Hi, 

I cannot open your file, since I do not have comsol anymore. 

However when convergence is difficult i used to prescribe charge instead of voltage:

- Model one of the conductors as a very high permittivity dielectric 

- give the other conductor a zero voltage constraint

- Prescribe the charge on one node or boundary of the high permittivity dielectric

- Increase the charge as the loading paramter

- log the resulting voltage and displacement

It is very easy to find the  equilibrium path this way (including the post pull in part). 

Regards, 

Stephan 

More detail:

. https://www.researchgate.net/publication/257022122_A_palette_of_methods_for_computing_pull-in_curves_for_numerical_models_of_microsystems

Article A palette of methods for computing pull-in curves for numeri...

put comp2.u, comp2.ODE1 and comp2.V in one seggregation step, and comp2.xyz in another step

Hi

Thank you for your all replies. After applying correct scales to comp2.u, comp2.ODE1, comp2.V, comp2.xyz and starting the Auxiliary Swept at a position close to the original position and with a small step, I finally did it. 

Best regards,

Duc