Pankaj Daga wrote:

> Hi,
> 
> There is something that is going on which I failed to notice before.
> It takes the absolute value of the filter bank returned output and
> then does the low pass filtering. What kind of information will that
> provide?
> 
> Thanks,
> 
> N
-- snip --

That will provide a dandy AM demodulator, assuming that you have 
carriers at 30Hz, 90Hz, etc.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Yup,

this must be it! Thanks a lot!!

Best,
N

"Jon Harris" <goldentully@hotmail.com> wrote in message news:<2tad00F1tfk7mU1@uni-berlin.de>...
> "Pankaj Daga" <pankajdaga@excite.com> wrote in message
> news:1a4b2aef.0410150117.67096704@posting.google.com...
> > Hi,
> >
> > There is something that is going on which I failed to notice before.
> > It takes the absolute value of the filter bank returned output and
> > then does the low pass filtering. What kind of information will that
> > provide?
> 
> As we suspected.  Absolute value followed by low pass filtering provides some
> measure of the average signal energy.  It looks like the process is trying to
> find the average energy in different frequency bands.

"Pankaj Daga" <pankajdaga@excite.com> wrote in message
news:1a4b2aef.0410150117.67096704@posting.google.com...
> Hi,
>
> There is something that is going on which I failed to notice before.
> It takes the absolute value of the filter bank returned output and
> then does the low pass filtering. What kind of information will that
> provide?

As we suspected.  Absolute value followed by low pass filtering provides some
measure of the average signal energy.  It looks like the process is trying to
find the average energy in different frequency bands.

"Pankaj Daga" <pankajdaga@excite.com> wrote in message
news:1a4b2aef.0410150117.67096704@posting.google.com...
> Hi,
>
> There is something that is going on which I failed to notice before.
> It takes the absolute value of the filter bank returned output and
> then does the low pass filtering. What kind of information will that
> provide?

As we suspected.  Absolute value followed by low pass filtering provides some
measure of the average signal energy.  It looks like the process is trying to
find the average energy in different frequency bands.

pankajdaga@excite.com (Pankaj Daga) wrote in message news:<1a4b2aef.0410150117.67096704@posting.google.com>...
> Hi,
> 
> There is something that is going on which I failed to notice before.
> It takes the absolute value of the filter bank returned output and
> then does the low pass filtering. What kind of information will that
> provide?

Eh... I'm getting very curious now. 

What is the task of the program system you "inherited"? Or are you 
completely in the dark as of what problem it was designed to solve?

Rune

Pankaj Daga wrote:

> Hi,
> 
> There is something that is going on which I failed to notice before.
> It takes the absolute value of the filter bank returned output and
> then does the low pass filtering. What kind of information will that
> provide?
> 
> Thanks,
> 
> N

Absolute value is what an ideal full-wave rectifier creates from an
analog signal. Low-passed absolute value is AM demodulation. What
happens after the filters?

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

On 15 Oct 2004 02:17:30 -0700, pankajdaga@excite.com (Pankaj Daga)
wrote:

>Hi,
>
>There is something that is going on which I failed to notice before.
>It takes the absolute value of the filter bank returned output and
>then does the low pass filtering. What kind of information will that
>provide?
>
>Thanks,
>
>N

Hi,
  that depends on what "follow on" operations are 
performed on the absolute value samples.

You'll have to keep decyphering the code 
***very*** carefully to find.

Keep asking questions.  The guys here will be able 
to answer them.

Good Luck,
[-Rick-]

Hi,

There is something that is going on which I failed to notice before.
It takes the absolute value of the filter bank returned output and
then does the low pass filtering. What kind of information will that
provide?

Thanks,

N



Tim Wescott <tim@wescottnospamdesign.com> wrote in message news:<10mt6oggn54m71b@corp.supernews.com>...
> Jerry Avins wrote:
> 
> > Pankaj Daga wrote:
> > 
> > 
> >>Hi everyone,
> >>
> >>Just got handed a 600000 line software application with lots of DSP. I
> >>am trying to understand a few things and as I am new to DSP, it is a
> >>bit complicated.
> >>
> >>Here is a scenario. There is a signal that is being fed into a
> >>cascaded Bessel filter bank that splits it into 0-60 Hz, 60-120 Hz,
> >>120-180 Hz, 180-240 Hz uptil 480 Hz... i.e. in bands of 60 Hz.
> >>
> >>After this, the output of these filters is again fed to a low pass
> >>filter at 60 Hz...now, what effect will this have on the signals... I
> >>mean, 0-60 Hz would not have any components above 60 HZ and the rest
> >>will have components above 60 Hz. So what si the point of low pass
> >>filtering it again? I am sure I am missing something very fundamental
> >>here and was hoping that someone would be able to help with this.
> >>
> >>Thanks,
> >>
> >>Sincerely,
> >>
> >>N
> > 
> > 
> > The scenario you describe is too stupid to be real. The 0-60 Hz signal
> > would hardly be changed by the low-pass, and the others would be
> > suppressed. There must be more to it that isn't clear yet.
> > 
> > The use of Bessel filters is an indication that phase linearity is more
> > important than sharp band edges, whatever that's worth.
> > 
> > Jerry
> 
> Is there anything else going on between the filter banks and the final 
> 60Hz filtering?  Squaring of the signals, perhaps?
> 
> If some sort of modulation is being extracted from the signals (i.e. by 
> squaring) then the resulting demodulated waveforms _would_ have content 
> below 60Hz, as well as unwanted content above.
> 
> This is just supposition, of course...

Tim Wescott wrote:
> Jerry Avins wrote:
> 
>> Pankaj Daga wrote:
>>
>>
>>> Hi everyone,
>>>
>>> Just got handed a 600000 line software application with lots of DSP. I
>>> am trying to understand a few things and as I am new to DSP, it is a
>>> bit complicated.
>>>
>>> Here is a scenario. There is a signal that is being fed into a
>>> cascaded Bessel filter bank that splits it into 0-60 Hz, 60-120 Hz,
>>> 120-180 Hz, 180-240 Hz uptil 480 Hz... i.e. in bands of 60 Hz.
>>>
>>> After this, the output of these filters is again fed to a low pass
>>> filter at 60 Hz...now, what effect will this have on the signals... I
>>> mean, 0-60 Hz would not have any components above 60 HZ and the rest
>>> will have components above 60 Hz. So what si the point of low pass
>>> filtering it again? I am sure I am missing something very fundamental
>>> here and was hoping that someone would be able to help with this.
>>>
>>> Thanks,
>>>
>>> Sincerely,
>>>
>>> N
>>
>>
>>
>> The scenario you describe is too stupid to be real. The 0-60 Hz signal
>> would hardly be changed by the low-pass, and the others would be
>> suppressed. There must be more to it that isn't clear yet.
>>
>> The use of Bessel filters is an indication that phase linearity is more
>> important than sharp band edges, whatever that's worth.
>>
>> Jerry
> 
> 
> Is there anything else going on between the filter banks and the final
> 60Hz filtering?  Squaring of the signals, perhaps?
> 
> If some sort of modulation is being extracted from the signals (i.e. by
> squaring) then the resulting demodulated waveforms _would_ have content
> below 60Hz, as well as unwanted content above.
> 
> This is just supposition, of course...

I'd be more inclined to think that it's demodulation if we knew that the
various bands were squared separately, each representing a separate
channel. What happens to the outputs of all the low-pass filters would
be an interesting bit of information.

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

"Tim Wescott" <tim@wescottnospamdesign.com> wrote in message
news:10mt6oggn54m71b@corp.supernews.com...
> > Pankaj Daga wrote:
> >
> >
> >>Just got handed a 600000 line software application with lots of DSP. I
> >>am trying to understand a few things and as I am new to DSP, it is a
> >>bit complicated.
> >>
> >>Here is a scenario. There is a signal that is being fed into a
> >>cascaded Bessel filter bank that splits it into 0-60 Hz, 60-120 Hz,
> >>120-180 Hz, 180-240 Hz uptil 480 Hz... i.e. in bands of 60 Hz.
> >>
> >>After this, the output of these filters is again fed to a low pass
> >>filter at 60 Hz...now, what effect will this have on the signals... I
> >>mean, 0-60 Hz would not have any components above 60 HZ and the rest
> >>will have components above 60 Hz. So what si the point of low pass
> >>filtering it again? I am sure I am missing something very fundamental
> >>here and was hoping that someone would be able to help with this.
>
> Is there anything else going on between the filter banks and the final
> 60Hz filtering?  Squaring of the signals, perhaps?
>
> If some sort of modulation is being extracted from the signals (i.e. by
> squaring) then the resulting demodulated waveforms _would_ have content
> below 60Hz, as well as unwanted content above.
>
> This is just supposition, of course...

That is exactly what I was thinking.  Squaring the signal or doing some other
non-linear operation could generate additional frequencies that might need to be
filtered out by the 60Hz LP.