The silicon pads occupy two disks at the end of the strip detector, and are used to measure dn/dy. They are arranged in equal bins in pseudorapidity and phi. As an alignment criterium one can require that the offset transverse to the beam be smaller than a fraction - say 1/10 - of the smallest pad. The smallest pad is 2x2 mm, which leads to an alignment criterion of about 0.2 mm in R.
The placement in z of the pads relative to the strips sets the scale for the y-measurements in the pads. If the interaction vertex is in the center, one could require that the intercept of a track with a pad should not shift by more than 1/10 of a pad size due to a shift in z of the pad plane. This leads to dz = 1.2 mm.
No physics agument leads to a requirement on the global placement accuracy in phi.
Clearly physics considerations do not require alignment of the pads relative to the strips that is more accurate than the approximately 0.1 mm expected to be achieved in constuction of the MVD/pad detector already.