Monte Carlo/Yield Graphs
Monte Carlo analysis allows you to see the effects of component variations over a large sample of your design. While an ideal circuit with precise design values might appear to provide acceptable performance, the cumulative effect of component tolerances may make the design unusable due to excessive failures during manufacturing. Eclipse helps you to determine the repeatability of your design by providing several graphical/tabular views of the sampling results. Using the results of the analysis, you can tighten or relax component tolerances and/or performance specifications in a more informative way.

The Monte Carlo/Yield Process

The Monte Carlo process performs statistical analysis over a sampling size defined by the user. Each sample can be thought of as a normal Eclipse solution, but with values randomly assigned according to the assigned tolerance profiles. During each sample, three basic steps occur:

  1. Parameter population - Monte Carlo analysis simulates component variations based on the distribution profile assigned to a particular parameter. Using its tolerance profile, each parameter candidate is assigned a statistically random value during the sample. Available distribution profiles are discussed later.

  2. Analysis – the analysis is performed using the currently selected Graph Sheet. The output parameters specified in the graphs contained on the sheet will individually have statistics collected.

  3. Data collection – output parameter data is collected during each sample with the final results displayed as mean, standard deviation, minimum and maximum values. Furthermore, a Yield table will tally failure results relative to yield goals specified in the netlist.

Typical output:

mc1.tif (25607 bytes)