A difficult and challenging problem not only to bayesian but also to those interested in maximum likelihood estimation is to express the precision matrix of the second order moving average processes in closed analytic from in terms of the parameters directly. The main objective of  this  article is to develop a convenient technique to  obtain  such closed analytic  from.the proposed technique is based on approximating  the covariance structure of the first two observations only. Then a homogeneous difference  equation  of  the transformation  matrix  and  an  exact solution  for  the  difference  equation  is  given  in  a  closed  and easily  computed from.