constructs to support high level application functionality on a distributed network of smart transducers. The model-based FDI methodology in this work has several aspects that may be realized on such a distributed network. In the current work, the FDI application operates on a workstation that appears on the network as another (virtual) transducer node. The concurrent tasks in the application may be associated with actual transducer nodes. It represents a first effort toward constructing capabilities for distributed FDI in complex dynamic systems.
We present a new procedure for creating non-linear behavioral models of microwave devices, based on techniques developed in Time-Series Analysis. We illustrate this procedure by creating a model for a HEMT device. Large-signal time-domain microwave measurements are used to generate the time-series data of the terminal currents and voltages. The dynamical model of the HEMT is defined by fitting multivariate polynomials to the terminal voltages and their higher-order derivatives. The model accurately predicts DC, small-signal, and large-signal behavior.
The capability of time domain simulation with frequency domain data exists in some time-domain simulators, but the computations can be time-consuming. We decrease this computational burden by exploiting S-parameter fitter and fast recursive convolution methods. With recursive convolution the frequencies at which the S-parameters are sampled may be spaced in any way. For example, a logarithmic frequency spacing allows S-parameters to be sampled over a broader band without increasing the number of frequencies measured.We use two different system identification techniques for extracting the closed-form equations describing measured or simulated S-parameter data. Lumped parameter systems are approximated as a rational polynomial modulated by a complex exponential. Distributed parameter systems are approximated by the sum of complex exponentials. The recursive convolution engine handles both forms. The HP SPICE circuit simulator has been extended to allow fast, recursive convolution methods to be employed for transient analysis. The resulting simulator is called SSpice v2. Significant speed-up in the simulation of several circuits have been achieved using this new technique compared to both the traditional approach in SPICE and our previous directconvolution-based approach in SSpice v l[TC95].In this paper, we present the algorithms and applications of the simulator. Three applications are demonstrated in the paper which are MR head flex line modeling, chip-to-chip signal modeling on MCM, and on-chip inductor modeling.
IntroductionThis paper is concerned with the following problem: Suppose one has measured the frequency-domain transfer functions of a passive, linear electronic "device" such as wiring, connectors, power/ground planes, and on-and off-chip interconnects. One would like to simulate the device along with other components connected to it with a timedomain simulator such as SPICE. It is desirable to be able to perform the simulation and done quickly. This problem may arise during design of a new RF or high-speed digital product.All techniques for solving this problem depend on the sampling of the transfer functions between all ports of the device at many frequencies. The transfer functions may be kept as impedance matrices, admittance matrices, or scattering (also called "S-parameter") matrices. All of these matrix types are equivalent in the sense that there are simple mathematical formulas for converting among them.' The capability of performing time-domain simulation based on S-parameter measurements exists in SSpice VI, a version of HP-SPICE that has FFT-based S-parameter convolution added [TC95]. However, for typical applications, accurate transient analysis requires that S-parameters be measured from DC to well over 1 GHz. The large number of data samples present in such measurements, along with the fact that convolution operations on these samples must be performed at each simulator time step, imply that the resulting simulation runs may be time consuming.When FFT-based convolution is used, the freque...
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.