Aaron
Posts:
1
Registered:
9/10/13


odd behavior in decimate
Posted:
Sep 10, 2013 1:46 PM


If I run this command: plot(decimate(ones(100000,1),10,n)1) (n is the IIR filter order for antialiasing) I get reasonable results for n=3,5,7,9,11,13 (randomlooking hash, on the order of 1e7 or smaller) However, if I use n=2,4,6,8 there is a significant DC error in the signal (values around 0.0114). n=10,12 gives a similar result to the odd orders. It looks like there may be some odd DC behavior with the Chebyshev type I filters at even orders <= 8. If so, this makes the TMW default of n=8 a little scary.

