If you consider a plant of the form $G(s)=k/s^2$ and a controller of the form $C(s)=K_p+K_i /s^{\lambda}+K_d s^{\mu}$ then here are a couple of points to consider.
1. You may want to cancel the effect of the second integrator. Thus the structure of the controller is of PD type.
2. By changing the parameters $K_p$, $K_d$, and $\mu$ it is possible to satisfy some specifications in both time and frequency domains. Keeping $K_p$ small and $\mu$ close to 1 will ensure the effect of (partial) cancellation one of the poles $s=0$ of the double integrator system.
This reply was modified 8 years, 9 months ago by Aleksei.
This reply was modified 8 years, 9 months ago by Aleksei.