A Quadrature Signals Tutorial: Complex, But Not Complicated

Understanding the 'Phasing Method' of Single Sideband Demodulation

Complex Digital Signal Processing in Telecommunications

Introduction to Sound Processing

Introduction of C Programming for DSP Applications

There are **9** messages in this thread.

You are currently looking at messages 1 to .

**Is this discussion worth a thumbs up?**

I read in one of the literature about a CIC filter with decimation of 3.5. As far as I know CIC decimates by integer only. Is there a way to build a CIC with decimation of 3.5 ? Tom

```
>I read in one of the literature about a CIC filter with decimation of 3.5.
>As far as I know CIC decimates by integer only.
>Is there a way to build a CIC with decimation of 3.5 ?
>
>Tom
Reference?
The only way I can think of is for there to be an interpolator included
within the CIC, e.g. decimate by 7 then interpolate by 2.
Mark
```

On Apr 30, 12:26 pm, "Tom" <tomda...@yahoo.com> wrote: > I read in one of the literature about a CIC filter with decimation of 3.5. > As far as I know CIC decimates by integer only. > Is there a way to build a CIC with decimation of 3.5 ? > > Tom CICs can be used in more complex structures to achieve non-integer resampling. Try Google cic noninteger Don't expect all or the best results to be on the first page or take only one click to access. http://coblitz.codeen.org:3125/citeseer.ist.psu.edu/cache/papers/cs/24224/http:zSzzSzwooster.hut.fizSzpublicationszSzfinsig01zSzdecimation_by_non.pdf/decimation-by-non-integer.pdf http://citeseer.ist.psu.edu/cache/papers/cs/30893/http:zSzzSztelin.ugent.bezSz~hwymeerszSzpaperszSzBaiona03.pdf/wymeersch03low.pdf http://www.telfor.org.yu/telfor2001/radovi/7-9.pdf Dale B. Dalrymple http://dbdimages.com

>You keept twice the information if you interpolate first. Should have said AND interpolate by 2. :) Mark

Hello Jerry, Does it matter which one comes first (interpolate or decimate) other than decimation is prefered first to run the logic at lower clock speed (f/7 and 3.5f rather than 2f and 3.5f if we interpolate first)? >markt wrote: >>> I read in one of the literature about a CIC filter with decimation of 3.5. >> >>> As far as I know CIC decimates by integer only. >>> Is there a way to build a CIC with decimation of 3.5 ? >>> >>> Tom >> >> Reference? >> >> The only way I can think of is for there to be an interpolator included >> within the CIC, e.g. decimate by 7 then interpolate by 2. > >You keept twice the information if you interpolate first. > >Jerry >-- >Engineering is the art of making what you want from things you can get. >??????????????????????????????????????????????????????????????????????? >

Jerry, I got it. Thank you very much. >ytach wrote: >> Hello Jerry, >> >> Does it matter which one comes first (interpolate or decimate) other than >> decimation is prefered first to run the logic at lower clock speed (f/7 and >> 3.5f rather than 2f and 3.5f if we interpolate first)? >> >> >>> markt wrote: >>>>> I read in one of the literature about a CIC filter with decimation of >> 3.5. >>>>> As far as I know CIC decimates by integer only. >>>>> Is there a way to build a CIC with decimation of 3.5 ? >>>>> >>>>> Tom >>>> Reference? >>>> >>>> The only way I can think of is for there to be an interpolator >> included >>>> within the CIC, e.g. decimate by 7 then interpolate by 2. >>> You keept twice the information if you interpolate first. > >If you interpolate first, the maximum allowed frequency in the result is > (fs is the original sample rate) fs/3.5. If you do it last, the >maximum frequency can be only fs/7. Once aliasing occurs, id won't be >undone by interpolating. > >Jerry >-- >Engineering is the art of making what you want from things you can get. >??????????????????????????????????????????????????????????????????????? >

```
On Apr 30, 7:45 pm, "ytach" <ytachw...@ou.edu> wrote:
> Does it matter which one comes first (interpolate or decimate) other than
> decimation is prefered first to run the logic at lower clock speed (f/7 and
> 3.5f rather than 2f and 3.5f if we interpolate first)?
Do you even have to actually interpolate?
What if you just double the clock?
And either repeate the samples or zero fill (meaning you only have to
do the math every other clock)
Does this come out to the same as alternating between two adjacent
integer ratios?
(Assuming for a minute that you are putting low-rate output samples
into a fifo - if you actually want to send them immediately to a DAC
or something where output interval matters, you'll need a sample clock
that doesn't jitter)
```