User manual

page 84
Analog System Lab Kit PRO
appendix B
Simulation models are very useful in the design phase of an electronic system.
Before a system is actually built using real components, it is necessary to perform
a‘softwarebreadboarding’exercisethroughsimulationtoverifythefunctionality
of the system and to measure its performance. If the system consists of several
building blocks B1, B2, ..., Bn, the simulator requires a mathematical representation
of each of these building blocks in order to predict the system performance. Let us
consideraverysimpleexampleofapassivecomponentsuchasaresistor.Ohm’slaw
can be used to model the resistor if we intend to use the resistor in a DC circuit. But
if the resistor is used in a high frequency application, we may have to think about
the parasitic inductances and capacitances associated with the resistor. Similarly,
the voltage and current may not have a strict linear relation due to the dependence
oftheresistivityontemperatureofoperation,skineect,andsoon.Thisexample
illustrates that there is no single model for an electronic component. Depending
on the application and the accuracy desired, we may have to use simpler or more
complex mathematical models.
We will use another example to illustrate the above point. The MOS transistor,
which is the building block of most integrated circuits today, is introduced at the
beginning of the course on VLSI design. In a digital circuit, the transistor may be
simplymodeledasanidealswitchthatcanbeturnedonorobycontrollingthe
gatevoltage.This model issucientifweareonlyinterestedinunderstanding
the functionality of the circuit. If we wish to analyze the speed of operation of the
circuit or the power dissipation in the circuit, we will need to model the parasitics
associated with the transistors. If the same transistor is used in an analog circuit,
the model that we use in the analysis would depend on the accuracy which we want
intheanalysis.Wemayperformdierentkindsofanalysisforananalogcircuit-
DC analysis, transient analysis, and steady-state analysis. Simulators such as SPICE
requiretheusertospecifythemodelforthetransistor.Therearemanydierent
models available today for the MOS transistor, depending on the desired accuracy.
The level-1 model captures the dependence of the drain-to-source current on the
gate-to-source and drain-to-source voltages, the mobility of the majority carrier,
the width and length of the channel, and the gate oxide thickness. It also considers
non-idealities such as channel length modulation in the saturation region, and the
dependence of the threshold voltage on the source-to-bulk voltage. More complex
models for the transistor are available, which have more than 50 parameters.
Ifyouhavebuiltanoperationalamplierusingtransistors,astraight-forwardway
to analyze the performance of the OP-Amp is to come up with the micromodel of the
OPAMAP, where each transistor is simply replaced with its corresponding simulation
model. Micromodels will lead to accurate simulation, but will prove computationally
A macromodel is a mathematical convenience that helps to reduce simulation
complexity. The idea is to replace the actual circuit by something that is simpler,
but is nearly equivalent in terms of input characteristics, output characteristics, and
feedforward characteristics. Simulation of a complete system becomes much more
simple when we use macromodels for the blocks. Manufacturers of semiconductors
provide macromodels for their products to help system designers in the process of
systemcongurationselection. The macromodelscanbe loaded intoa simulator.
B.1 Micromodels
B.2 Macromodels
intensive. As the number of nodes in the circuit increases, the memory requirement
will be higher and the convergence of the simulation can take longer.
A macromodel is a way to address the problem of space-time complexity mentioned
above. In today’s electronic systems we make use of analog circuits such as
operationalampliers,dataconverters,PLL,VCO,voltageregulators,andsoon.
The goal of the system designer is not only to get a functionally correct design,
but also to optimize the cost and performance of the system. The system-level
cost and performance depend on the way the building blocks B1, B2,..., Bn have
been implemented. For example, if B1 is an OP-Amp, we may have several choices
ofoperationalampliers.TexasInstrumentsoersalargenumberofoperational
ampliers that a system designer can choose from.Referto Table B.1. As you
willsee,therearecloseto2000typesofoperationalampliersavailable!These
arecategorized into 17 dierentbins tomake the selection simpler.However,
you will notice that 240 varieties are available in the category of Standard Linear
amplifers! How does a system designer select from this large collection? To
understandthis,youmustlookatthecharacteristicsofastandardlinearamplier
- these include the number of operational ampliers in a single package, the
GainBandwidth Productoftheamplier,theCMRR,Vs(min),Vs(max),and so
on. See http://tinyurl.com/ti-std-linear. The website allows you to specify these
parameters and narrow your choices.
But how does one specify the parameters for the components? The overall
system performance will depend on the way the parameters for the individual
components have been selected. For example, the gain-bandwidth product of an
operationalamplierB1willinuenceasystem-levelparametersuchasthenoise
immunity or stability. If one has n components in the system, and there are m
choices for each component, there are m·npossiblesystemcongurations.Even
if we are able to narrow the choices through some other considerations, we may
stillhavetoevaluateseveralsystemcongurations.Performingsimulationsusing
micromodels will be a painstaking and non-productive way of selecting system
congurations.