MLS stands for Maximum Length Sequence. When the MLS measurement method was first introduced, the maximum length sequence was generated using shift registers. By connecting the shift registers in a special way it was found that a Maximum Length Sequence could be generated. This was the longest sequence that could be made before it repeated itself. If the number of shift registers is N the length of the sequence is 2^N - 1. The MLS signal can be used to measure any type of LTI (Linear Time Invariant) system, therefore nowadays Maximum Length Sequence measurements are used in different fields in addition to acoustics.
The MLS method gives the impulse response of the measured system using cross-correlation between input and output signals. The impulse response can be easily windowed in the time domain. Using windowing, reflections (e.g. from the walls of the a room) may easily be removed from the frequency response. Using this feature it is possible to simulate the frequency response of a loudspeaker measured in an anechoic room.
The number of averages determines how many times the MLS is repeated during the measurement. The reason for averaging is that it will decrease uncorrelated noise and thus increase the quality of the measurements. For each time the number of averages is doubled, the signal-to-noise ratio is theoretically increased with 3 dB. Note that the total measurement time increases when the number of averages increases. Be also aware of that if the system you are measuring is somewhat time-variant, a long measurement time is not good. An example of a time-variant system is a concert hall where people are moving around. Try to avoid using MLS during such situations, we recommend selecting Single sweep as excitation signal. For room acoustical measurements, a measurement time of more that 60 seconds is seldom necessary. Electrical or mechanical systems are usually very little time-invariant, so the long measurement times are seldom a problem.
The MLS technique has many advantages when compared with other methods of measuring the response of a system. The MLS has a flat spectrum (contains equal amount of all frequencies. The DC component is rejected and may not be measured. The signal-to-noise ratio of a MLS measurement is high and may be increased by increasing the total measurement interval. The measured distortion of the system is spread throughout the computed impulse response as spurious peaks. Every MLS sequence has his own characteristic distortion pattern. For robustness against or detection of harmonic distortion, instead of MLS, use a single sweep as excitation signal. In general, we recommend using a single sweep instead of MLS.