This paper discusses the problem of preserving approximated feedback l
inearization under digital control. Starting from a partially feedback
linearizable affine continuous-time dynamics, a digital control proce
dure which maintains the dimension of the maximally feedback lineariza
ble par? up to any order of approximation with respect to the sampling
period is proposed. The result is based on the introduction of a samp
led normal form, a canonical structure which naturally appears when st
udying feedback linearization.