Forums

How to get a segment of the magtitude response of a filter?

Started by agump June 15, 2009
Hi,
     I am designing a decimation multirate filter with CIC and the
decimated rate is very large . It could be larger than 1,000,000. The
difficult I face is that when the decimated rate is too large , I
can't view the details of the magnitude response of the passpand. In
order to get the details of the magnitude response , the freqz
function in matlab need to carry out a FFT with millions of points .
It's beyond the capability of my PC. I am thinking whether there is
some methods , which can get needed segment of the magnitude response
with relatively small calculation. Please give some advise , thanks.
Best Regards.
On Sun, 14 Jun 2009 20:30:06 -0700, agump wrote:

> Hi, > I am designing a decimation multirate filter with CIC and the > decimated rate is very large . It could be larger than 1,000,000. The > difficult I face is that when the decimated rate is too large , I can't > view the details of the magnitude response of the passpand. In order to > get the details of the magnitude response , the freqz function in matlab > need to carry out a FFT with millions of points . It's beyond the > capability of my PC. I am thinking whether there is some methods , which > can get needed segment of the magnitude response with relatively small > calculation. Please give some advise , thanks. Best Regards.
Try doing it symbolically, with a pencil and paper. A CIC filter has to be just about the easiest 'real' filter out there to do the math on. -- www.wescottdesign.com
On 6&#26376;15&#26085;, &#19979;&#21320;12&#26102;51&#20998;, Tim Wescott <t...@seemywebsite.com> wrote:
> On Sun, 14 Jun 2009 20:30:06 -0700, agump wrote: > > Hi, > > &nbsp; &nbsp; &nbsp;I am designing a decimation multirate filter with CIC and the > > decimated rate is very large . It could be larger than 1,000,000. The > > difficult I face is that when the decimated rate is too large , I can't > > view the details of the magnitude response of the passpand. In order to > > get the details of the magnitude response , the freqz function in matlab > > need to carry out a FFT with millions of points . It's beyond the > > capability of my PC. I am thinking whether there is some methods , which > > can get needed segment of the magnitude response with relatively small > > calculation. Please give some advise , thanks. Best Regards. > > Try doing it symbolically, with a pencil and paper. &nbsp;A CIC filter has to > be just about the easiest 'real' filter out there to do the math on. > > --www.wescottdesign.com
Hi , Tim, Maybe I forget to mention that I need to put a FIR behind the CIC . The difficult focuses on analyzing the overall frequency response.Thanks for your reply. Best Regards.
On Jun 15, 3:44&nbsp;am, agump <agump...@gmail.com> wrote:
> On 6&#26376;15&#26085;, &#19979;&#21320;12&#26102;51&#20998;, Tim Wescott <t...@seemywebsite.com> wrote: > > > > > > > On Sun, 14 Jun 2009 20:30:06 -0700, agump wrote: > > > Hi, > > > &nbsp; &nbsp; &nbsp;I am designing a decimation multirate filter with CIC and the > > > decimated rate is very large . It could be larger than 1,000,000. The > > > difficult I face is that when the decimated rate is too large , I can't > > > view the details of the magnitude response of the passpand. In order to > > > get the details of the magnitude response , the freqz function in matlab > > > need to carry out a FFT with millions of points . It's beyond the > > > capability of my PC. I am thinking whether there is some methods , which > > > can get needed segment of the magnitude response with relatively small > > > calculation. Please give some advise , thanks. Best Regards. > > > Try doing it symbolically, with a pencil and paper. &nbsp;A CIC filter has to > > be just about the easiest 'real' filter out there to do the math on. > > > --www.wescottdesign.com > > Hi , Tim, > &nbsp; &nbsp; &nbsp; &nbsp; Maybe I forget to mention that I need to put a FIR behind the > CIC . The difficult focuses on analyzing the overall frequency > response.Thanks for your reply. > Best Regards.- Hide quoted text - > > - Show quoted text -
Still as Tim stated, use the symbolic form of the CIC filter's response. Then multiply it by the FIR's response. Clay

agump wrote:

> Hi, > I am designing a decimation multirate filter with CIC and the > decimated rate is very large . It could be larger than 1,000,000. The > difficult I face is that when the decimated rate is too large , I > can't view the details of the magnitude response of the passpand. In > order to get the details of the magnitude response , the freqz > function in matlab need to carry out a FFT with millions of points . > It's beyond the capability of my PC. I am thinking whether there is > some methods , which can get needed segment of the magnitude response > with relatively small calculation. Please give some advise , thanks. > Best Regards.
Thanks you all . I'll try as you suggest.
Best Regards.