[Octave-bug-tracker] [bug #45811] signal package: ellipord returns complex for band-stop filter

[Charles Praplan]

Follow-up Comment #2, bug #45811 (project octave):

Hi,

I worked a little bit on the ellipord function and found some
corrections and improvements that I have implemented and tested:

1) Correction for complex order for notch filter

2) Correction for minimum order in case of not symetrical
   requirements (notch)

     In case of not symmetrical requirements, the symmetry was
     always done by preserving the passband and modifying the
     stopband. This could lead to non minimum order for notch
     filters. Note that Matlab also has this problem.

3) Implementation of functionality for analog filters as in
   Matlab (this includes a new 'validate_filter_bandss.m'
   function for analog filters).


I append the file ellipord_case_2015_10_15.zip containing
the corrected files, 8 test cases and 2 plots to compare
matlab to octave:

- ellipord.m
- validate_filter_bandss.m

- ellipord_test_ana_LP.m
- ellipord_test_ana_HP.m
- ellipord_test_ana_BP.m
- ellipord_test_ana_notch.m
- ellipord_test_dig_LP.m
- ellipord_test_dig_HP.m
- ellipord_test_dig_BP.m
- ellipord_test_dig_notch.m

- result_ellipord_test_ana_notch_with_matlab.png
- result_ellipord_test_ana_notch_with_new_ellipord_octave.pdf

One can run the two notch test cases on matlab to see non
optimal order.

If you find this contribution usefull, please integrate them
into the next release.

Ch. Praplan

(file #35202)
    _______________________________________________________

Additional Item Attachment:

File name: ellipord_case_2015_10_15.zip   Size:51 KB


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/bugs/?45811>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/