User`s guide

16
-
14 Di
g
ital Worst-Case Timin
g
Analysis
Digital worst-case timing simulation does not yield such results
without an applied stimulus; it is not a static timing analysis tool.
The level of confidence that you establish for your design’s
timing-dependent characteristics is directly a function of the
applied stimulus.
Generally, the most productive approach to stimulus definition
is to use functional testing: a stimulus designed to operate the
design in a normal manner, exercising all of the important
features in combination with a practical set of data. For
example, if you are designing a digital ADDER circuit, you will
likely want to ensure that no timing race conditions exist in the
carry logic.
To be effective, any timing simulation methodology should
include the following key steps:
Accurate specification of device delay characteristics
Functional specification of circuit behavior, including all
don’t care states or conditions
A set of stimuli designed to verify the operation of all
functions of the design
One common design verification strategy is stepwise
identification of the sections of the design that are to be
exercised by particular subsets of the stimulus, followed by
verification of the response against the functional specification.
This phase would be done using normal (not worst-case)
simulation, with typical delays selected for the elements. The
key metric here is the state response of the design. Note that
(with rare exception) this response consists of defined states and
does not include X’s.
The second phase of design verification is to use digital worst-
case simulation, reapplying the functionally correct stimulus,
and comparing the resulting state response to that obtained
during normal simulation. Differences at primary observation
points (such as circuit outputs and internal state variables),
particularly those due to X states (such as critical hazards), must
be investigated to determine their cause. Starting at those
points, use Probe in conjunction with the circuit schematic to
trace back through the network. Continue until the reason for the
hazard is located.