next up previous contents
Next: Optimization of Dynamic Range Up: Dynamic Range Previous: Noise   Contents


Weighting Functions

By default the input signal source is assumed to have a `white' spectrum, that is, its power spectral density is assumed to be independent of frequency. It is also assumed that the circuit which is connected to the output of the filter is equally sensitive to noise at all frequencies. If this does not represent the actual situation, assumptions can be modified using an input and output weighting function.

The input weighting function Si is used to determine | fi|22 in (19). Similarly, the output weighting functions So is used to determine | gi|22 , which is used in (24). In FA specifications, the input weighting function is called si, and the output weighting function is so. They must directly or indirectly be specified as a function of frequency, represented by the variable fr. If the input spectrum is expected to be a sinc function it could be specified like this,

	si = (sin(fr * 1000) / (fr * 1000))^2;
or equivalently like this.
	si = (sin(x) / x)^2;
	x = fr * 1000;

For the change to take effect the flag num must be set. If this is done the optimizations and the dynamic range evaluations will use the weighting functions.


next up previous contents
Next: Optimization of Dynamic Range Up: Dynamic Range Previous: Noise   Contents
2009-06-03