On the problem of optimum signal to noise ratio in amplitude measurements