The problem stated in your example model looks a bit strange: Why would you want to modify the starting point x0 for the call to fsolve during the simulation?
I suspect that you rather want to solve a parameterised set of equations like the following
- C*x1 - x2 - exp(-x1) = 0
- -x1 + C*x2 - exp(-x2) = 0
with C being a parameter that you wish to vary throughout the simulation.
This can be done using the Algebraic Constraint block introduced in PLECS 4.1. Please see the attached model. The two Function blocks implement the left hand sides of the two equations with u(1) and u(2) representing x1 and x2 and u(3) representing the parameter C. The Algebraic Constraint block finds x1 and x2 such that the Function blocks will output 0.