Linear Phase via 'Phase Cloning' New Method

I have an algorythm that yields a Linear phase response of a lowpass filter
that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE IN
BOTH PASS AND STOP bands over several octaves.

 (***however, to process a signal I must insert a known time delay to the
filter output***)

I believe it is of the first of its kind.  I would like your help in
qualifying the algorithm. The proof is in the showing.

The method is based on linear 'PHASE CLONING', a method i developed where
we force a signal to mimic, 'clone' the phase of another.

Send me a txt/xls file of data with a known singal (time domain), Ill
filter and send back the low pass filtered data.  We can start with a
bessel 4 pole.  Once i prove the filter, i will 'clone' a butterworth, or
cheby filter.

someone wrote:
> I have an algorythm that yields a Linear phase response of a lowpass filter
> that is neither FIR nor IIR. �

Is it a non-LTI method? What is the advantage over standard linear-
phase FIR filters?

> In fact it can approximate LINEAR PHASE IN
> BOTH PASS AND STOP bands over several octaves.

Sounds reasonable. However, the linear-phase response in the stop-band
is usually a waste of coefficients.

>
> �(***however, to process a signal I must insert a known time delay to the
> filter output***)

Sounds reasonable.

>
> I believe it is of the first of its kind. �I would like your help in
> qualifying the algorithm. The proof is in the showing.

Well, describe your algorithm so that we can qualify it.

>
> The method is based on linear 'PHASE CLONING', a method i developed where
> we force a signal to mimic, 'clone' the phase of another.

Some people call that delay.

>
> Send me a txt/xls file of data with a known singal (time domain), Ill
> filter and send back the low pass filtered data. �

Around here, we've been having linear-phase lowpass filter for a while
now. Where do you live?

> We can start with a
> bessel 4 pole. �Once i prove the filter, i will 'clone' a butterworth, or
> cheby filter.

What's the use of that? Can you actually state what advantage your
method brings that the world has been waiting for?

Regards,
Andor

westocl wrote:

> I have an algorythm that yields a Linear phase response of a lowpass filter
> that is neither FIR nor IIR. In fact it can approximate LINEAR PHASE IN
> BOTH PASS AND STOP bands over several octaves.
> 
>  (***however, to process a signal I must insert a known time delay to the
> filter output***)
> 
> I believe it is of the first of its kind.  I would like your help in
> qualifying the algorithm. The proof is in the showing.

So you just filter in the forward and backward directions, and use some 
sort of overlap and add to make it continuous time, right?

> The method is based on linear 'PHASE CLONING', a method i developed where
> we force a signal to mimic, 'clone' the phase of another.

Ingenious.

> Send me a txt/xls file of data with a known singal (time domain), Ill
> filter and send back the low pass filtered data.  We can start with a
> bessel 4 pole.  Once i prove the filter, i will 'clone' a butterworth, or
> cheby filter.

After how many semesters you have been kicked out from the university?


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On 13 Feb, 13:02, "westocl" <cwest...@hotmail.com> wrote:
> I have an algorythm that yields a Linear phase response of a lowpass filter
> that is neither FIR nor IIR. &#4294967295;
...
> I believe it is of the first of its kind. &#4294967295;

Then you are wasting your time here. This is the sort
of thing that research journals are for, I would suggest
IEEE Transactions on Signal Processing. If you are right,
that's the kind of place to publish your algorithm.

Rune

On Wed, 13 Feb 2008 06:02:06 -0600, "westocl" <cweston_@hotmail.com>
wrote:

>I have an algorythm that yields a Linear phase response of a lowpass filter
>that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE IN
>BOTH PASS AND STOP bands over several octaves.
>
  (snipped)

Hi,
   Your words are hard for me to understand.
How can a filter's time-domain impulse response 
not be FIR and not be IIR?  If the filter's phase 
response is linear, then doesn't that mean the 
filter has an FIR impulse response?

[-Rick-]

westocl wrote:
> I have an algorythm that yields a Linear phase response of a lowpass filter
> that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE IN
> BOTH PASS AND STOP bands over several octaves.

That's nonsense. If the impulse response is neither finite nor infinite, 
it must be zero. What good is that?

Phase in the stopband? By measurement, how could anyone tell the phase 
of a suppressed signal? Why would anyone want to?

>  (***however, to process a signal I must insert a known time delay to the
> filter output***)

Well! It might be a filter after all!

> I believe it is of the first of its kind.  I would like your help in
> qualifying the algorithm. The proof is in the showing.

OK, show me. What is the algorithm?

> The method is based on linear 'PHASE CLONING', a method i developed where
> we force a signal to mimic, 'clone' the phase of another.

Interesting.

> Send me a txt/xls file of data with a known singal (time domain), Ill
> filter and send back the low pass filtered data.  We can start with a
> bessel 4 pole.  Once i prove the filter, i will 'clone' a butterworth, or
> cheby filter.

If you do this in secret, how will we know that you didn't simply 
construct a plausible output by hand instead of with an algorithm?

Jerry

P.S. I suspect that you can't get a three-pole response.
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

>On Wed, 13 Feb 2008 06:02:06 -0600, "westocl" <cweston_@hotmail.com>
>wrote:
>
>>I have an algorythm that yields a Linear phase response of a lowpass
filter
>>that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE
IN
>>BOTH PASS AND STOP bands over several octaves.
>>
>  (snipped)
>
>Hi,
>   Your words are hard for me to understand.
>How can a filter's time-domain impulse response 
>not be FIR and not be IIR?  If the filter's phase 
>response is linear, then doesn't that mean the 
>filter has an FIR impulse response?
>
>[-Rick-]
>

Sorry for the confusion, everyone.   I called the 'filter responses'
neither...because by classical definitions they dont really fall in either
catagory).  Miming 'filter responses' are just one of applications of the
algorithm. 

I wont post here anymore, thanx.

On Wed, 13 Feb 2008 16:46:08 -0600, "westocl" <cweston_@hotmail.com>
wrote:

>>On Wed, 13 Feb 2008 06:02:06 -0600, "westocl" <cweston_@hotmail.com>
>>wrote:
>>
>>>I have an algorythm that yields a Linear phase response of a lowpass
>filter
>>>that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE
>IN
>>>BOTH PASS AND STOP bands over several octaves.
>>>
>>  (snipped)
>>
>>Hi,
>>   Your words are hard for me to understand.
>>How can a filter's time-domain impulse response 
>>not be FIR and not be IIR?  If the filter's phase 
>>response is linear, then doesn't that mean the 
>>filter has an FIR impulse response?
>>
>>[-Rick-]
>>
>
>Sorry for the confusion, everyone.   I called the 'filter responses'
>neither...because by classical definitions they dont really fall in either
>catagory).  Miming 'filter responses' are just one of applications of the
>algorithm. 
>
>I wont post here anymore, thanx.

Hi westocl,
   Why won't you post here anymore?

See Ya',
[-Rick-]

>westocl wrote:
>> I have an algorythm that yields a Linear phase response of a lowpass
filter
>> that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE
IN
>> BOTH PASS AND STOP bands over several octaves.
>
>That's nonsense. If the impulse response is neither finite nor infinite,

>it must be zero. What good is that?
>
>Phase in the stopband? By measurement, how could anyone tell the phase 
>of a suppressed signal? Why would anyone want to?
>
>>  (***however, to process a signal I must insert a known time delay to
the
>> filter output***)
>
>Well! It might be a filter after all!
>
>> I believe it is of the first of its kind.  I would like your help in
>> qualifying the algorithm. The proof is in the showing.
>
>OK, show me. What is the algorithm?
>
>> The method is based on linear 'PHASE CLONING', a method i developed
where
>> we force a signal to mimic, 'clone' the phase of another.
>
>Interesting.
>
>> Send me a txt/xls file of data with a known singal (time domain), Ill
>> filter and send back the low pass filtered data.  We can start with a
>> bessel 4 pole.  Once i prove the filter, i will 'clone' a butterworth,
or
>> cheby filter.
>
>If you do this in secret, how will we know that you didn't simply 
>construct a plausible output by hand instead of with an algorithm?
>
>Jerry
>
>P.S. I suspect that you can't get a three-pole response.
>-- 
>Engineering is the art of making what you want from things you can get.
>???????????????????????????????????????????????????????????????????????
>

Jerry, thanks for sending a decent reply back and there are obvious
reasons why i wont just post the algorithm. One, im still developing and
two are the people like the first two that replied to me.

Once again, when i say its neither, is because by FIR/IIR classical
definitions the response doesnt really apply.

Heres a BIG reason why you might want to do this....  If you can have an
ouptut phase of a signal comming from a filter clone the phase of a signal
that you do know and you forced them to be the same and they come out
different in a LTI system....you might just possible conclude that that
phase differnce comes from noise... Hmmm... you think we can subtract some
of that noise out of there??  reduced NF??

jerry -
>If you do this in secret, how will we know that you didn't simply 
>construct a plausible output by hand instead of with an algorithm?

RE
-because my friend,jerry... You know whats in the text file... I dont, so
how can i have done it by hand??  I have no clue what that data means.
Send some data in the stop band, with a known phase... and i claim to be
able to have rejected it in magnitude and kept the phase.

or how bout i take a chebychev or eliptical filter and give it the phase
response of a bessel.

>westocl wrote:
>> I have an algorythm that yields a Linear phase response of a lowpass
filter
>> that is neither FIR nor IIR.  In fact it can approximate LINEAR PHASE
IN
>> BOTH PASS AND STOP bands over several octaves.
>
>That's nonsense. If the impulse response is neither finite nor infinite,

>it must be zero. What good is that?
>
>Phase in the stopband? By measurement, how could anyone tell the phase 
>of a suppressed signal? Why would anyone want to?
>
>>  (***however, to process a signal I must insert a known time delay to
the
>> filter output***)
>
>Well! It might be a filter after all!
>
>> I believe it is of the first of its kind.  I would like your help in
>> qualifying the algorithm. The proof is in the showing.
>
>OK, show me. What is the algorithm?
>
>> The method is based on linear 'PHASE CLONING', a method i developed
where
>> we force a signal to mimic, 'clone' the phase of another.
>
>Interesting.
>
>> Send me a txt/xls file of data with a known singal (time domain), Ill
>> filter and send back the low pass filtered data.  We can start with a
>> bessel 4 pole.  Once i prove the filter, i will 'clone' a butterworth,
or
>> cheby filter.
>
>If you do this in secret, how will we know that you didn't simply 
>construct a plausible output by hand instead of with an algorithm?
>
>Jerry
>
>P.S. I suspect that you can't get a three-pole response.
>-- 
>Engineering is the art of making what you want from things you can get.
>???????????????????????????????????????????????????????????????????????
>

Jerry, thanks for sending a decent reply back and there are obvious
reasons why i wont just post the algorithm. One, im still developing and
two are the people like the first two that replied to me.

Once again, when i say its neither, is because by FIR/IIR classical
definitions the response doesnt really apply.

Heres a BIG reason why you might want to do this....  If you can have an
ouptut phase of a signal comming from a filter clone the phase of a signal
that you do know and you forced them to be the same and they come out
different in a LTI system....you might just possible conclude that that
phase differnce comes from noise... Hmmm... you think we can subtract some
of that noise out of there??  reduced NF??

jerry -
>If you do this in secret, how will we know that you didn't simply 
>construct a plausible output by hand instead of with an algorithm?

RE
-because my friend,jerry... You know whats in the text file... I dont, so
how can i have done it by hand??  I have no clue what that data means.
Send some data in the stop band, with a known phase... and i claim to be
able to have rejected it in magnitude and kept the phase.

or how bout i take a chebychev or eliptical filter and give it the phase
response of a bessel.