Stiff ode solvers failure

I have a stiff system of nonlinear 1st order ODEs represents the equation of motion of a large dynamic system, can be expressed mathematically in the following form:

• {r_dot}= [A] {F(r)}

{r_dot} and {F(r)} are vectors of size n. F is a function of the system states (r). A is a symmetric square matrix of size n*n, I have used all the stiff ode solvers (ode15s, ode23s, ……..) to solve for the system response for t=[0 200] seconds, however the execution stopped at t=22 seconds with warning:

Warning: Failure at t=2.204902e+001.  Unable to meet integration tolerances without reducing
the step size below the smallest value allowed (5.684342e-014) at time t.

Is there any idea to fix this problem like modifying the ode options or mass matrix, Jacobin or any other suggestions?. Are there any implicit integration techniques can be used to solve this system either with a better computational stability?

[Mike Hosea]

I don't think the question can be answered in such generality. Are all the solvers giving up at about the same t value or are they just making it to various, quite different t values before failing?