Solving ODE becomes unstable when using a decaying signal

I'm solving a stiff, time-dependent ode using mass matrix and a jacobian matrix with the ode15s solver. The problem is vectorized.

As long as I have signals (i.e. the time-dependent term) that rise with time there are no problems. But as soon as the signal is decaying the solver canot compute the solution anymore. This is not true for sinusoidal signals, then it works fine.

Is there any "general thing" I overlooked?

Thanks!

Answer 1

[Babak]

I had same problem before.

Try deriving the "dimensionless" form of your equations and solve those. usually if you are solving stiff ODEs in which the variables have dimensions the solver may fail.