This paper studies the optimum control of multivariable systems by considering the behaviour of the two-variable on-off system described by the equation x + Ax = u where x = (x1 x2), u = (u1 u2), A = (a11 a12 a21 a22), and the drives ui =??1. The error e is defined as z - x, where z = (z1 z2) is the system desired output.