A active multi-order filter has a cutoff frequency at -3dB.
A bunch of identical passive filters in series produce a droopy response that reduces the cutoff amount depending on how many there are so its -3dB frequency is not what is simply calculated.
I think, this applies to Butterworth (and in some cases to Thomson-Bessel) filters only.
Filter circuits following the Chebyshev (or similar) approximations have cutoff frequencies which are defined in relation to the width of the ripple region within the passband.
Hence, this may lead to cutoff frequencies which are 0.1 or 0.2 or.....or 3 dB below the maximum.
-3 dB (half power) cutoff frequency criterion is used for filters with non-ripple passband. For equi-ripple Chebyshev and elliptical (Cauer) filters, the ripple related criterion described by LvW is used by most authors, see graph below. Some books are however calculating a -3 dB cutoff frequency for these filters.
-3 dB (half power) cutoff frequency criterion is used for filters with non-ripple passband. For equi-ripple Chebyshev and elliptical (Cauer) filters, the ripple related criterion described by LvW is used by most authors, see graph below. Some books are however calculating a -3 dB cutoff frequency for these filters.
Yes - and everybody may decide if it really makes sense to define the passband at a point where the magnitude is another 3 dB down with respect to the value at 0 Hz.