Basic Questions??

Setanta wrote:
> "Jerry Avins" <jya@ieee.org> wrote in message
> news:bcydnd-NI9KZ78TZnZ2dnUVZ_v-dnZ2d@rcn.net...
> > Setanta wrote:
> > > "mobi" <mobien@gmail.com> wrote in message
> > > news:1146655522.379792.202490@y43g2000cwc.googlegroups.com...
> > >
> > >
> > >>(4) We can have digital filters for anti-aliasing when we do resampling
> > >>in digital domain. But if you think that analog anti-aliasing filter
> > >>can be replaced by digital after the sampler, then we cannot have a
> > >>digital anti-alaising fitler.
> > >
> > >
> > >
> > > It is possible to sample under the nyquist frequency and still represent the
> > > actual signal - without pre-filtering.
> > > What is needed are a number channels - sampling at different frequencies -
> > > selecting - for instance - three channels sampling at well chosen
> > > frequencies wrt to the system frequency up to 2000 times the maximum
> > > sampling rate can be detected in the signal - and aliasing can be removed
> > > simply.
> >
> > Please tell me more.
>
> It is difficult to illiterate these ideas here in this arena
>
> If you want to no more check out a paper in 'Sound & Vibration Journal'
>
> Called
>
> Nyquist-overcoming the limitations
>
> R.F. McLeana,*, S.H. Alsopa, J.S. Flemingb
>

Just looked at this paper, and its content can best be described by the
word "guff".


-- 
Oli

On 5 May 2006 01:44:51 -0700, "Oli Filth" <catch@olifilth.co.uk>
wrote:

  (snipped)

>> It is difficult to illiterate these ideas here in this arena
>>
>> If you want to no more check out a paper in 'Sound & Vibration Journal'
>>
>> Called
>>
>> Nyquist-overcoming the limitations
>>
>> R.F. McLeana,*, S.H. Alsopa, J.S. Flemingb
>>
>
>Just looked at this paper, and its content can best be described by the
>word "guff".
>-- 
>Oli

Hi Oli,

  over the years I've seen two magazine articles 
that promote signal processing schemes that  
violate Nyquist sampling.  In both cases, the only
thing those sig proc methods achieved was to badly 
embarrass the authors.

[-Rick-]
 

"Oli Filth" <catch@olifilth.co.uk> wrote in message
news:1146818691.706740.93090@j33g2000cwa.googlegroups.com...
> Setanta wrote:
> > "Jerry Avins" <jya@ieee.org> wrote in message
> > news:bcydnd-NI9KZ78TZnZ2dnUVZ_v-dnZ2d@rcn.net...
> > > Setanta wrote:
> > > > "mobi" <mobien@gmail.com> wrote in message
> > > > news:1146655522.379792.202490@y43g2000cwc.googlegroups.com...
> > > >
> > > >
> > > >>(4) We can have digital filters for anti-aliasing when we do
resampling
> > > >>in digital domain. But if you think that analog anti-aliasing filter
> > > >>can be replaced by digital after the sampler, then we cannot have a
> > > >>digital anti-alaising fitler.
> > > >
> > > >
> > > >
> > > > It is possible to sample under the nyquist frequency and still
represent the
> > > > actual signal - without pre-filtering.
> > > > What is needed are a number channels - sampling at different
frequencies -
> > > > selecting - for instance - three channels sampling at well chosen
> > > > frequencies wrt to the system frequency up to 2000 times the maximum
> > > > sampling rate can be detected in the signal - and aliasing can be
removed
> > > > simply.
> > >
> > > Please tell me more.
> >
> > It is difficult to illiterate these ideas here in this arena
> >
> > If you want to no more check out a paper in 'Sound & Vibration Journal'
> >
> > Called
> >
> > Nyquist-overcoming the limitations
> >
> > R.F. McLeana,*, S.H. Alsopa, J.S. Flemingb
> >
>
> Just looked at this paper, and its content can best be described by the
> word "guff".
>
>
> --
> Oli
>

Dear Oli

The claims in the paper and the subsequent development of the system
outlined in it work very well.

Perhaps you don't understand it??

Setanta

Jerry Avins wrote:
...
> > It is possible to sample under the nyquist frequency and still represent the
> > actual signal - without pre-filtering.
> > What is needed are a number channels - sampling at different frequencies -
> > selecting - for instance - three channels sampling at well chosen
> > frequencies wrt to the system frequency up to 2000 times the maximum
> > sampling rate can be detected in the signal - and aliasing can be removed
> > simply.
>
> Please tell me more.

That technique is concerned with sampling _periodic_ wide-band signals.
They propose a system using three samplers, wich can reconstruct
periodic signals with highest frequency up to 5000 times the frequency
of the highest sampler.

I believe them. The simple reason is because I can do even better :-).
I propose a system using only two (regular) samplers, both of which can
have an arbitrary low upper bound on their sampling rate. Using my
system, one can perfectly reconstruct continuous periodic signals with
no bound on the highest harmonic (ie non-bandlimited signals).

Interested? I wouldn't be surprised if you already knew the answer,
Jerry :-).

Regards,
Andor

Andor wrote:
> Jerry Avins wrote:
> ...
>>> It is possible to sample under the nyquist frequency and still represent the
>>> actual signal - without pre-filtering.
>>> What is needed are a number channels - sampling at different frequencies -
>>> selecting - for instance - three channels sampling at well chosen
>>> frequencies wrt to the system frequency up to 2000 times the maximum
>>> sampling rate can be detected in the signal - and aliasing can be removed
>>> simply.
>> Please tell me more.
> 
> That technique is concerned with sampling _periodic_ wide-band signals.
> They propose a system using three samplers, wich can reconstruct
> periodic signals with highest frequency up to 5000 times the frequency
> of the highest sampler.
> 
> I believe them. The simple reason is because I can do even better :-).
> I propose a system using only two (regular) samplers, both of which can
> have an arbitrary low upper bound on their sampling rate. Using my
> system, one can perfectly reconstruct continuous periodic signals with
> no bound on the highest harmonic (ie non-bandlimited signals).
> 
> Interested? I wouldn't be surprised if you already knew the answer,
> Jerry :-).

Andor,

You spoil the fun. I wanted the OP to explain that. Also to estimate how 
long a sample must be taken and how stable the signal needs to be over 
that time. You need two samplers only if you can't know that your sample 
and signal periods don't have a non-trivial common divisor.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Jerry Avins wrote:
...
> You need two samplers only if you can't know that your sample
> and signal periods don't have a non-trivial common divisor.

Or if you want the scheme to work for periodic signals with no
restriction on the period. The extension to two samplers should have
occured to me when I posted the original riddle.

Will anybody please try to reply the questions rather than starting out
a new topic.

Please
Andor wrote:
> Jerry Avins wrote:
> ...
> > You need two samplers only if you can't know that your sample
> > and signal periods don't have a non-trivial common divisor.
>
> Or if you want the scheme to work for periodic signals with no
> restriction on the period. The extension to two samplers should have
> occured to me when I posted the original riddle.

Ok James, let's see what your question was:

James wrote:
> (5) If we have a 10Khz Processor and DAQ Card which samples at
> 100Khz/Samples......Can we have such configuration ????

Think about this for a moment: the postman rings at your door 10 times,
but you are still sleepy and only have the energy to answer the door
once. Can you arrange yourself with the postman so that you can still
collect all 10 letters?

On 6 May 2006 05:38:10 -0700, Andor <andor.bariska@gmail.com> wrote:
> Ok James, let's see what your question was:
> 
> James wrote:
>> (5) If we have a 10Khz Processor and DAQ Card which samples at
>> 100Khz/Samples......Can we have such configuration ????
> 
> Think about this for a moment: the postman rings at your door 10 times,
> but you are still sleepy and only have the energy to answer the door
> once. Can you arrange yourself with the postman so that you can still
> collect all 10 letters?

Given that today is Saturday, the answer is obvious. I'd (re-)invent
the ultra-low-speed high-capacity packetized buffer, go back to sleep,
and check the contents at a later date. Tuesday, maybe.

Oh, and I'd disconnect the interrupt signal line. (Ack! Phlbbt!)


Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut minds pring dawt cahm (y'all)
--
    Ambition is a poor excuse for not having enough sense to be lazy.
-- 

programmer.james@gmail.com wrote:
> Will anybody please try to reply the questions rather than starting out
> a new topic.
> 
> Please

I guess you're new around here :-)

> Andor wrote:
> 
>>Jerry Avins wrote:
>>...
>>
>>>You need two samplers only if you can't know that your sample
>>>and signal periods don't have a non-trivial common divisor.
>>
>>Or if you want the scheme to work for periodic signals with no
>>restriction on the period. The extension to two samplers should have
>>occured to me when I posted the original riddle.

Steve