DSPRelated.com
Forums
   Forums    More Forums    comp.dsp  

Linear phase and impulse response symmetry

Started by Unknown ●March 11, 2008
Dear All,

First of all I am not DSP engineer, scientist whatever. I am
researcher working with microwave and RF systems, so real bare iron is
well known to me.
I have following question: Is it possible to have nonlinear phase in
frequency domain (nonconstant group delay) and non-symmetrical impulse
response in time domain?

Regards

Pawel
Reply by Pawel ●March 11, 20082008-03-11
On Mar 11, 11:04 am, prulikow...@gmail.com wrote:

> Dear All,
>
> First of all I am not DSP engineer, scientist whatever. I am
> researcher working with microwave and RF systems, so real bare iron is
> well known to me.
> I have following question: Is it possible to have nonlinear phase in
> frequency domain (nonconstant group delay) and non-symmetrical impulse
> response in time domain?
>
> Regards
>
> Pawel


OOOps actually it should be Is it possible to have nonlinear phase in
 frequency domain (nonconstant group delay) and symmetrical impulse
 response in time domain?
Reply by bharat pathak ●March 11, 20082008-03-11
Hello Pawel,

      Yes it is possible to have.

      For example if you have a filter whose
      impulse response is

      h[n] = [1 2 3 4 5]/15;

      then such a filter will have nonlinear phase
      response and a group delay which is not constant
      over the frequency range of interest.

Regards
Bharat Pathak

Arithos Designs
www.Arithos.com



>
>First of all I am not DSP engineer, scientist whatever. I am
>researcher working with microwave and RF systems, so real bare iron is
>well known to me.
>I have following question: Is it possible to have nonlinear phase in
>frequency domain (nonconstant group delay) and non-symmetrical impulse
>response in time domain?
>
>Regards
>
>Pawel
>
Reply by Pawel ●March 11, 20082008-03-11
On Mar 11, 11:13 am, "bharat pathak" <bha...@arithos.com> wrote:

> Hello Pawel,
>
>       Yes it is possible to have.
>
>       For example if you have a filter whose
>       impulse response is
>
>       h[n] = [1 2 3 4 5]/15;
>
>       then such a filter will have nonlinear phase
>       response and a group delay which is not constant
>       over the frequency range of interest.
>
> Regards
> Bharat Pathak
>
> Arithos Designswww.Arithos.com
>
>
>
> >First of all I am not DSP engineer, scientist whatever. I am
> >researcher working with microwave and RF systems, so real bare iron is
> >well known to me.
> >I have following question: Is it possible to have nonlinear phase in
> >frequency domain (nonconstant group delay) and non-symmetrical impulse
> >response in time domain?
>
> >Regards
>
> >Pawel


Thanks Bharat,

However,

how is Your impulse response symmetrical

h[n] = [1 2 3 4 5]/15;

As for me it is not, Symmetrical is for ie. h[n]=[1 2 3 4 5 4 3 2 1]
or h[n]=[1 2 3 4 5 5 4 3 2 1]  and why do You scale by 15?

Pawel
Reply by bharat pathak ●March 11, 20082008-03-11
>OOOps actually it should be Is it possible to have nonlinear phase in
> frequency domain (nonconstant group delay) and symmetrical impulse
> response in time domain?


No, it is not possible. For linear phase and constant group delay,
the impulse response has to be symmetrical.
Reply by bharat pathak ●March 11, 20082008-03-11
>
>how is Your impulse response symmetrical


Pawel, I did not say that this example impulse response is
symmetrical. I was giving example of non-symmetrical impulse
response.


>h[n] = [1 2 3 4 5]/15;


why do You scale by 15?

scaling by 15 is just some normalization criteria. Since this
is a low pass filter i am using the condition that a dc in
should give me a dc out (no scaling).

Regards
Bharat
Reply by Pawel ●March 11, 20082008-03-11
On Mar 11, 11:42 am, "bharat pathak" <bha...@arithos.com> wrote:

> >how is Your impulse response symmetrical
>
> Pawel, I did not say that this example impulse response is
> symmetrical. I was giving example of non-symmetrical impulse
> response.
>
> >h[n] = [1 2 3 4 5]/15;
>
> why do You scale by 15?
>
> scaling by 15 is just some normalization criteria. Since this
> is a low pass filter i am using the condition that a dc in
> should give me a dc out (no scaling).
>
> Regards
> Bharat


Ok, I think there is misunderstanding, probably You did not read my
further post

OOOps actually it should be Is it possible to have nonlinear phase in
 frequency domain (nonconstant group delay) and symmetrical impulse
 response in time domain?

I am specifically interested in the possibility of having symmetrical
impulse response (ie. symmetrical FIR filter) and nonlinear phase. Can
nonsymmetrical impulse response (ie nonsymmetrical FIR filter) have
linear phase?

Is it possible, is it physical?

Regards

Pawel
Reply by banton ●March 11, 20082008-03-11
>On Mar 11, 11:04 am, prulikow...@gmail.com wrote:
>> Dear All,
>>
>> First of all I am not DSP engineer, scientist whatever. I am
>> researcher working with microwave and RF systems, so real bare iron is
>> well known to me.
>> I have following question: Is it possible to have nonlinear phase in
>> frequency domain (nonconstant group delay) and non-symmetrical impulse
>> response in time domain?
>>
>> Regards
>>
>> Pawel
>
>OOOps actually it should be Is it possible to have nonlinear phase in
> frequency domain (nonconstant group delay) and symmetrical impulse
> response in time domain?


I am pretty sure, you can't.
All symmetrical IR filters have linear phase.

gr.
Bjoern
Reply by Manolis C. Tsakiris ●March 11, 20082008-03-11
Hello Powell,

the answer is no.
When a filter has a symmetrical impulse response then it's frequency
response is striclty linear.

Manolis
Reply by Pawel ●March 11, 20082008-03-11
On Mar 11, 12:06 pm, "Manolis C. Tsakiris" <el01...@mail.ntua.gr>
wrote:

> Hello Powell,
>
> the answer is no.
> When a filter has a symmetrical impulse response then it's frequency
> response is striclty linear.
>
> Manolis


Hello Manolis,

Thanks For Your help.

Is the following possible:

Can nonsymmetrical impulse response (ie nonsymmetrical FIR filter)
have
linear phase?

Pawel