I would like to know if I can predict what the group delay of a CIC filter is. I have an agressive FIR BPF where the sample rate is 2000 x the filter bandwidth. The group delay is longer than I would like it to be. Could a CIC filter offer a lower group delay? I'll need to interpolate back to the original sample rate, so I need to take that into consideration as part of the answer. Thanks, in advance. Jim
CIC Group Delay
Started by ●October 12, 2005
Reply by ●October 12, 20052005-10-12
jim_nospam_beasley@yahoo.com wrote:> I would like to know if I can predict what the group delay of a CIC > filter is.What is a CIC filter? Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●October 12, 20052005-10-12
"Bob Cain" <arcane@arcanemethods.com> wrote in message news:dii59h19o4@enews4.newsguy.com...> > jim_nospam_beasley@yahoo.com wrote: >> I would like to know if I can predict what the group delay of a CIC >> filter is.http://www.google.com/search?hl=en&q=%22cic+filter%22+%22group+delay%22> > What is a CIC filter? >http://www.google.com/search?hl=en&q=cic+filter> > Bob > -- > > "Things should be described as simply as possible, but no simpler." > > A. EinsteinSimple enough, Bob? ;-) Cheers, Syms.
Reply by ●October 12, 20052005-10-12
<jim_nospam_beasley@yahoo.com> wrote in message news:1129092761.120373.45800@z14g2000cwz.googlegroups.com...>I would like to know if I can predict what the group delay of a CIC > filter is. > > I have an agressive FIR BPF where the sample rate is 2000 x the filter > bandwidth. The group delay is longer than I would like it to be. Could > a CIC filter offer a lower group delay?I assume your current filter is linear phase. The delay through a linear phase FIR filter is half the filter length. Since a CIC filter is also a linear phase, the delay through the CIC will be half the length of its impulse response, which is the sum of the lengths of each stage. For a given delay, a CIC will not filter as well as a low-pass FIR designed without restrictions. Similarly, for a given set of performance criteria, a well designed low-pass will require less delay than a CIC to meet them. If you need to minimize group delay in the passband, you want a minimum phase IIR. Keep in mind, though, that a filter as narrow as you indicate is going to have a lot of passband delay no matter what you do. -- Matt
Reply by ●October 13, 20052005-10-13
Jim, A CIC filter is only good for the start of the decimation or end of the interpolation, you can't reasonably replace your filter with only a CIC filter or filters; a CIC filter is only a good filter if you are GROSSLY oversampled. For a multirate implementation (which your 1000x oversampling screams for) you might start with some CIC filters followed by some halfband filters and a final FIR filter, take the rate back up using the halfband and CIC filters in reverse order. I am curios how long is the FIR you have designed? What are the required design specs (normalize to Fs if you like)? Is linear phase a requirement? Dirk jim_nospam_beasley@yahoo.com wrote:> I would like to know if I can predict what the group delay of a CIC > filter is. > > I have an agressive FIR BPF where the sample rate is 2000 x the filter > bandwidth. The group delay is longer than I would like it to be. Could > a CIC filter offer a lower group delay? > > I'll need to interpolate back to the original sample rate, so I need to > take that into consideration as part of the answer. > > Thanks, in advance. > > Jim
Reply by ●October 13, 20052005-10-13
Thanks, all, for your replies. Yes, it's pretty clear I can't get something for nothing. While addressing some system level concerns, to test some basic principals, the order of the FIR was about 5700. A multi-rate filter is preferable, and will reduce the resource requirements tremendously. I was most concerned with what to do about the group delay. A quick check with cascaded multirate filters didn't give much improvement in delay. I'll probably explore some compromises, such as with filter bandwidth, and delay is going to be on the order of 1/BW.
Reply by ●October 13, 20052005-10-13
If you can ease up on whatever transition bandwidth requirements you have, or the stop band requirements, that will help too. Dirk
Reply by ●October 14, 20052005-10-14
It's a damn good article - one of the best I've ever read within the class of engineering magazine articles. --Randy
Reply by ●October 14, 20052005-10-14
On Tue, 11 Oct 2005 22:00:37 -0700, Bob Cain <arcane@arcanemethods.com> wrote:> > >jim_nospam_beasley@yahoo.com wrote: >> I would like to know if I can predict what the group delay of a CIC >> filter is. > >What is a CIC filter? > > >BobHi, ya' have a look at: http://www.embedded.com/shared/printableArticle.jhtml?articleID=160400592 [-Rick-]
Reply by ●October 14, 20052005-10-14
On 11 Oct 2005 21:52:41 -0700, jim_nospam_beasley@yahoo.com wrote:>I would like to know if I can predict what the group delay of a CIC >filter is. > >I have an agressive FIR BPF where the sample rate is 2000 x the filter >bandwidth. The group delay is longer than I would like it to be. Could >a CIC filter offer a lower group delay? > >I'll need to interpolate back to the original sample rate, so I need to >take that into consideration as part of the answer. > >Thanks, in advance. > >JimHi, as far as I recall, the delay through a single-stage CIC filter having D delay elements in its comb filter will be Delay = (D - 1)/2 samples. [-Rick-]