Test and Measurement 1623ment can be remedied by tapering the energy at the
beginning and end of the window to zero using a mathe-
matical function. A number of data window shapes are
available for performing the smoothing.
These include the Hann, Hamming, Blackman-
Harris, and others. In the same way that a physical
polishing process removes some good material from
what is being rubbed, data windows remove some good
data in the process of smoothing the discontinuity. Each
window has a particular shape that leaves the data
largely untouched at the center of the window but tapers
it to varying degrees toward the edges. Half windows
only smooth the data at the right edge of the time record
while full windows taper both (start and stop) edges.
Since all windows have side effects, there is no clear
preference as to which one should be used. The Hann
window provides a good compromise between time
record truncation and data preservation. Figs. 46-22 and
46-23 show how a data window might be used to reduce
spectral leakage.
46.3.5.10 A Methodical Approach
Since there are an innumerable number of tests that can
be performed on a system, it makes sense to establish a
methodical and logical process for the measurement
session. One such scenario may be as follows:
- Determine the reason for and scope of the measure-
ment session. What are you looking for? Can you
hear it? Is it repeatable? Why do you need this
information?- Determine what you are going to measure. Are you
looking at the room or at the sound system? If it is
the room, possibly the only meaningful measure-
ments will be the overall decay time and the noise
floor. If you are looking at the sound system,
decide if you need to switch off or disconnect some
loudspeakers. This may be essential to determine
whether the individual components are working
properly, or that an anomaly is the result of interac-
tion between several components. “Divide and
conquer” is the axiom.
Figure 46-21. Increasing the length of the time window increases the frequency resolution, but lets more of the room into
the measurement, SIA-SMAART.
Figure 46-22. The impulse response showing both early
and late energy arrivals.Interval 1: Early energy
Interval 2: Late energyTimeAmplitude1 2400