Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
I learn my polyphase filtering from a couple of app notes in the ADSP-2100
family application guide. But I can't find it any more. I can give you a hand in
understanding it but you will have to be specific. Down Sampling filtering is
very simple but concept still involve understanding of filter design.
Unfortunately I don't know what CIC decimation filter is. I use polyphase filter for
both audio and video. And I am not lucky enough to have Simulink. Anyway, I will
give you my 2 cents of experiences.
1. since we will be using a digital FIR filter for decimation, the filter will have to design
based on input sampling frequency and output sampling frequency. (Now I really wish
I can locate the paper I learn it from. It is hard to explaint things without drawing it out.)
For example, if you are downsample by 2, which means you are throwing away every
other sample. That also means you only keep half the bandwidth from the original signal.
You will need a FIR filter that cuts off at half of your incoming frequency bandwidth.
2. The number of taps you have determine how sharp the cut off and how much riinging
you will get in both pass and stop band. The more taps you have the better. But that
also depends on number of bits you have and number of fractional bit you are keeping
during calculation. There is no need to have close to zero ripple if you only need 10-bits
in and out. Here is the rule of thumb. Figure out the lowest level of you input/output
system (number of bits you used), that will determine your maximum allowable ripple in
both pass and stop band of your filter. The adjust your number of taps to give you
the best cutoff slope that you can realistically implement. Then calculate the filter coefficients.
I know this email may not help too much. It is a subject that takes at least a chaper
of a text book to explain. Kind of hard to do that in a email.