User Guide

Characteristics of Functions 4
-7
the units of the Lagrange multipliers are the reciprocal of the
units of their associated constraint.
Example: A multiplier associated with a width constraint has
units of meter
-1
.
Characteristics of
Functions
The success of an optimization depends highly on the behavior
of the functions related to each specification.
Generally, the PSpice Optimizer obtains values (measures
performance) for each of the specifications by evaluating a trace
which results from a simulation with varying parameter values.
The optimizer can experience difficulties if the accuracy of the
measurement is decreased by:
• Discontinuities in the simulation results.
• An error in the Probe goal function definition.
• Inaccuracies in the simulation results on which the
evaluation is based.
Generally, simulations and measurements using AC and DC
analyses behave better than simulations using transient analyses.
This is particularly true if the evaluations are set up to measure
the value of a single point (e.g., the time when a trace crosses a
specified level). This kind of measurement tends to behave
discontinuously as the parameters change, creating difficulties
for the optimizer.
To improve measurement accuracy, consider any of the
following techniques:
• Use several points rather than a single point for the Probe
goal function. That is, specify several points on a waveform
instead of a single point.
• Reduce the step ceiling in Transient analysis to produce a
more finely sampled set of data.
To review, the PSpice Optimizer
measures performance in one of
three ways: by takin
g
the value
of the sin
g
le-point trace (trace
function) in Probe, by applyin
g
a
Probe
g
oal function to the trace,
or by evaluatin
g
a PSpice
Optimizer expression. Trace
functions and Probe
g
oal
functions require a simulation;
PSpice Optimizer expressions
do not.