Sandia National Laboratories develops and maintains code and models for its own internal Simulation Program for Integrated Circuit Engineering (SPICE). Our simulation environment is distinguished from commercial tool sets by its massively parallel capabilities and its ability to include radiation effects models. Numerous improvements have been made to its speed and model capabilities, and models continue to be updated with ongoing Verification and Validation data. Uncertainty modeling is included, and Verilog-AMS, Multi-Time PDE, and Harmonic Balance capabilities are being developed. Sandia has also developed a framework for analog/mixed-signal capability and has demonstrated its utility to critical system elements. This capability is critical in enabling simulations of VHDL designs against analog models and observing the interactions in waveforms. Signal integrity analysis capabilities are also regularly employed. Sandia is leading the way in developing models for multi-parameter aging effects for modeling electrical performance. These aging effects include numerous environmental influences including temperature, humidity, and radiation. Careful geometric and physical construction analysis (e.g., doping profiles) are obtained to build accurate models. Sandia has many Commercial Off-The-Shelf (COTS) and custom device models that are developed and verified. Sandia employs an extensive calibration methodology that involves parameter extraction and thorough simultaneous parameter optimization across multiple parameters and data sets. Optimization involves finding the minima of a predetermined objective function. Sandia analysts are experts in careful data review for each device type and application to determine the best optimization procedures. Device modeling from internal Sandia 2D and 3D Poisson solvers is helpful in calibrating the model physics as well. Sandia emphasizes quantifying margins of uncertainty, and has defined a proven methodology for limiting the propagation of parameters to the most critical sets. The method employs design of experiments and analysis of variance to classify primary factors for uncertainty propagation.
Sandia performs electrical modeling and simulation work to support many different systems and design efforts. Modeling of circuits of varying complexities from a few discrete elements to full-system hierarchical models has been demonstrated. Other capabilities demonstrated include electromagnetic coupling analysis and voltage droop in various structured ASICs. One method of uncertainty quantification employed uses Monte Carlo techniques for model and parameter variations to model the impact to a system. While not all sources of uncertainty may be included (such as parasitic inductances or package parasitics), a useful methodology has been demonstrated and work continues to add parasitics and other uncertainties.
Sandia analysts are experts in Model Based Performance Analysis. The method employs various model fidelities as needed to support the sensitivity and uncertainty studies of a large system response. The environment space includes temperature, radiation, and circuit bias and load conditions. A probabilistic approach is taken to account for device uncertainty, and data are acquired to reduce uncertainties and improve confidence. Large numbers of simulations are performed for all the various bias, load, temperature and radiation effects. The parallel processing capabilities at Sandia are useful for this approach. This approach has been used for assurance qualifications on several large systems, and numerous COTS circuit models have been developed to support this approach. Sandia analysts are experts at performing requirements validations as well.
For additional information or questions, please email us at Custom Components.