LTSpice "World Tour"

[jackinnj]

Saw a link to this in an email from Linear:
http://www.nuhorizons.com/xpresstrack/ltcswitcher/

Looks like it's free!

 

[iko]

Thank you!

 

[jcx]

anyone planning on attending?

now if we could just get at least one person per event to ask for "Save State/Load State" functionality starting from .TRAN analysis that would allow running a .AC analysis at operating points that cannot be reached by the DC solver

the .AC analysis cannot be complete when the DC operating point won't solve for capacitor charging currents or Vdrops across circuit inductances -"dynamic" state variables that .TRAN calculates, maintains and that are useful for more complete stability analysis when these dynamic states heavily modulate the gains of the nonlinear devices (transistors) in the sim

if the .AC could take the full saved dynamic state of the sim from a .TRAN it would allow small signal transfer function analysis of stability while the output for a audio amp was slewing, entering/leaving clipping/current limiting, at dynamically determined output currents with reactive loads...

the DC solver doesn't work at all for many of my high gain circuits where the .TRAN sims fine - I can get them started by various hacks like local feedback switches that reduce gain/remove internal positive feedback for DC op point calc and then open the switches to get to the full circuit which the .TRAN solver works with after a big settling transient from the switching – obviously the initial DC state is useless for even beginning AC analysis in these instances – and yes I have tried .IC/UIC Load/SaveBias, checked “save subcircuit node/current” ect. – the DC solver simply doesn’t work well enough with really high gain circuits


I have asked Mike a few times over the years for this functionality - maybe requests coming from several directions would get more attention

 

[Eokboy]

Thanks for the heads up

 

[pjp]

jackinnj, thanks for posting that!
I attended Mike's talk, and it was MOST impressive.

There was a little for everyone.
Beginners got a stunning demo of what LTSpice is capable of (LTSpice plots compared to Oscilloscope measurements), and advanced users got tips on how to exploit the strengths of the tool and what its limitations are (including a discussion of the math algorithms behind LTSpice).

Definitely recommend this to everyone.

 

[Tazzz]

now if we could just get at least one person per event to ask for "Save State/Load State" functionality starting from .TRAN analysis that would allow running a .AC analysis at operating points that cannot be reached by the DC solver

the .AC analysis cannot be complete when the DC operating point won't solve for capacitor charging currents or Vdrops across circuit inductances -"dynamic" state variables that .TRAN calculates, maintains and that are useful for more complete stability analysis when these dynamic states heavily modulate the gains of the nonlinear devices (transistors) in the sim

if the .AC could take the full saved dynamic state of the sim from a .TRAN it would allow small signal transfer function analysis of stability while the output for a audio amp was slewing, entering/leaving clipping/current limiting, at dynamically determined output currents with reactive loads...

the DC solver doesn't work at all for many of my high gain circuits where the .TRAN sims fine - I can get them started by various hacks like local feedback switches that reduce gain/remove internal positive feedback for DC op point calc and then open the switches to get to the full circuit which the .TRAN solver works with after a big settling transient from the switching – obviously the initial DC state is useless for even beginning AC analysis in these instances – and yes I have tried .IC/UIC Load/SaveBias, checked “save subcircuit node/current” ect. – the DC solver simply doesn’t work well enough with really high gain circuits


I have asked Mike a few times over the years for this functionality - maybe requests coming from several directions would get more attention

I don't think savebias ever saves any current information, just node voltages.

Have you tried to save all note voltages / component currents and try forcing that as initial conditions for the dc solver?

 

[hvogt]

the DC solver doesn't work at all for many of my high gain circuits where the .TRAN sims fine

Do you have any examples for me to test these issues? I have just implemented a transient operating point calculation into ngspice and would need some cases for testing the code.


This is a first step using the initial conditions. When standard op determination (normal iterations, gmin stepping, source stepping) fails, there is a try using transient simulation to get the op before the intended simulation starts.