There are two poles. one at 1 and the other at 2. Both of the there will cause what ever it is to ''blow up" or go to infinity shortly.This isn't really a PLC problem, is it?
Integrating the differential equations using Runge-Kutta is better because the differential equation can be non-linear and it is easier to model disturbances.In general, you need to convert the differential equation into an approximately equivalent difference equation.
I never had that problem because I know where to place the closed loop poles and the sample times on a motion controller are fast and deterministic.Beware that higher order equations may exhibit oscillatory response (limit cycling) when converted to discreet time form when the the continuous time form was stable. For this reason it is better to design the system with discreet time/difference equations from the outset.