Maximum length sequences

The method used so far for measuring the input impulse response involves injecting a pulse into the source tube and measuring the reflections. By using a signal that continues over a longer time interval we may put more energy into the system, improving the signal to noise ratio and removing the need to average over 1000 measurements. Obviously, white noise cannot be used because we need the phase information as well as frequency response. Equivalently there must be some way of analysing the measured reflections to recover the time domain response to a single pulse.

One solution is to use a pseudo-random binary signal called a maximum length sequence (MLS). An MLS signal consists of an apparently random sequence of 0's and 1's that has a flat frequency spectrum for all frequencies up to the Nyquist frequency with the exception of the dc value. It is also computationally efficient to generate such a sequence on computer and unlike white noise an MLS signal is deterministic and therefore repeatable. The input impulse response of a system can be extracted by a cross-correlation procedure. This method of excitation has been employed frequently in measuring the input impulse response of rooms for reverberation measurement [65,66,67]. Here we discuss how MLS signals may be generated and the resulting measurements analysed. The discussion leads on to the application of MLS excitation in acoustic pulse reflectometry.