Lambda SR1, Operation Manual, Page: 133 / 317

SR1 Operation 133
© 2014 Stanford Research Systems
Window Selection
It is well known that the application of a window functions is typically necessary to obtain maximum
dynamic range for FFT measurements. The discrete Fourier Transform implemented by SR1 calculates a
spectrum assuming the time record repeats continuously. Thus, if the signal being analyzed is not
perfectly repetitive in the time record interval the the calculated spectrum will include the discontinuities
between the beginning and end of the time record which show up in the frequency domain as wide skirts
and a high noise floor around the actual spectrum. The spectrum shown below on the left is the
unwindowed spectrum of two sinewaves of arbitrary frequency. The spectrum is completely distorted by
the artifacts associated with the time record discontinuities.
Unw indow ed Spectrum
Spectrum w ith Blackm a Window
To eliminate the discontinuities between the beginning and end of the time record the time record is
typically multiplied by a window function which is large in the middle of the time record and tapers off
near the beginning and end of the time record. The spectrum on the right shows the spectrum of the
same two sine waves after application of a window function.
Many window functions have been developed over the years, but few have the >120 dB dynamic range
necessary for modern audio measurements. For most measurements with SR1, the
Rife Vincent 4-