Learning Forum

[Mariah Ha]

Hi all, 

I have a simulation in Charge( small signal analysis) which is diverging for a long time. I tried every single solution ( reducing the mesh size, reducing the initialization step, more iterations, kkt, ....) which are in the charge guide for divergence issues. 

This is the error I am getting. What type of error is this? I am not sure why it is saying also that the mesh is not adequately refined, I tried different mesh even the tiny ones.

[Guilin Sun]

This error indicates the automeshing is not well resolving the geometry/doping/high-E-fields so the equation is not solved correctly. Think this way, using a Newton method to solve an question numerically, the error is oscillating and has not reached a desired tolerance.

The message in yellow shows that "after 0 interation", meaning the solver may not start the interation. So pleae check your device configuration and make sure partition works fine before simulation. If there is a curved surface on a flate surface, you may consider to have a few nano meter overlap to avoid perfect geometry; you may also reduce the element quanlity, modify some tolerances, and use more number of iteration than the default one, if it is truely the divergence issue.

 

We will suggest to check this article   

1.      Troubleshooting convergence errors in CHARGE

Other than this, we do not have further suggestions.