Simulations which utilize SPICE or functional models are often
composed of one or more of the following types of analyses:
DC, AC, Transient, Monte Carlo, etc.
PCB level simulations, which usually employ IBIS behavioral models,
are also a combination of several different classes of analyses,
Let us consider some notable features of each simulation type:
All or some of the nets to be simulated are not routed. Some
simulation programs assume Manhattan routing for un-routed segments and
assign default values for characteristic impedance and propagation
delay. An undesirable property of pre-layout simulation is that
crosstalk is usually ignored. On the other hand, this class of
simulation offers a high level of flexibility. Waveforms and flight
time data can be generated and analyzed. Important objectives of
pre-route simulation include determination of optimum component
placement, trace lengths and termination types.
Verification is one of the main aims of post-route simulation. The
changes implemented during the pre-layout simulation should be carefully
evaluated, to ensure that signal integrity degradations and timing
violations have been successfully eliminated. An advantage of
post-route over pre-route simualtion is that the effects of crosstalk
coupling can be accurately evaluated.
Frequently presents the simplest type of simulation. Creation of a
NET file aimed at limiting the simulation task to the nets or buses of
interest, rather than the whole board, is highly recommended. The model
files and topology should be carefully examined for completeness and
Multiboard hierarchical cases involve two or more boards linked
via cables or connectors, such as a motherboard with plug-in expansion
cards or memory modules. System simulation can be significantly more
complex than single board analysis. It is recommended that each board
to be simulated as a "stand alone", before performing hierarchical
simulation. Simulating each board separately, prior to system
simulation, suffers from the limitation that some nets will be
incomplete (without drivers or receivers) and therefore no waveforms can
be generated for those nets, but it can provide useful insight related
to the models, topology and impedances. Again, use of a NET file, to
contain the simulation to the critical sub-set, provides an effective
means to simplify the simulation task and to increase efficiency.
It appears to me that one possible way to classify SI simulations is
SI simulaions can be broadly divided as PRE-ROUTE and POST-ROUTE.
Each of the two includes SINGLE BOARD and MULTI-BOARD analyses as
subgroups. It should be noted that each simulation is usually carried
out for three corners: Maximum, Nominal and Minimum (or eqivalently,
Fast, typical and Slow) to verfiy the design under all conditions.
Furthermore, Single and Multi-board simulations in turn may contain
several types of analyses such as WHAT-IF and MONTE-CARLO, which are
briefly described below:
WHAT-IF anlysis can be carried out at any simulation stage, but it is
especially valuable during the pre-route phase. It provides a quick
method for investigating capabilities of various drivers, different
types of termination, and alterations in topology. WHAT-IF simlation
can serve as an efficient means for identification and elimination of
signal integrity and timing violations.
Finally, MONTE-CARLO is also regarded as an importnat class of
simulation. It can be utilized to determine the worst case spread
expected in a large sample of manufacured boards. This is accomplised
by a random variation of the driver strength, voltage and speed.
In concusion, an understanding of how SI simulations are classifed and
main objectives of each simulation type, is essential when formulating
an effective simulation strategy, its successful execution and
Thank you in advance for your comments.
**** To unsubscribe from si-list: send e-mail to firstname.lastname@example.org. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****