Extended measurement options

General Parameters

FFT size/cycle length

Number of samples used for frequency analysis by FFT. Therefore it must be a power of two.

This is also the number of samples after that the reference signal repeats itself. This implies that all frequencies used for measurement are multiples of the cycle time = cycle length / sampling rate.
Reasonable values depend on the room you intend to measure. Typically 262144 is enough for standard measurements. But if you have long reverberation you need to increase the value.

Discard first cycles

After pressing Start the first n samples of the test recording are discarded. This is required to give the room a chance to reach a steady state. It is also useful to skip glitches at the start of the recording. Recommended values are in the order of one or two cycle times.

Frequency bin factor

Join neighbor frequencies in the result file if they fit into an interval [f, f*(1+binfactor)]. The gain and the group delay will be averaged for all frequencies in the bin.

Frequency range

This is the range of frequencies that is used for the measurement. Outside this interval the test signal contains no energy.
This also controls the X axis of the graphs.

Display settings

Frequency response gain

Range of the left Y axis of the frequency response graph.

Frequency response delay

Range of the right Y axis of the frequency response graph.

Record level indicator range

Set the level of the left boundary of the record level indicators.

Record level indicator limits

Levels where the color of the record level indicator turns from green to yellow and red respectively.

Reference signal

Energy distribution exponent

This is the weight exponent used for the relative intensity of different frequencies.
I ~ f^exponent
Reasonable values are in the range [-0.25, 0].

Minimum factor to next frequency

To limit the phase noise it could be helpful to set this factor to a value slightly larger than 1. This causes the energy of frequencies closer than the given factor to be collected in bins with are at least this factor away.

Line frequency notch

Although good equipment will not be sensitive in practice often the power line frequency and its harmonics appear as noise. Selecting a nonzero value will eliminate the line frequency and the selected number of harmonics from the measurement. Selecting 1 removes only the line frequency itself, 2 removes also the double line frequency.
The drawback is that high group delays cannot longer be handled correctly at the line frequency.
Reasonable values are 0 if you do not need this work around and 3 to remove up to the 3^rd harmonic. Of course, you need to setup the line frequency of your country correctly for this to work. Usually 50 or 60 Hz.

Use only odd frequencies in spectrum

If this option is checked all frequencies that fit an even number of times in the cycle time are not used. This causes second order intermodulations, i.e. sums and differences of other frequencies, always to appear at even frequencies that do not contribute to the result.

Skip some frequencies in stereo mode

When entering stereo mode every second frequency is used for each channel. This introduces a unwanted symmetry into the autocorrelation function of the test signal. When this option is enabled some randomly chosen frequencies in the spectrum are eliminated. This breaks the symmetry and enables the measurement to deal with twice as large delays - the same as in single channel mode.