SPICE Transient Analysis Research Articles

Life After SPICE

I graduated among the last wave of students who really focused on circuit simulation at the University of California (UC), Berkeley. Those students included Tom Quarles, who wrote SPICE3 as a way of improving the underlying architecture of SPICE, and Jacob White, who explored new numerical algorithms such as waveform relaxation as a way of accelerating traditional SPICE transient analysis. These students, like most of the ones who preceded them, accepted the basic capabilities of SPICE and were working to improve them by making them faster, more accurate, and more robust. My work was a bit different; I focused on teaching the old dog new tricks. I tried to develop new analyses specifically tailored for designers of high-frequency circuits. In particular, I worked to develop methods for efficiently computing the steady-state response of a circuit. This work culminated in the development of a harmonic balance simulator that eventually formed the basis of the commercial microwave simulator from Hewlett Packard (now Agilent). After graduation, I went in another direction. I took the harmonic balance simulator I wrote as a student to Cadence and converted it to a general purpose SPlCE-like simulator. At the time HSPICE was the dominant commercial simulator for integrated circuits, and some of us at Cadence had aspirations of building a new simulator to replace it. I took the lessons I learned about simulator architecture from Tom Quarles and simulator algorithms from Jacob White and produced a new simulator named Spectre that was several times faster than HSPICE as well as being more accurate and more robust (actually, Jacob and I working together completed the first version in two weeks). Essentially, I took everything that had been learned since the development of SPICE2 and put it into Spectre. It was not enough.

Limit-cycle oscillations in a log-domain-based filter

A circuit is presented for a class-AB fully differential log-domain parallel resonator based on a capacitively loaded gyrator. It is shown that if floating rather than shunt capacitive loads are used, the circuit is stable for small stimuli, but can be triggered into steady-state limit-cycle oscillations by large enough transient inputs. The nature of the instability is explored and it is shown how SPICE transient analysis can be used to detect the instability and locate the separatrix between stable-equilibrium-point-seeking and limit-cycle-seeking regions of state space. Analysis is also presented to demonstrate that the circuit with floating capacitors does not implement an externally linear transfer function, and therefore, it is not a true log-domain circuit.

Introduction to RF simulation and its application

Radio-frequency (RF) circuits exhibit several distinguishing characteristics that make them difficult to simulate using traditional SPICE transient analysis. The various extensions to the harmonic balance and shooting method simulation algorithms are able to exploit these characteristics to provide rapid and accurate simulation for these circuits. This paper is an introduction to RF simulation methods and how they are applied to make common RF measurements. It describes the unique characteristics of RF circuits, the methods developed to simulate these circuits, and the application of these methods.

Determination of bit-rate and sensitivity limits of an optimized p-i-n/HBT OEIC receiver using SPICE simulations

The sensitivity of an OEIC receiver depends essentially on the physical sources of device and circuit noise referred to its input, provided that the inter-symbol interference (ISI) makes no significant contribution. For well designed receivers, the latter situation can be realized only at an optimum bandwidth (f/sub 3 dB.opt/) for a given bit rate (B) or vice versa. In this paper, we have determined the relationship between the bit rate and the 3-dB bandwidth for negligible and pre-set levels of ISI for an optimized p-i-n/HBT transimpedance receiver with adjustable bandwidth. We have used SPICE simulations in the frequency domain to determine the effect of device and circuit noise, and SPICE transient analysis to determine the effect of ISI on the sensitivity. The ratio f/sub 3 dB.opt//B has been found to vary from 0.65 to 0.45 when B changes from 10 to 20 Gbps for the OEIC receiver used.

Modern guide to spectral analysis with SPICE

Either SPICE's own command or a post-processor applying an FFT can be used to deliver a high-quality spectral analysis, provided attention is paid to ensure that circuit settling time is adequate, that the signals are harmonically suitable or an appropriate window applied, and that the signal to be analysed is adequately sampled. While only a good understanding of the discrete and fast Fourier transforms will lead to fool-proof analysis, it is possible to tackle most situations by adhering to simple guidelines described here, We present our own post-processing algorithm. This is a relatively robust tool for performing spectral analysis on SPICE transient analysis output data.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Adaptation of "SPICE3" to simulation of lossy multiple-coupled transmission lines

Design of high-speed, high-performance electronic circuits requires simulation of transients in networks that include multiple, coupled lossy transmission lines. A widely used circuit simulator, SPICE, does not have facilities for simulation of multiconductor transmission lines. Recently a modification to SPICE3 to include multi-conductor lossless lines was reported. In many situations line losses must be included in the model and networks with lossy transmission simulated. In general simulation of lossy transmission lines is complex and computationally very intensive. The situation is simpler when D-C losses are considered. A recent study concluded that D-C losses provide an adequate modeling of signal transmission in many practical situations. This paper describes the implementation of multi-conductor, coupled lines with D-C losses in SPICE3e2. This implementation ties SPICE transient analysis with a special algorithm for the analysis of a transmission line and is more efficient than another approach based an the concept of a transmission line subcircuit. Internal tests of the new program, LSPICE3, were successfully performed and now this program is available to SPICE3 users. >

SPICE transient analysis errors estimated in frequencyspace

Comparison of the circuit simulator SPICE numerical integrator Z transform with the original system Laplace transform reveals that the trapezoidal method causes only frequency wrapping, while the Gear method generates frequency and amplitude errors. A universal expression for the magnitude of the Gear method error is derived. The theoretical analysis is tested by an example.

The simulation errors introduced by the SPICE transient analysis

The SPICE transient simulations of simple circuits that can be easily solved by hand often give erroneous results. The numerical errors are investigated and analyzed using the analogy of the simulation program with a digital integrator. The transfer function of the digital integrator is defined through the z-transform. The solving method implemented in SPICE is one of the possible approximations of the z-transform. This approximation has some good properties but also some drawbacks; it warps the frequency and it is not step invariant.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A charge conserving non-quasi-state (NQS) MOSFET model for SPICE transient analysis

An analytic charge-conserving non-quasi-static (NQS) model is derived for long-channel MOSFETs and has been implemented in SPICE3. The model is based on approximate solutions to the transient current continuity equation, an analytic equations are derived for node charges using the charge-sheet formulation. The NQS effects in several test circuits, which include a pass transistor, a CMOS inverter chain, and a differential sample-hold circuit, are stimulated. Excellent agreements have been observed among this work, PISCES (2-D device simulation), the 1-D numerical simulation, the multiple lump model, and CODECS (a mixed device and circuit simulation). However, large differences have been observed between this work and conventional quasi-static (QS) models. The model computation time of this work implemented in SPICE3 is about 2-3 times larger than those of QS models (BSIM, Level-2 Meyer) in SPICE3.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Novel techniques to solve sets of coupled differential equations with SPICE

The solution of sets of coupled differential equations using SPICE is treated. The equations are rearranged so that each of the variables appears as a current through a 1- Omega resistor, and the SPICE transient analysis option TRAN is used to invoke the DCTRAN module. This module performs the DC operating point analysis, the initial-condition of transient analysis, the DC transfer characteristic analysis, and the transient analysis. By initially invoking the subroutine LOAD, the linearized system of nodal equations for the specific iteration is constructed. Separate computation of the nonlinear devices' contribution to the Y matrix is carried out by the respective subroutines. >

Modeling Time-Dependent Elements for SPICE Transient Analyses

A technique is presented which uses existing SPICE elements for simulating circuits containing time-dependent element values. An example of a SPICE transient analysis of a circuit having a time-dependent capacitance is presented. The modeling technique is extended to time-dependent resistors and inductors with SPICE subcircuit descriptions given for each element model.