Spectrum DSP M2 User Manual page 44

Spectrum DSP M2 User Manual V1.1
tendency of a signal (say, a carrier) to "leak" onto the display above and below the
displayed frequency.
The use of "FFT Windowing" - the pre-processing of spectral data before display – can
greatly improve the visual performance of the Spectrum Scope and Waterfall display by
decreasing "bin leakage" (e.g. "sidelobe") - that is, the tendency of a signal (say, a
carrier) to "leak" onto the display above and below the displayed frequency.
The default setting is "Blackman" which is quite good in making the spectrum scope
look sharper and preventing signals from "bleeding" into each other, but most of these
windowing functions take a degree of processor time and can slightly slow the
response to knobs and buttons, particularly if the waterfall speed is increased to
"Yellow" or "Red" speeds.
The following windowing functions are available:
Rectangular: This is the same as no window function and results in rather
•
poor visual performance, particularly in the presence of strong signals amongst
weak ones. This setting has negligible processor load.
Sine (a.k.a. "Cosine Window"): This is slightly narrower than "Rectangular"
•
but still fairly wide and causes minimal processor loading.
Bartlett (a.k.a. Fejér): A "Triangular" window – somewhat better than
•
"Rectangular" and minimal processor load.
Welch (Parabolic): Comparable to Bartlett.
•
Hann (Raised Cosine – fastest as enhanced): Very good sidelobe rejection,
•
but not as narrow as Hamming or Blackman. This has higher processor load
than the previous functions.
Hamming (Raised Cosine): Narrower than Hann, but not quite as good
•
sidelobe rejection.
Blackman (Default): This is a good, general-purpose function with about the
•
same "narrowness" as Hamming but not quite as good sidelobe rejection.
Nuttall: Slightly wider than Blackman, comparable sidelobe rejection. This has
•
the highest processor loading of the available functions.
It may be observed that with the "wide" window function (Rectangular, Sine, Bartlett,
Welch – in decreasing tendency) that strong signals will tend to "smear" across the
display.
In contrast, for the narrowest function (e.g. Hamming and Blackman) these can cause
some weaker signals to be a bit more difficult to see as with wider, more spread-out
energy of voice signals may not be integrated into multiple FFT bins and may not be
quite as visible.
The function with the best sidelobe rejection is the Hann, but even though it is wider
than the Hamming or Blackman this can also cause some weaker signals to become
more difficult to see owing to its excellent sidelobe rejection and the lack of strong
spectral energy from one particular bin leaking into adjacent bins and being "visually
integrated".
In experimentation I have found that the Blackman is the most visually appealing,
P 44