An experimental system has been developed for in situ measurement of the acoustical impedance of surfaces using a single microphone, pseudorandom sequences, and transfer-function techniques. It is well known that the impedance determined from two transfer functions is very sensitive to phase errors. The biased impedance has been expressed as a function of equivalent errors [J.-F. Li and J.-C. Pascal, J. Acoust. Soc. Am. 99, 969--978 (1996)]. In the system used here only one channel is used; thus the error does not include the electronic phase mismatch. However, since no experimental system is perfectly linear and repeatable, an equivalent phase error occurs. This error can be represented as the difference in phases between two average transfer functions measured sequentially at two locations. The objective of this work is to show how this error affects the impedance. The error causes a large bias at low frequencies. The results suggest that to reduce errors the distance between the two microphone locations should be large enough at low frequencies, and the number of transfer functions averaged should be as large as possible.