The upper limit on the deepest step of the surface shape that can be accurately determined is an important performance measure associated with the fringe projection profilometry. This metric is evaluated as the variance of height between two adjacent pixels on a fringe patterns reflected from the object surface. This paper presents novel results on this metric based on the Nyquist sampling theorem originally developed in the area of communication theory. Compared to existing results, we indicate that the fringe width and digital image resolution also affect the height variance range significantly. This new result could be used to increase the measurement range for projection system.