SLASH Tutorials

Fast Fourier Transform (FFT) based measurements are widely used by designers to verify their circuits in many fields such as signal conditioning, instrumentation, or audio applications. However, this powerful mathematical tool is subject to errors when used without caution. Indeed, a designer can experience aliasing or spectral leakage which reduces the accuracy of the FFT results. The goal of this tutorial is to go through the basics of FFT-based measurements and to show the functionalities of SMASH to compute the FFT of a time- domain signal.

This tutorial introduces the designer to SLASH through the discovery of design and simulation, including netlisting configuration in SLED and how to start basic analyses in SMASH. It introduced the concepts related to Design Contexts and their configurations, as well as how to setup the dynamic link between SLED and SMASH for netlisting, back-annotation and cross-probing.

Transfer function and pole / zero extraction are mathematical tools commonly used by analog designers to characterize a circuit and then to model it as a black box in order to validate the behavior of a design or to speed up untractable simulations. However, classical analyses such as the Small-Signal analysis are not sufficient for accurate modeling.

This tutorial presents SMASH capabilities to extract the Transfer function of a design to go beyond small signal analysis:

• For low complexity circuits, the Pole Zero analysis (with the .PZ directive) allows the extraction of Poles and Zeros of the design and the automatic generation of a behavioral model using a Laplace function.
• For higher complexity circuits, the Domain Coloring analysis (with the .ACMAP directive) allows the visualization and identification on the complex map of the Poles and Zeros of the design and the automatic generation of a behavioral model using a Laplace function.

Then, a true frequency analysis can be done on the designs with these transfer function models. Moreover, verifications can be sped up by replacing circuit implementations by their Laplace models.

Testbench approach is widely used in the verification process of a circuit. But, as the complexity of designs is increasing, it becomes a challenge to guaranty that the test cases have sufficiently exercised the RTL code. Code coverage can be used to determine how a set of test cases exercise the RTL code of a circuit. To do this, SMASH gathers information about which parts of the code of the design under test were exercised during simulation and which parts were not. As a result, using RTL code coverage with SMASH enables to improve the verification coverage of your design.

KEYWORDS: DFT, FFT, Fourier, sampling, aliasing, spectral leakage, windowing

Fast Fourier Transform (FFT) based measurements are widely used by designers to verify their circuits in many fields such as signal conditioning, instrumentation, or audio applications. However, this powerful mathematical tool is subject to errors when used without caution. Indeed, a designer can experience aliasing or spectral leakage which reduces the accuracy of the FFT results. The goal of this tutorial is to go through the basics of FFT-based measurements and to show the functionalities of SMASH to compute the FFT of a time- domain signal.

KEYWORDS: application hardware modeling, multi-level, multi-domain, mixed language, behavioral modeling

Full-circuit simulations are very quickly limited by excessive simulation time as well as by the availability of appropriate models. Simulation of a system cannot be effectively performed at the most accurate level (commonly SPICE) both because it requires too much simulation time and because multi-domain systems cannot be accurately modeled using a single design language. Furthermore, when integrating components into a system, the system designer must necessarily consider that they meet the specifications and that they individually work as specified. The focus is then on verifying that the assembly in the system works as specified.

This tutorial presents the unique capabilities of SMASH for simulation of the Application Hardware Model (AHM) of a system specific function by combining the models of each component of the function at the appropriate description levels required for the analyses. The goal of such simulations is to simulate the overall function to detect design defects in the assembly and to verify the function performances.

KEYWORDS: audio, filtering, sampling, aliasing

The designer of an audio application will use a technical approach in order to check that the design meets the requirements. For instance, he may verify that profiles are in accordance with templates in the frequency domain or that SNR levels are correct for a sample of sine waveforms (with fixed amplitude and frequency)...
However, evaluating the quality of an audio application is more subjective as it is linked to the perception of the human hearing: it is much less a quantitative measurement. SMASH enables designers to generate audio output files from the results of a time domain simulation in order to listen to the effects of the application schematic.
SMASH also provides the ability to use an audio file as input of a transient simulation using dedicated voltage or current sources. This enables direct driving of the input of an analog circuit or driving of the input of a digital circuit through an ADC.

KEYWORDS: load, impedance, high-impedance net, floating net, HiZ, SPICE

High impedance (also known as hi-Z, tri-stated, or floating) is the state of a net which is not currently driven by the circuit or the nets that do not have any low impedance path to ground. Such high impedance nets are particularly sensitive to their environment and can pull-up or pull-down to an unexpected voltage value. This state is particularly difficult to check during design or test whereas it can lead to random circuit failure as well as yield drops. An innovative feature provided by SMASH aims at avoiding these failures by detecting Hi-Z nets and characterizing impedances of net.

KEYWORDS: multiple operating-points, steady state, SPICE

Finding DC operating points is the first and maybe the most determinant task for analog simulation. With the operating point analysis, SMASH proceeds through a complete search, using all heuristics in sequence, and gives the first stable result that is found. With multiple operating-points analysis, SMASH helps the designer to determine whether multiple operating-points exist and to identify meaningful ones in order to simplify eradicating unwanted operating-points which render the circuit useless when self-biased in such operating conditions.

KEYWORDS: Jitter, Behavioral, Multi-level, Calibration, Noise

Evaluation of jitter in a design is a key issue difficult to address. Even if RF simulation solutions have provided some answers, all types of jitters can not be determined efficiently. SMASH provides a complete methodology combining multi-level simulation with calibration that provides a breakthrough solution for simulating properly all forms of Jitter with shortened simulation times.
High level models of different noise sources are elaborated to describe all components of a design at a behavioral level. Noise sources can then be included in the behavioral models, enabling a fast simulation of jitters. Transient or small signal noise analyses are used to calibrate a behavioral model at system level.
Illustration of this approach is given on the design of a basic PLL. The results obtained attest a reduced simulation time with a good accuracy.

KEYWORDS: Design yield, Monte Carlo, reliability, dispersion

Sensitivity to dispersion is a new and efficient way to address circuit yield issue. For DC transfer and small signal analyses, SMASH now allows reducing time spent in dispersion analysis of circuits. From foundry data, you extract relevant parameters to introduce dispersion in component models. Indeed, threshold voltages VT, gain factor β dispersions are included in transistor models to take into account mismatch between components. SMASH then uses proprietary algorithms to evaluate local dispersion up to thousands of time faster than classic Monte Carlo analysis. Sensitivity to dispersion with SMASH is a faster method to perform circuit reliability analysis.

KEYWORDS: Design yield, Monte Carlo, reliability, dispersion, matching

The main drawback of classic yield analysis solutions based on Monte Carlo is that they do not provide the means to diagnose the causes of yield losses.
The patented Imbalance Locate innovation consists in providing the means to reproduce the error cases and to determine the disturbing devices.
Thanks to this extension to classic Monte Carlo, SMASH provides an efficient methodology allowing designers to diagnose which devices (transistors, resistors, capacitors...) are sensitive to dispersion and matching effects and cause yield losses.
With Imbalance Locate, go beyond detecting the error cases in a circuit to find the elements at the origin of the error cases. This methodology is an efficient way to accelerate analog design while increasing the robustness of the designs.

KEYWORDS: ABV, PSL, RTL, virtual verification, physical verification

The complexity of circuits has grown to such a degree that the verification can consume up to 70% of the design cycle. Productivity of logic designers has been enhanced thanks to assertion-based verification (ABV) which allows verifying temporal design properties. Such properties can be checked using either static verification (e.g. formal proving) or dynamic verification (e.g. simulation).
SLASH, the bundle of SLED with SMASH, allows designers to conveniently perform assertion-based verification through simulation. Furthermore, SLED SDG (Synthesizable Detector Generator) enables generating synthesizable verification units that can be embedded in a prototype or in the circuit itself.

KEYWORDS: Mixed signal, power consumption, conversion, ModelSim, compatibility

Mixed-signal IC and SoC design often implies interaction between different EDA solutions. This interaction (or even cosimulation) requires that simulators be interfaced through “backplanes” with cumbersome assembly of netlists, or explicit declaration of interfaces.
In this context, SMASH™ provides an easy to use feature that reduces and eases the time spent to convert a ModelSim based circuit description to a SMASH™ and SCROOGE readable format. This conversion is straightforward and very useful to simulate mixed-signal designs or to analyze the power consumption of complex digital or mixed-signal chips.

KEYWORDS: SPICE, netlist compatibility, flavor, accurate, solver

SMASH is a powerful circuit simulator that provides compatibility with most SPICE flavors and SPICE-like netlist syntaxes with an accurate analog solver. You can work with your preferred netlist syntax thanks to the FLAVOR directive of SMASH and you can easily transfer netlists from Hspice to SMASH in order to benefit from unique features of SMASH.

KEYWORDS: System C, behavioral, system level, VHDL

The SystemC description language is more and more frequently used to model and simulate new hardware designs at the system level. The need to integrate system level models into heterogeneous multi-level simulations is essential, thereby filling the design gap between system level design and HDL or SPICE design.
SMASHTM provides a SystemC model wrapper (C++) allowing the integration of SystemC models into a co-simulation with SMASHTM.

KEYWORDS: Spice library, encryption, protection

For most component library or model parameter set providers, protection of component or parameter descriptions is necessary. In order to protect proprietary parameters, sub-circuits, models, or netlists, and to distribute libraries to customers without revealing sensitive information, SMASH provides an encryption solution. Its ease of use enables designers to protect all kinds of SPICE like libraries and even complete designs.

KEYWORDS: analog, mixed signal, waveform, compression, post-processing

Simulation of large analog and mixed signal systems requires lots of disk space. Observation of all signals and even sometimes only a subset of them can also take a long time to load. To face these drawbacks, SMASH offers a waveform compression algorithm which allows compression at two different levels:
1. when displaying the waveforms: all signals displayed in the waveform viewer are compressed, but the master file is preserved. Post-processing performed after simulation, such as measures, remain unchanged after compression.
2. when saving simulation data: depending on a user defined accuracy setting, only part of the calculated points are saved. This data compression is lossless as long as compression tolerances respect simulation accuracy setup, in which case all measurements will stay relevant. This functionality is quite useful for large simulations, such as parametric or statistic.
This useful functionality aims at reducing disk space occupied by large simulations or accelerating parametric and statistic analyses.