36
8
Methods of optimising measuring results
The location is based on the correct determination of the time-
delay difference. If the leakage noise is sufficiently loud and if
there are no background noise sources, the time-delay differ-
ence is adequately well-displayed after only a few averaging
processes (4 to 16).
But what can be done if no sufficiently clear peak can be gener
-
ated? The following instructions cannot substitute practice and
experience still required in difficult situations. However, they will
help you to learn the basic operations.
Be that as it may, you must always remember one thing: If the
leakage noise does not reach the microphones with sufficient
intensity, no correlation is possible!
8.1
Changing the number of averaging processes
If a result is unsatisfactory after the initial series of averaging proc-
esses, you can improve it by increasing the number of cycles. The
result is improved because the calculation is based on a greater
body of information on the chronological behaviour of the leakage
noise. However, experience has shown that the result can only
rarely be additionally improved if more than 64 to 128 averaging
processes are selected.
8.2
Using filters
Use the features made available by the mathematical filters.
Unfortunately it is hardly possible to give any generally valid
„formula“. Only your own experience, i.e., learning by doing, will
lead to success.
In most cases, it is only practical to select frequency ranges for
the correlation function in which the coherence function does
clearly stand out from its environment, i.e., where there is peak.
Individual frequencies from various interference sources generate
a uniformly sinusoidal correlation function. They can be recog
-
nised in the spectra as sharp lines.
8 Methods of optimising measuring results
