FIR Filter limitation (or not?)

rover8898 wrote:
> Hello,
> 
> 
>>It's not the digital filters -- it's the digitizing process of an
>>analog signal what requires this filtering;  it's not always done
>>with analog signals;  some times it's easier to sample the signal
>>at a far higher rate, then apply a digital low-pass filter, and
>>then "sample" the resulting signal  (since it is already in the
>>digital domain, we're talking simply about keeping one every N
>>samples, or "downsample" by a factor of N)
> 
> 
> The filtering can be done with a [oversampling at higher sample rate] +
> [digital low pass filter] +[downsampling] scheme, I guess. But if the
> input signal (prior to A/D) has frequency components beyond the
> [oversampling rate/2] threshold, then there will aliasing, digital
> lowpass filter or not. Is there a reason why a precautionary ~broad
> analog filter cannot be placed ahead of the A/D (aside from cost and
> maybe gadget size)  ?

The technique of sampling at a much higher frequency and then do
a low-pass digital filtering does not replace the analog filter;
it just makes the analog filter trivial -- if you know that your
signal has valuable spectral contents up to 20 kHz and want to
sample at 44.1 kHz, the analog filter required would have to be
very precise and very "wall-like" -- it would be quite hard to
design an analog filter with such a "wall" frequency response
and without truly ugly phase distorsion (mostly at frequencies
near the cutoff).

So, instead, if you sample at, say, 8 times the intended rate
(i.e., at 8 x 44.1 kHz), then a very simple, perhaps first-
order RC filter with cutoff at 50 kHz would do a more-than-
excellent job, since now you only need to worry to eliminate
frequencies above 4x44.1, or approx. 170 kHz -- that's trivial,
since the RC already has a good attenuation at that point,
and also, the audio signal really has very low contents at
those ultra-high frequencies.

The part that you really were worried about -- a wall-like
cutoff above 20 but below 22, that you get with a digital
filter with nice phase response.

Notice, however, that I'm not describing a universal technique
that is applied unconditionally in every design -- it's just
that it may be very practical and easy, so you do encounter it
quite often -- for speech, for instance, where you want a
sampling rate of 8 kHz (typically), it's quite easy to do the
sampling at a higher rate and then bring it down after it is
in the digital domain.

HTH,

Carlos
--

Jerry,

Yes you are rigth. Aliasing ocurring above the downsampled Fs/2 should
be moot for it will removed later with the decimation digital filter.
However, (ever being accused of being too cautious), one can never
predict ALL parasitics that will find their way onto the input signal,
even if only sligthly. The VERY high frequency parasitics (RF and
above) are usually difficult to predict; depends on environmental
geometry and such ... .Therefore, since at least in theory, all high
frequency components should fold back and thus cause aliasing ...well,
I doubt that we will be lucky enough that all parasitics will fold back
only between [downsampled Fs/2] and  [oversampling Fs/2].
Hence, this leads me back to my precautionnary "broad analog filter"
solution (ahead of A/D) that would (at least in theory) filter the very
high unforseen frequencies of the input signal that would cause at
least some minor aliasing (how minor? probably below 1%, but better
safe than sorry I say ).
I would think that in today's world of cell phone, sattelite, radio,
wireless gadgets ...., one would "clean" a signal as much as possible,
cost-permitting obviously.
As for Servo work concerning the harmfull effect of the delay of an
analog antialiasing filter, well I would have to look into that for I
am sure there is merit into that as well.

-Roger

rover8898 wrote:
> Jerry,
> 
> Yes you are rigth. Aliasing ocurring above the downsampled Fs/2 should
> be moot for it will removed later with the decimation digital filter.
> However, (ever being accused of being too cautious), one can never
> predict ALL parasitics that will find their way onto the input signal,
> even if only sligthly. The VERY high frequency parasitics (RF and
> above) are usually difficult to predict; depends on environmental
> geometry and such ... .Therefore, since at least in theory, all high
> frequency components should fold back and thus cause aliasing ...well,
> I doubt that we will be lucky enough that all parasitics will fold back
> only between [downsampled Fs/2] and  [oversampling Fs/2].
> Hence, this leads me back to my precautionnary "broad analog filter"
> solution (ahead of A/D) that would (at least in theory) filter the very
> high unforseen frequencies of the input signal that would cause at
> least some minor aliasing (how minor? probably below 1%, but better
> safe than sorry I say ).
> I would think that in today's world of cell phone, sattelite, radio,
> wireless gadgets ...., one would "clean" a signal as much as possible,
> cost-permitting obviously.
> As for Servo work concerning the harmfull effect of the delay of an
> analog antialiasing filter, well I would have to look into that for I
> am sure there is merit into that as well.

Roger,

Engineering is compromise. No filter is perfect. Alias components 
significantly smaller than one LSB are generally harmless, and going to 
great lengths to make them smaller can degrade other aspects of a 
design. "Better safe than sorry" is a fine motto, but one must not put 
oneself in the position of being sorry for trying to be too safe.

Jerry
-- 
      You know that the outhouse is in the right place if       ���
      it seems too close in summer and too far in winter.       ���
�������������������������������������������������������������������

>but one must not put oneself in the position of being sorry for trying to be too safe

Been there. Done that. :-).

Ok. Gotcha. Have to learn to comprise between solutions for potential
problems and solutions for problable problems.

 -Thanks
Roger