SWCAD111 Transient Analysis question

[Harry Dellamano]

I am doing a transient analysis of Genome's 3ph PFC and being a Micro-Cap SPICE jockey, SWCAD111 is being a little stubborn.

I want to run an initial simulation to settle all nodes for 60mS as it does just fine. Then I want it to leave all values in the circuit and easily change the simulation time, let's say for 0 to100uS so I can view other nodes in the circuit. I would like to go back and do many fast (100uS) sweeps, at different nodes without having it start from T=0, leaving the last sweeps values (V+I) for starting values.

Sweeping from 0 to 60.100ms and viewing from 60.000mS to 60.100mS takes too long and too much data.

Micro-Cap allows you to restart all analysis from the previous data points very quickly. Is there a way with SWCAD-3?

Tanks,

Harry

 

[Jim Thompson]

I am doing a transient analysis of Genome's 3ph PFC and being a Micro-Cap SPICE jockey, SWCAD111 is being a little stubborn.

I want to run an initial simulation to settle all nodes for 60mS as it does just fine. Then I want it to leave all values in the circuit and easily change the simulation time, let's say for 0 to100uS so I can view other nodes in the circuit. I would like to go back and do many fast (100uS) sweeps, at different nodes without having it start from T=0, leaving the last sweeps values (V+I) for starting values.

Sweeping from 0 to 60.100ms and viewing from 60.000mS to 60.100mS takes too long and too much data.

Micro-Cap allows you to restart all analysis from the previous data points very quickly. Is there a way with SWCAD-3?

Tanks,

Harry

PSpice can do that. Knowing Mikey I'd expect his code to do the same.

...Jim Thompson

 

[Genome]

"Harry Dellamano" [email protected] wrote in message

I am doing a transient analysis of Genome's 3ph PFC and being a Micro-Cap
Tanks,
Harry

Don't Know, but if you click Help About in SWcadIII then you can ask the
question directly. I just did so you might get an answer sometime.

Last time I asked it was my fault so I e-mailed again to explain my error
and Mike replied.......

Cheers

DNA

 

[Genome]

"Harry Dellamano" [email protected] wrote in message


Don't Know, but if you click Help About in SWcadIII then you can ask the
question directly. I just did so you might get an answer sometime.

Last time I asked it was my fault so I e-mailed again to explain my error
and Mike replied.......

Cheers

DNA

From the help files....

..SAVEBIAS -- Save Operating Point to Disk

Syntax: .savebias <filename> [internal]

+ [temp=<value>] [time=<value> [repeat]] [step=<value>]
+ [DC1=<value>] [DC2=<value>] [DC3=<value>]

This command writes a text file to disk that is reloaded with a .loadbias
command in a subsequent simulation. If you have a circuit that has a
difficult-to-solve DC operating point, you can save that solution to disk so
that the next analysis can save time finding the DC solution before
proceeding to the rest of the simulation.

The keyword "internal" can be added to indicate that the internal nodes of
some devices should also be kept so that a more complete version of the DC
solution is kept.

If you want to save a particular DC operating point from a .tran analysis,
you can give specify a time. The first solved time point after the
stipulated time will be written. The modifier "repeat" will cause the DC
solution to be written after every period specified by this time. The file
will contain only the most recently solved DC point. DC1, DC2, and DC3 can
be given to extract a single operating point from .dc sweep analysis.

The savebias command writes a text file in the form of a .nodeset command.
Note that nodeset statements are only recommendations of the solution. That
is, the solver will start iterating the solution with the node voltages
given in the nodeset statements, but will continue iterating until it's
satisfied that the solution is valid. If you want to restart a .tran
solution from the DC operating point, you can edit the file from a .nodeset
to a .ic to try to coercive the solver to start from this DC state.

Since the integration state of all the circuit reactances isn't saved in the
..savebias file, success with this technique varys.

 

[Harry Dellamano]

"Harry Dellamano" [email protected] wrote in message


Don't Know, but if you click Help About in SWcadIII then you can ask the
question directly. I just did so you might get an answer sometime.

Last time I asked it was my fault so I e-mailed again to explain my error
and Mike replied.......

Cheers

DNA

From the help files....

.SAVEBIAS -- Save Operating Point to Disk

Syntax: .savebias <filename> [internal]

+ [temp=<value>] [time=<value> [repeat]] [step=<value>]
+ [DC1=<value>] [DC2=<value>] [DC3=<value>]

This command writes a text file to disk that is reloaded with a .loadbias
command in a subsequent simulation. If you have a circuit that has a
difficult-to-solve DC operating point, you can save that solution to disk
so that the next analysis can save time finding the DC solution before
proceeding to the rest of the simulation.

The keyword "internal" can be added to indicate that the internal nodes of
some devices should also be kept so that a more complete version of the DC
solution is kept.

If you want to save a particular DC operating point from a .tran analysis,
you can give specify a time. The first solved time point after the
stipulated time will be written. The modifier "repeat" will cause the DC
solution to be written after every period specified by this time. The
file will contain only the most recently solved DC point. DC1, DC2, and
DC3 can be given to extract a single operating point from .dc sweep
analysis.

The savebias command writes a text file in the form of a .nodeset command.
Note that nodeset statements are only recommendations of the solution.
That is, the solver will start iterating the solution with the node
voltages given in the nodeset statements, but will continue iterating
until it's satisfied that the solution is valid. If you want to restart a
.tran solution from the DC operating point, you can edit the file from a
.nodeset to a .ic to try to coercive the solver to start from this DC
state.

Since the integration state of all the circuit reactances isn't saved in
the .savebias file, success with this technique varys.

Are we saying I cannot just check a box for the starting point of any
sweep?
Possible starting points should be; Zero, File, Leave (from previous run)
I my SPICE work, 90% start from Leave, the remaining 10 % composed of Zero
and File. This saves a ton of time in power supplies where time constants
vary from 10nS to 100mS and 10nS rise times must be analyzed.
I know that sweep can be expanded to see small time spans but if you have
many waveforms on different V+H scales, that can be very difficult without
the Leave feature.
Confused,
Harry

 

[Genome]

Are we saying I cannot just check a box for the starting point of any
sweep?
Possible starting points should be; Zero, File, Leave (from previous run)
I my SPICE work, 90% start from Leave, the remaining 10 % composed of
Zero and File. This saves a ton of time in power supplies where time
constants vary from 10nS to 100mS and 10nS rise times must be analyzed.
I know that sweep can be expanded to see small time spans but if you have
many waveforms on different V+H scales, that can be very difficult
without the Leave feature.
Confused,
Harry

OK, I've tried this. Yes it looks like you cannot just check a box.
Sometimes life is hard.......

There is one caveat I have seen. Although you can use this method you have
to pick a moment in time when the 'major' time dependant voltage/current
sources in the circuit have the same values as they have at t=0. This is
because the 'engine' will reset them to these values when you 'restart' the
analysis.

With 50Hz phase voltages your choice of 60mS is a good one, perhaps you know
more than you are giving away(?).

1) Run your first analysis with something like...

..savebias test60m.dat time=60m
..tran 0 61m uic

2) Edit the resulting file and change .nodeset to be .ic

3) Run your subsequent analysis with

..loadbias test60m.dat
..tran 0 100u

Don't set uic, the loadbias takes care of that.

This seems to work.

Cheers

DNA