On Monday, January 10, 2011 1:48:44 PM UTC-5, Fred Marshall wrote:
...
> assuming that they both start at t=0,
> then
> the "filter" introduces a delay of N/2
> and
> the output of the filter is transient for N/2 samples
> it is in steady state more or less for N samples
> is in transient again for N/2 samples.
...
What am I missing? The transversal filter than best illustrates my dilemma is a boxcar: N coefficients, each of magnitude 1/N. If the signal is a unit step, the output will steadily build for N (not N/2) samples. If eventually there is a trailing edge, the decay will be equally long.
Jerry