FDAtool: Cutoff for "Chebyshev" and "Elliptic"

Hi all.

I'm about to design a couple of IIR filters in FDA tool. The cutoff-frequency normally is -3 dB, but for Chebyshev and Elliptic it seems like it's only -1 dB.

Am I doing something wrong, or is it simple because I don't know the correct definitions? :)

I would appreciate some sort of verification, if possible.

Thanks!

Alex.

[Honglei Chen]

Could you provide an example of this -1 dB behavior? Thanks.

[Alex]

Sure.

Here is a plot for a 2. order, with bandpass 20-150 Hz.

http://i162.photobucket.com/albums/t244/the_julle/001-1.png

4. order, with bandpass 20-150 Hz.

http://i162.photobucket.com/albums/t244/the_julle/002-1.png

Also, the Butterworth filter drops quicker... I would expect the opposite - that Chebyshev and Elliptic drops quicker than the Butterworth?

Answer 1

[Honglei Chen]

Hi Alex,

I think there is some misunderstanding regarding the design specification. In IIR filter, if you specify the filter order, the parameters for Butterworth method and the parameters for Chebyshev/Elliptic are different. For Butterworth, like you said, the parameter is the cutoff frequency, i.e. 3dB point. However, for Chebyshev/Elliptic method, you specify the passband and passband ripples. Based on your picture, I assume you are using the default passband ripple (Apass), which is 1dB. This means that between 20 and 150 Hz, the fluctuation cannot exceed 1dB. If you think you can live with a 3dB drop, you can then change your passband ripple (Apass) to 3dB, although such setting is rare in a Chebyshev/Elliptic design.

HTH

Honglei

[Alex]

Thanks!

You are correct, I'm using a 1dB passband.