"percent" Flag in Monte Carlo Simulation

stub

TSMC model: difference between nch and nch_mac

Check Monte Carlo Simulation Simulation Netlist in ADE (IC5141)

Get Monte Carlo simulation netlist in IC5141

The Monte Carlo portion of the netlist is shown below:

mc1 montecarlo numruns=100 seed=1 variations=mismatch donominal=yes \

    scalarfile="../monteCarlo/mcdata" paramfile="../monteCarlo/mcparam" \

    saveprocessparams=yes processparamfile="../monteCarlo/processParam" \

    processscalarfile="../monteCarlo/processData" savefamilyplots=yes \

    sampling=lhs {

        dcOp dc write="spectre.dc" maxiters=150 maxsteps=10000 annotate=status

        dcOpInfo info what=oppoint where=rawfile

        modelParameter info what=models where=rawfile

        element info what=inst where=rawfile

        outputParameter info what=output where=rawfile

        designParamVals info what=parameters where=rawfile

        primitives info what=primitives where=rawfile

        subckts info what=subckts  where=rawfile

        export noExprs=oceanEval("0")

}

Mismatch Simulation

When simulating with mismatch, spectre will generate a random value for each of the varied global parameters for each subckt instance. 

There is a commonly mistaken belief that you must use inline subckts for mismatch simulation. In fact, the subckt does not need to be inline, that is merely to make back-annotation of operating point data simpler because the result database does not contain an additional level of hierarchy for the inline device.

Pick a Particular Iteration from Monte Carlo Simulation Result

value(VT("/myOut")   'iteration <iterationNumberHere>)

Specifying Parameter Distributions in Spectre -- Use Statistics Blocks

Statistics block can contain:

• One or more process blocks;
• One or more mismatch blocks;
• One or more correlation statements;
• A truncate statement (for generating truncated distributions);

Process blocks:

• sampled once per Monte Carlo run

Mismatch blocks:

• sampled once per subcircuit instance

Correlation statements:
Two types of correlation statements that you can use:

• Process parameter correlation statements
• Instance or mismatch correlation statements (matched devices)

Sample Monte Carlo Analysis

mc2 montecarlo donominal=no variations=mismatch seed=1234 numruns=200 {

      dcop2 dc

      tran2 tran start=0 stop=1u

      export slewrate=oceanEval("slewRate(v(\"vout\"),10n,t,30n,t,10,90 )"

}

Notes:

1. Monte Carlo analysis statement applies only mismatch variations (variations=mismatch);
2. It does not perform a nominal run (donomial=no).