The dreaded time step to small error

[ronv]

I'm trying to model a simple PWM and get this error.
I've seen it before with zeners. The thing that is interesting is that with the pot set at .5 it runs (10K pot), but at .8 or so it fails.
Any ideas??

 

[crutschow]

I ran you simulation and got the "step size too small" error once, but then I couldn't get it to repeat. Apparently just one of those quirks of the model/simulator.

 

[eTech]

I'm trying to model a simple PWM and get this error.
I've seen it before with zeners. The thing that is interesting is that with the pot set at .5 it runs (10K pot), but at .8 or so it fails.
Any ideas??

One thing I noticed is that R8 is "1meg". change it to "1MEG". There is also an unconnected dangling wire off to the right of schematic.
I made the changse but still received time step error..

 

[MikeMl]

Ran for me using ver 4.19v, with default sim settings. Sometimes the "alternate solver" works...

 

[alec_t]

I frequently get the 'time step too small' error in my sims. As you found, changing a component value sometimes cures it.
I'd be interested to know if there's any setting in Control Panel which can increase the time step and guarantee a fix? I've no real understanding of what most of the settings do (e.g. Stol, MinDeltaGmin ...) although I tinker with them sometimes in a vain attempt to unstick a sim.

 

[eTech]

This LTspice directive seems to fix it. Add it as a directive to the schematic and the analysis will converge.
Use the Normal resolver.

.options gmin=1e-10

What it does:
Add a small conductance of 1e10(=10GOhm) parallel to every diode of transistors and diodes.

eT

 

[eTech]

I frequently get the 'time step too small' error in my sims. As you found, changing a component value sometimes cures it.
I'd be interested to know if there's any setting in Control Panel which can increase the time step and guarantee a fix? I've no real understanding of what most of the settings do (e.g. Stol, MinDeltaGmin ...) although I tinker with them sometimes in a vain attempt to unstick a sim.

The time step setting is adjusted in Transient settings "Maximum Timestep".
The way I understand this, if you break the analysis down into time slices where each time slice represent an analysis sample, the time step setting reflects the length of each sample.

As far as I now, there's no settings that can guarantee convergence...

eT

 

[ronv]

Ahhh. I fell better all over. It fixed that circuit and the more complicated one. I read something about this but didn't know how to do anything about it. I'll write it down with my pass words so I don't loose it! Thanks!