Netlist/Options

Q: Determining the actual values used for simulation.

A: To determine the actual options used for a simulation, include the following line in your netlist:

.OPTION OPTS

This forces Star-Hspice to print out the options and corresponding values used in the simulation to stdout. Emphasis is generally redirected to a .lis file.

If you would like to find the actual element parameters used in the simulation, then you can use the element templates as described in Chapter 8, Element Template Output in the 2000.4 Star-Hspice Manual .

Solution: 143

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: My netlists, which contain both IBIS elements and .ALTER statements, do not always complete successfully or the answers are odd.

A: This was corrected in Star-Hspice v1999.4 release and after.

Solution: 147

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: Can I use more than one interface option in my netlist?

A: No, the first interface option will be overwritten by the second. If you first set:

.OPTION CSDF

and later in the netlist set:

.OPTION POST

Star-Hspice first writes all the outputs in Viewlogic format and later the outputs are overwritten in AvanWaves format. You cannot see the output in Viewlogic but you can see them in AvanWaves.

Solution: 208

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: Why doesn't .OPTION SCALE work in my simulation?

A: See this solution in Models.

Solution: 209

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: Why does my extracted netlist fail in Star-Hspice?

A: To use an extracted netlist, do a global search for the colon symbol (:). Replace this character globally with some unique pattern that is not likely to be in the netlist already. "_c_" is a possible replacement for the colon symbol.

Solution: 342

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: What is the magnitude difference between the NORM and UNORM options when using windowing in the .FFT statement?

A: If you use the option UNORM instead of the default NORM setting in your .FFT statement, you will see a difference in the signal magnitude. The reason for this is because of the equations that govern the different windows. The equations are given in Window Weighting Characteristics in FFT Analysis in the 2000.4 Star-Hspice Manual .

The difference in magnitude is due to "Coherent Power Gain." The definition for Coherent Power Gain is that it is a measure of the reduction in signal power due to the window function suppressing a coherent signal at the end of the measurement interval. This value is really given for each window in the equation column (it is not obvious in all cases). Here is a table to show what the coherent power gain is for each case.

Solution: 351

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: How can I find out what components are connected to the same node?

A: The option you probably need is: .OPTION NODE. This option creates a node cross-reference table that can be printed. The table lists each node and all the elements connected to it. The terminal is indicated by a code, separated from the element name with a colon (:). The codes are:

Solution: 360

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: In PETL, what is the significance of GRIDFACTOR?

A: The program uses the predefined number of segments according to the specified accuracy mode. In some cases, users may want higher accuracy than they can specify using the accuracy mode. Then they can double or triple the number of the predefined segments by setting GRIDFACTOR to 2 or 3.

Solution: 392

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: Which integration method (TRAP or GEAR) is more accurate?

A: The TRAP (trapezoidal) method is faster and more accurate than GEAR. The GEAR method is used to remove the oscillation that can occur due to the TRAP algorithm. Circuits that are nonconvergent with TRAP will often converge with GEAR. The GEAR method is suitable for the circuits that are inductive in nature, such as switching regulators.

Solution: 409

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: Why doesn't losstangent have any affect on my W Element?

A: For the dielectric loss matrix to be computed and included in the simulation, ".FSOPTIONS computegd" has to be set in the SPICE deck.

Solution: 505

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: I have 64 .ALTER statements but only get 36 .mtx files. Why?

A: By default, Star-Hspice only generates 36 unique output files for .tr*, .mt* etc. After 36 .ALTER statements, Star-Hspice starts from the beginning and writes over *.mt0 on up. To change this you can use either:

.OPTION ALT999

or:

.OPTION ALT9999

The first option lets Star-Hspice write 999 files before going back to the 0th file and alt9999 lets you write 9999 output files.

Solution: 511

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Q: What does the option MU do? The manual just says it is a trapezoidal integration parameter.

A: The option MU acts as both a flag and a parameter. The equations for trapezoidal integration and backward-Euler integration are very similar to each other. The difference between these two is the parameter MU. By setting MU to 0.5 (the default), Star-Hspice uses trapezoidal integration. If MU=0, then Star-Hspice uses backward-Euler integration. If you set MU to a value between 0 and 0.5, then you get a blend of both methods.

Here is the order of the three integration methods available in Star-Hspice (trapezoidal, backward-Euler, Gear) from most accurate to most stable:

Solution: 556

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *