Overview
Robust Design is a proven development philosophy focused on improving reliability to the point of zero defects. Approaching this aggressive goal requires that Robust Design principles be an early and integral part of the development cycle. The objective is to make the end-product immune to factors that could adversely affect performance.

Robust Design requires that the following four factors be considered in the design process: signal, response, noise, and control.

Noise factors are disturbances that cause the systems response to shift from specification. These factors are likely beyond the designer’s control, such as manufacturing tolerances, aging, usage patterns, environmental conditions, etc. Noise factors must be identified and quantified so that accurate choices can be made about which effects require compensation. Control factors are used by the designer to compensate for noise factors that could significantly influence the system away from nominal performance.

Once the critical noise factors are identified and the control factors selected, a Robust Design flow is used to implement and analyze the design to ensure system reliability. The objective of a Robust Design flow is to meet performance requirements with the highest possible system reliability and the most reasonable systems cost.

Robust Design Flow
Adopting Robust Design principles to improve reliability means making system performance immune to variations in design technologies, component parameters, manufacturing processes, and environmental and operational conditions. In a Robust Design flow, these variations become the noise factors affecting system performance. The method of control for each variation may be as simple as selecting high precision components or as involved as implementing new control algorithms. The matrix of possible variations and control combinations becomes so complex that the traditional design-prototype-test flow is not practical. Designers must move their design activities to the virtual world, where powerful simulation tools support complete system design and verification using Robust Design techniques as outlined in the following steps

Nominal Design
The first step in a Robust Design flow is to complete the system’s nominal design, which becomes the target for the remaining analysis. The system must perform within specification under nominal conditions.

Sensitivity Analysis
Following the nominal design stage, a sensitivity analysis is performed. The designer must determine which parameters have the most effect on system performance. Once the effects of each parameter variation is calculated, the designer then analyzes the data and selects which parameters to focus on during the rest of the design process.

Parametric Analysis
A parametric analysis allows a designer to fine-tune the component parameters that most affect system performance. The objective is to vary specific parameters over a limited range in order to determine the set of values that best meet performance specifications. Once parameter values are selected, it is important to also verify performance over a range of environmental conditions.

Statistical Analysis
A statistical analysis investigates how random combinations of parameter values can affect system performance and reliability. A series of simulations are performed, with parameter values randomly changed between each simulation run. Hundreds or even thousands of runs may be required to get statistically meaningful results.

Stress Analysis
During a stress test, the system is simulated to see if meeting performance specifications pushes components beyond their safe operational limits. Component parameters are assigned maximum ratings which are monitored to see if they are exceeded.

Failure Modes Analysis
The final step in a Robust Design flow is determining how the system will perform when individual components fail. Failure mode requirements are often mandated in the design specification and must be verified during the design process.

Choosing the Right Tool
Implementing an effective and efficient Robust Design process requires simulation tools with specialized capabilities. The key tool requirements are simulation support, model library support, modeling language support, and advanced data analysis. A simulator must have special, built-in capabilities for each of the steps in the Robust Design process.

Saber Advantages

• Efficient implementation of all Robust Design analyses through a single simulation environment
• Quickly create a virtual system design supported by more than 30,000 behavioral and characterized models
• Perform extremely compute intensive statistical analyses with grid computing
• Create models quickly and accurately with model characterization tools
• Increase design portability through model language standards MAST & VHDL-AMS
• Protect intellectual property with model encryption

This summary was adapted from the article, Improving System Reliability Using the Saber Simulator in a Robust Design Flow