Re: Order of filter

Hello Muhammad - Analog filters are designed much like Digital filters,
but the implementation is different.

In any filter there are these general parameters,

Filter function: Lpw Pass, High Pass, Bandpass
Filter type (butterworth, chebychev, bessel, linear phase, etc)
Passband ripple,
Cutoff frequency (with some dB specification at the cutoff),
Stopband frequency
Stopband attenuation

In an analog filter one can look at the classic textbook attenuation
curves to determine the filter 'Order' for the specifications above.

In a digital filter we can use a tool like the GNU Radio supplied
'Filter Design Tool' to determine the number of 'Taps'.

Point 1) There really isn't an equivalent between the Analog Filter
Order and the Digital Filter 'Taps', but for a specific design you can
do the design back and forth between both domains and you will probably
see some correlation between Order and Taps for that specific design.

Point 2) If you wish to get a rough equivalent to the 'Analog Order' of
a digital filter you can go look in any filter textbook and check the
curves out for the specifications given.

Point 3) If you stack two say: 1st Order, 1 kHz cutoff low pass filters
Active Analog Filters together and measure the response you will find
that the cutoff frequency measures around 640 Hz. This is because the
filters react with each other to make a narrower passband. For higher
order filters the design must take into account the number of stages (ot
Taps) to properly calculate the pole locations. The same thing happens
with a digital filter, namely stacking identical filters will result in
the passband getting narrower, although the ultimate stopband
attenuation will increase. (I talked about the LPF case here, the
classic BPF and HPF equivalents can be found in any filter textbook).

A quick way to play with filters in GNURadio is to design a filter
block, add a noise source as the input and look at the 'Averaged' output
with a QT Frequency Sink Block - there you will see the basic response
curve.

Hope this brief overview helps.

Steve Hageman