You need to approach this more systematically. Changing the interest rate rule mechanically will change the roots of the difference equation so that the model runs. But that does not mean that the model makes sense. In the paper you referenced, the inflation feedback needed to be bigger than 1 in order to satisfy the Taylor principle. For some reason in your model, the feedback must be outside of the usual range. That sounds as if there is a more fundamental problem in your model and changing the interest rule simply looks as if it were the solution.