When permanent shock is large, how to solve perfect foresight model

Dear all,
I am trying to use dynare to solve a perfect foresight model. When the permanent shock is small, such as from 0.4 to 0.41 for a parameter, dynare can solve the model. But when the shock is larger, such as from 0.1 to 0.41, dynare failed to solve perfect foresight model. I‘ve tried stack_solve_algo from 1 to 7, but it failed. I really don’t know how to solve it .
(Attached is my code)
Thank you very much
solve_ss1.m (1.1 KB)
steady_state2.m (2.6 KB)
tran_path.mod (5.3 KB)

There is something strange going on here. As soon as tau becomes smallerr than about 0.36 the solver fails. Maybe it has something to do with the economics of the problem. It seems some variables that can only be positive hit 0 during the adjustment path