unsteady state calculation with adaptive time step
 

I have been using coupled solver for unsteady state calculation with adaptive time step. I find that the time step becomes very small (1x10e-08 seconds) and so it takes unusually long time to compute. Is there way to achieve convergence with a larger step size. Is linearization of the source term possible. Also can we fix the lower limit of the scalar variables i.e say it should always be positive. sm